diff options
Diffstat (limited to 'clang/test')
141 files changed, 19769 insertions, 7857 deletions
diff --git a/clang/test/AST/ByteCode/bitfields.cpp b/clang/test/AST/ByteCode/bitfields.cpp index df8d567..a583282 100644 --- a/clang/test/AST/ByteCode/bitfields.cpp +++ b/clang/test/AST/ByteCode/bitfields.cpp @@ -128,3 +128,82 @@ namespace NonConstBitWidth { // both-note {{read of non-const variable}} }; } + +namespace IncDecOverflow { + constexpr bool test1() { + struct {unsigned u: 5; } a {}; + a.u--; + return a.u == 31; + } + static_assert(test1(), ""); + + constexpr bool test2() { + struct {unsigned u: 5; } a {}; + --a.u; + return a.u == 31; + } + static_assert(test2(), ""); + + constexpr bool test3() { + int x = 0; + struct {unsigned u: 5; } a {}; + x = a.u--; + return a.u == 31; + } + static_assert(test3(), ""); + + constexpr bool test4() { + int x = 0; + struct {unsigned u: 5; } a {}; + x = --a.u; + return a.u == 31; + } + static_assert(test4(), ""); + + constexpr bool test5() { + struct {unsigned u: 5; } a {}; + a.u = 31; + ++a.u; + + return a.u == 0; + } + static_assert(test5(), ""); + + constexpr bool test6() { + struct {unsigned u: 5; } a {}; + a.u = 31; + ++a.u; + + return a.u == 0; + } + static_assert(test6(), ""); + + constexpr bool test7() { + struct {unsigned u: 5; } a {}; + a.u = 31; + a.u++; + + return a.u == 0; + } + static_assert(test7(), ""); + + constexpr bool test8() { + int x = 0; + struct {unsigned u: 5; } a {}; + a.u = 31; + x = a.u++; + + return a.u == 0; + } + static_assert(test8(), ""); + + constexpr bool test9() { + int x = 0; + struct {unsigned u: 5; } a {}; + a.u = 31; + x = ++a.u; + + return a.u == 0; + } + static_assert(test9(), ""); +} diff --git a/clang/test/AST/ByteCode/builtin-functions.cpp b/clang/test/AST/ByteCode/builtin-functions.cpp index f47bc49..0b7d51b 100644 --- a/clang/test/AST/ByteCode/builtin-functions.cpp +++ b/clang/test/AST/ByteCode/builtin-functions.cpp @@ -63,6 +63,19 @@ constexpr int test_address_of_incomplete_array_type() { // both-error {{never pr static_assert(test_address_of_incomplete_array_type() == 1234, ""); // both-error {{constant}} \ // both-note {{in call}} +namespace LocalExternRedecl { + constexpr int externRedecl1() { + extern int arr[]; + return 0; + } + constexpr int externRedecl2() { // both-error {{never produces a constant expression}} + extern int arr[]; + __builtin_memmove(&arr, &arr, 4 * sizeof(arr[0])); // both-note 2{{incomplete type}} + return 1234; + } + static_assert(externRedecl2() == 1234); // both-error {{not an integral constant expression}} \ + // both-note {{in call to}} +} struct NonTrivial { constexpr NonTrivial() : n(0) {} diff --git a/clang/test/AST/ByteCode/cxx14.cpp b/clang/test/AST/ByteCode/cxx14.cpp new file mode 100644 index 0000000..9622311 --- /dev/null +++ b/clang/test/AST/ByteCode/cxx14.cpp @@ -0,0 +1,9 @@ +// RUN: %clang_cc1 -std=c++14 -verify=both,expected %s -fexperimental-new-constant-interpreter +// RUN: %clang_cc1 -std=c++14 -verify=both,ref %s + + + +constexpr int(*null_ptr)() = nullptr; +constexpr int test4 = (*null_ptr)(); // both-error {{must be initialized by a constant expression}} \ + // both-note {{evaluates to a null function pointer}} + diff --git a/clang/test/C/C11/n1285_1.c b/clang/test/C/C11/n1285_1.c index 345ec94..a1422f2 100644 --- a/clang/test/C/C11/n1285_1.c +++ b/clang/test/C/C11/n1285_1.c @@ -32,9 +32,9 @@ struct X f(void); // C11-O2-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_X]], ptr [[REF_TMP]], i32 0, i32 0 // C11-O2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [5 x i32], ptr [[A]], i64 0, i64 0 // C11-O2-NEXT: call void @llvm.lifetime.end.p0(ptr [[REF_TMP]]) #[[ATTR5]] -// C11-O2-NEXT: store ptr [[ARRAYDECAY]], ptr [[P]], align 8, !tbaa [[INTPTR_TBAA2:![0-9]+]] -// C11-O2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[P]], align 8, !tbaa [[INTPTR_TBAA2]] -// C11-O2-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA7:![0-9]+]] +// C11-O2-NEXT: store ptr [[ARRAYDECAY]], ptr [[P]], align 8, !tbaa [[INTPTR_TBAA6:![0-9]+]] +// C11-O2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[P]], align 8, !tbaa [[INTPTR_TBAA6]] +// C11-O2-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA2:![0-9]+]] // C11-O2-NEXT: call void @llvm.lifetime.end.p0(ptr [[P]]) #[[ATTR5]] // C11-O2-NEXT: ret i32 [[TMP1]] // @@ -91,18 +91,18 @@ int func_return(void) { // C11-O2: [[COND_END]]: // C11-O2-NEXT: [[A1:%.*]] = getelementptr inbounds nuw [[STRUCT_X]], ptr [[REF_TMP]], i32 0, i32 0 // C11-O2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [5 x i32], ptr [[A1]], i64 0, i64 0 -// C11-O2-NEXT: store ptr [[ARRAYDECAY]], ptr @p, align 8, !tbaa [[INTPTR_TBAA2]] +// C11-O2-NEXT: store ptr [[ARRAYDECAY]], ptr @p, align 8, !tbaa [[INTPTR_TBAA6]] // C11-O2-NEXT: call void @llvm.lifetime.end.p0(ptr [[REF_TMP]]) #[[ATTR5]] // C11-O2-NEXT: call void @llvm.lifetime.start.p0(ptr [[Q]]) #[[ATTR5]] // C11-O2-NEXT: call void @llvm.memset.p0.i64(ptr align 4 [[DOTCOMPOUNDLITERAL]], i8 0, i64 20, i1 false) // C11-O2-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_X]], ptr [[DOTCOMPOUNDLITERAL]], i32 0, i32 0 // C11-O2-NEXT: [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_X]], ptr [[DOTCOMPOUNDLITERAL]], i32 0, i32 0 // C11-O2-NEXT: [[ARRAYDECAY4:%.*]] = getelementptr inbounds [5 x i32], ptr [[A3]], i64 0, i64 0 -// C11-O2-NEXT: store ptr [[ARRAYDECAY4]], ptr [[Q]], align 8, !tbaa [[INTPTR_TBAA2]] -// C11-O2-NEXT: [[TMP0:%.*]] = load ptr, ptr @p, align 8, !tbaa [[INTPTR_TBAA2]] -// C11-O2-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA7]] -// C11-O2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[Q]], align 8, !tbaa [[INTPTR_TBAA2]] -// C11-O2-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4, !tbaa [[INT_TBAA7]] +// C11-O2-NEXT: store ptr [[ARRAYDECAY4]], ptr [[Q]], align 8, !tbaa [[INTPTR_TBAA6]] +// C11-O2-NEXT: [[TMP0:%.*]] = load ptr, ptr @p, align 8, !tbaa [[INTPTR_TBAA6]] +// C11-O2-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA2]] +// C11-O2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[Q]], align 8, !tbaa [[INTPTR_TBAA6]] +// C11-O2-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4, !tbaa [[INT_TBAA2]] // C11-O2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP1]], [[TMP3]] // C11-O2-NEXT: call void @llvm.lifetime.end.p0(ptr [[Q]]) #[[ATTR5]] // C11-O2-NEXT: ret i32 [[ADD]] @@ -138,10 +138,10 @@ int ternary(void) { // C11-O2-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[REF_TMP]], ptr align 4 [[X]], i64 20, i1 false), !tbaa.struct [[TBAA_STRUCT9:![0-9]+]] // C11-O2-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_X]], ptr [[REF_TMP]], i32 0, i32 0 // C11-O2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [5 x i32], ptr [[A]], i64 0, i64 0 -// C11-O2-NEXT: store ptr [[ARRAYDECAY]], ptr @p, align 8, !tbaa [[INTPTR_TBAA2]] +// C11-O2-NEXT: store ptr [[ARRAYDECAY]], ptr @p, align 8, !tbaa [[INTPTR_TBAA6]] // C11-O2-NEXT: call void @llvm.lifetime.end.p0(ptr [[REF_TMP]]) #[[ATTR5]] -// C11-O2-NEXT: [[TMP0:%.*]] = load ptr, ptr @p, align 8, !tbaa [[INTPTR_TBAA2]] -// C11-O2-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA7]] +// C11-O2-NEXT: [[TMP0:%.*]] = load ptr, ptr @p, align 8, !tbaa [[INTPTR_TBAA6]] +// C11-O2-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA2]] // C11-O2-NEXT: call void @llvm.lifetime.end.p0(ptr [[X]]) #[[ATTR5]] // C11-O2-NEXT: ret i32 [[TMP1]] // @@ -175,10 +175,10 @@ int comma(void) { // C11-O2-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[REF_TMP]], ptr align 4 [[X]], i64 20, i1 false), !tbaa.struct [[TBAA_STRUCT9]] // C11-O2-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_X]], ptr [[REF_TMP]], i32 0, i32 0 // C11-O2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [5 x i32], ptr [[A]], i64 0, i64 0 -// C11-O2-NEXT: store ptr [[ARRAYDECAY]], ptr @p, align 8, !tbaa [[INTPTR_TBAA2]] +// C11-O2-NEXT: store ptr [[ARRAYDECAY]], ptr @p, align 8, !tbaa [[INTPTR_TBAA6]] // C11-O2-NEXT: call void @llvm.lifetime.end.p0(ptr [[REF_TMP]]) #[[ATTR5]] -// C11-O2-NEXT: [[TMP0:%.*]] = load ptr, ptr @p, align 8, !tbaa [[INTPTR_TBAA2]] -// C11-O2-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA7]] +// C11-O2-NEXT: [[TMP0:%.*]] = load ptr, ptr @p, align 8, !tbaa [[INTPTR_TBAA6]] +// C11-O2-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA2]] // C11-O2-NEXT: call void @llvm.lifetime.end.p0(ptr [[X]]) #[[ATTR5]] // C11-O2-NEXT: ret i32 [[TMP1]] // @@ -217,10 +217,10 @@ int cast(void) { // C11-O2-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[REF_TMP]], ptr align 4 [[X]], i64 20, i1 false), !tbaa.struct [[TBAA_STRUCT9]] // C11-O2-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_X]], ptr [[REF_TMP]], i32 0, i32 0 // C11-O2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [5 x i32], ptr [[A]], i64 0, i64 0 -// C11-O2-NEXT: store ptr [[ARRAYDECAY]], ptr @p, align 8, !tbaa [[INTPTR_TBAA2]] +// C11-O2-NEXT: store ptr [[ARRAYDECAY]], ptr @p, align 8, !tbaa [[INTPTR_TBAA6]] // C11-O2-NEXT: call void @llvm.lifetime.end.p0(ptr [[REF_TMP]]) #[[ATTR5]] -// C11-O2-NEXT: [[TMP0:%.*]] = load ptr, ptr @p, align 8, !tbaa [[INTPTR_TBAA2]] -// C11-O2-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA7]] +// C11-O2-NEXT: [[TMP0:%.*]] = load ptr, ptr @p, align 8, !tbaa [[INTPTR_TBAA6]] +// C11-O2-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA2]] // C11-O2-NEXT: call void @llvm.lifetime.end.p0(ptr [[S]]) #[[ATTR5]] // C11-O2-NEXT: call void @llvm.lifetime.end.p0(ptr [[X]]) #[[ATTR5]] // C11-O2-NEXT: ret i32 [[TMP1]] @@ -232,13 +232,13 @@ int assign(void) { return *p; } //. -// C11-O2: [[INTPTR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// C11-O2: [[META3]] = !{!"p1 int", [[META4:![0-9]+]], i64 0} -// C11-O2: [[META4]] = !{!"any pointer", [[META5:![0-9]+]], i64 0} -// C11-O2: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// C11-O2: [[META6]] = !{!"Simple C/C++ TBAA"} -// C11-O2: [[INT_TBAA7]] = !{[[META8:![0-9]+]], [[META8]], i64 0} -// C11-O2: [[META8]] = !{!"int", [[META5]], i64 0} +// C11-O2: [[INT_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// C11-O2: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// C11-O2: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// C11-O2: [[META5]] = !{!"Simple C/C++ TBAA"} +// C11-O2: [[INTPTR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// C11-O2: [[META7]] = !{!"p1 int", [[META8:![0-9]+]], i64 0} +// C11-O2: [[META8]] = !{!"any pointer", [[META4]], i64 0} // C11-O2: [[TBAA_STRUCT9]] = !{i64 0, i64 20, [[META10:![0-9]+]]} -// C11-O2: [[META10]] = !{[[META5]], [[META5]], i64 0} +// C11-O2: [[META10]] = !{[[META4]], [[META4]], i64 0} //. diff --git a/clang/test/CIR/CodeGen/aapcs-volatile-bitfields.c b/clang/test/CIR/CodeGen/aapcs-volatile-bitfields.c index 92eae6a..19362cf 100644 --- a/clang/test/CIR/CodeGen/aapcs-volatile-bitfields.c +++ b/clang/test/CIR/CodeGen/aapcs-volatile-bitfields.c @@ -1,11 +1,11 @@ -// RUN: %clang_cc1 -triple aarch64-unknown-linux-gnu -fclangir -emit-cir -fdump-record-layouts %s -o %t.cir 1> %t.cirlayout +// RUN: %clang_cc1 -triple aarch64-unknown-linux-gnu -fclangir -emit-cir -fdump-record-layouts %s -o %t.cir > %t.cirlayout // RUN: FileCheck --input-file=%t.cirlayout %s --check-prefix=CIR-LAYOUT // RUN: FileCheck --input-file=%t.cir %s --check-prefix=CIR // RUN: %clang_cc1 -triple aarch64-unknown-linux-gnu -fclangir -emit-llvm %s -o %t-cir.ll // RUN: FileCheck --input-file=%t-cir.ll %s --check-prefix=LLVM -// RUN: %clang_cc1 -triple aarch64-unknown-linux-gnu -emit-llvm -fdump-record-layouts %s -o %t.ll 1> %t.ogcglayout +// RUN: %clang_cc1 -triple aarch64-unknown-linux-gnu -emit-llvm -fdump-record-layouts %s -o %t.ll > %t.ogcglayout // RUN: FileCheck --input-file=%t.ogcglayout %s --check-prefix=OGCG-LAYOUT // RUN: FileCheck --input-file=%t.ll %s --check-prefix=OGCG diff --git a/clang/test/CIR/CodeGen/atomic.c b/clang/test/CIR/CodeGen/atomic.c index 440010a..6579988 100644 --- a/clang/test/CIR/CodeGen/atomic.c +++ b/clang/test/CIR/CodeGen/atomic.c @@ -514,3 +514,596 @@ void atomic_exchange_n(int *ptr, int value) { // OGCG: %{{.+}} = atomicrmw xchg ptr %{{.+}}, i32 %{{.+}} acq_rel, align 4 // OGCG: %{{.+}} = atomicrmw xchg ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 } + +void test_and_set(void *p) { + // CIR-LABEL: @test_and_set + // LLVM-LABEL: @test_and_set + // OGCG-LABEL: @test_and_set + + __atomic_test_and_set(p, __ATOMIC_SEQ_CST); + // CIR: %[[VOID_PTR:.+]] = cir.load align(8) %{{.+}} : !cir.ptr<!cir.ptr<!void>>, !cir.ptr<!void> + // CIR-NEXT: %[[PTR:.+]] = cir.cast bitcast %[[VOID_PTR]] : !cir.ptr<!void> -> !cir.ptr<!s8i> + // CIR-NEXT: %[[RES:.+]] = cir.atomic.test_and_set seq_cst %[[PTR]] : !cir.ptr<!s8i> -> !cir.bool + // CIR-NEXT: cir.store align(1) %[[RES]], %{{.+}} : !cir.bool, !cir.ptr<!cir.bool> + + // LLVM: %[[PTR:.+]] = load ptr, ptr %{{.+}}, align 8 + // LLVM-NEXT: %[[RES:.+]] = atomicrmw xchg ptr %[[PTR]], i8 1 seq_cst, align 1 + // LLVM-NEXT: %{{.+}} = icmp ne i8 %[[RES]], 0 + + // OGCG: %[[PTR:.+]] = load ptr, ptr %{{.+}}, align 8 + // OGCG-NEXT: %[[RES:.+]] = atomicrmw xchg ptr %[[PTR]], i8 1 seq_cst, align 1 + // OGCG-NEXT: %{{.+}} = icmp ne i8 %[[RES]], 0 +} + +void test_and_set_volatile(volatile void *p) { + // CIR-LABEL: @test_and_set_volatile + // LLVM-LABEL: @test_and_set_volatile + // OGCG-LABEL: @test_and_set_volatile + + __atomic_test_and_set(p, __ATOMIC_SEQ_CST); + // CIR: %[[VOID_PTR:.+]] = cir.load align(8) %{{.+}} : !cir.ptr<!cir.ptr<!void>>, !cir.ptr<!void> + // CIR-NEXT: %[[PTR:.+]] = cir.cast bitcast %[[VOID_PTR]] : !cir.ptr<!void> -> !cir.ptr<!s8i> + // CIR-NEXT: %[[RES:.+]] = cir.atomic.test_and_set seq_cst %[[PTR]] volatile : !cir.ptr<!s8i> -> !cir.bool + // CIR-NEXT: cir.store align(1) %[[RES]], %{{.+}} : !cir.bool, !cir.ptr<!cir.bool> + + // LLVM: %[[PTR:.+]] = load ptr, ptr %{{.+}}, align 8 + // LLVM-NEXT: %[[RES:.+]] = atomicrmw volatile xchg ptr %[[PTR]], i8 1 seq_cst, align 1 + // LLVM-NEXT: %{{.+}} = icmp ne i8 %[[RES]], 0 + + // OGCG: %[[PTR:.+]] = load ptr, ptr %{{.+}}, align 8 + // OGCG-NEXT: %[[RES:.+]] = atomicrmw volatile xchg ptr %[[PTR]], i8 1 seq_cst, align 1 + // OGCG-NEXT: %{{.+}} = icmp ne i8 %[[RES]], 0 +} + +void clear(void *p) { + // CIR-LABEL: @clear + // LLVM-LABEL: @clear + // OGCG-LABEL: @clear + + __atomic_clear(p, __ATOMIC_SEQ_CST); + // CIR: %[[VOID_PTR:.+]] = cir.load align(8) %{{.+}} : !cir.ptr<!cir.ptr<!void>>, !cir.ptr<!void> + // CIR-NEXT: %[[PTR:.+]] = cir.cast bitcast %[[VOID_PTR]] : !cir.ptr<!void> -> !cir.ptr<!s8i> + // CIR: cir.atomic.clear seq_cst %[[PTR]] : !cir.ptr<!s8i> + + // LLVM: store atomic i8 0, ptr %{{.+}} seq_cst, align 1 + + // OGCG: store atomic i8 0, ptr %{{.+}} seq_cst, align 1 +} + +void clear_volatile(volatile void *p) { + // CIR-LABEL: @clear_volatile + // LLVM-LABEL: @clear_volatile + // OGCG-LABEL: @clear_volatile + + __atomic_clear(p, __ATOMIC_SEQ_CST); + // CIR: %[[VOID_PTR:.+]] = cir.load align(8) %{{.+}} : !cir.ptr<!cir.ptr<!void>>, !cir.ptr<!void> + // CIR-NEXT: %[[PTR:.+]] = cir.cast bitcast %[[VOID_PTR]] : !cir.ptr<!void> -> !cir.ptr<!s8i> + // CIR: cir.atomic.clear seq_cst %[[PTR]] volatile : !cir.ptr<!s8i> + + // LLVM: store atomic volatile i8 0, ptr %{{.+}} seq_cst, align 1 + + // OGCG: store atomic volatile i8 0, ptr %{{.+}} seq_cst, align 1 +} + +int atomic_fetch_add(int *ptr, int value) { + // CIR-LABEL: @atomic_fetch_add + // LLVM-LABEL: @atomic_fetch_add + // OGCG-LABEL: @atomic_fetch_add + + return __atomic_fetch_add(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch add seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw add ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw add ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_add_fetch(int *ptr, int value) { + // CIR-LABEL: @atomic_add_fetch + // LLVM-LABEL: @atomic_add_fetch + // OGCG-LABEL: @atomic_add_fetch + + return __atomic_add_fetch(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch add seq_cst %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[OLD:.+]] = atomicrmw add ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // LLVM-NEXT: %[[RES:.+]] = add i32 %[[OLD]], %[[VAL]] + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[OLD:.+]] = atomicrmw add ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // OGCG-NEXT: %[[RES:.+]] = add i32 %[[OLD]], %[[VAL]] + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int c11_atomic_fetch_add(_Atomic(int) *ptr, int value) { + // CIR-LABEL: @c11_atomic_fetch_add + // LLVM-LABEL: @c11_atomic_fetch_add + // OGCG-LABEL: @c11_atomic_fetch_add + + return __c11_atomic_fetch_add(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch add seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw add ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw add ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_fetch_sub(int *ptr, int value) { + // CIR-LABEL: @atomic_fetch_sub + // LLVM-LABEL: @atomic_fetch_sub + // OGCG-LABEL: @atomic_fetch_sub + + return __atomic_fetch_sub(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch sub seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw sub ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw sub ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_sub_fetch(int *ptr, int value) { + // CIR-LABEL: @atomic_sub_fetch + // LLVM-LABEL: @atomic_sub_fetch + // OGCG-LABEL: @atomic_sub_fetch + + return __atomic_sub_fetch(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch sub seq_cst %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[OLD:.+]] = atomicrmw sub ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // LLVM-NEXT: %[[RES:.+]] = sub i32 %[[OLD]], %[[VAL]] + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[OLD:.+]] = atomicrmw sub ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // OGCG-NEXT: %[[RES:.+]] = sub i32 %[[OLD]], %[[VAL]] + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int c11_atomic_fetch_sub(_Atomic(int) *ptr, int value) { + // CIR-LABEL: @c11_atomic_fetch_sub + // LLVM-LABEL: @c11_atomic_fetch_sub + // OGCG-LABEL: @c11_atomic_fetch_sub + + return __c11_atomic_fetch_sub(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch sub seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw sub ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw sub ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +float atomic_fetch_add_fp(float *ptr, float value) { + // CIR-LABEL: @atomic_fetch_add_fp + // LLVM-LABEL: @atomic_fetch_add_fp + // OGCG-LABEL: @atomic_fetch_add_fp + + return __atomic_fetch_add(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch add seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!cir.float>, !cir.float) -> !cir.float + + // LLVM: %[[RES:.+]] = atomicrmw fadd ptr %{{.+}}, float %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw fadd ptr %{{.+}}, float %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 +} + +float atomic_add_fetch_fp(float *ptr, float value) { + // CIR-LABEL: @atomic_add_fetch_fp + // LLVM-LABEL: @atomic_add_fetch_fp + // OGCG-LABEL: @atomic_add_fetch_fp + + return __atomic_add_fetch(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch add seq_cst %{{.+}}, %{{.+}} : (!cir.ptr<!cir.float>, !cir.float) -> !cir.float + + // LLVM: %[[OLD:.+]] = atomicrmw fadd ptr %{{.+}}, float %[[VAL:.+]] seq_cst, align 4 + // LLVM-NEXT: %[[RES:.+]] = fadd float %[[OLD]], %[[VAL]] + // LLVM-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[OLD:.+]] = atomicrmw fadd ptr %{{.+}}, float %[[VAL:.+]] seq_cst, align 4 + // OGCG-NEXT: %[[RES:.+]] = fadd float %[[OLD]], %[[VAL]] + // OGCG-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 +} + +float c11_atomic_fetch_sub_fp(_Atomic(float) *ptr, float value) { + // CIR-LABEL: @c11_atomic_fetch_sub_fp + // LLVM-LABEL: @c11_atomic_fetch_sub_fp + // OGCG-LABEL: @c11_atomic_fetch_sub_fp + + return __c11_atomic_fetch_sub(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch sub seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!cir.float>, !cir.float) -> !cir.float + + // LLVM: %[[RES:.+]] = atomicrmw fsub ptr %{{.+}}, float %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw fsub ptr %{{.+}}, float %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_fetch_min(int *ptr, int value) { + // CIR-LABEL: @atomic_fetch_min + // LLVM-LABEL: @atomic_fetch_min + // OGCG-LABEL: @atomic_fetch_min + + return __atomic_fetch_min(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch min seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw min ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw min ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_min_fetch(int *ptr, int value) { + // CIR-LABEL: @atomic_min_fetch + // LLVM-LABEL: @atomic_min_fetch + // OGCG-LABEL: @atomic_min_fetch + + return __atomic_min_fetch(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch min seq_cst %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[OLD:.+]] = atomicrmw min ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // LLVM-NEXT: %[[OLD_LESS:.+]] = icmp slt i32 %[[OLD]], %[[VAL]] + // LLVM-NEXT: %[[RES:.+]] = select i1 %[[OLD_LESS]], i32 %[[OLD]], i32 %[[VAL]] + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[OLD:.+]] = atomicrmw min ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // OGCG-NEXT: %[[OLD_LESS:.+]] = icmp slt i32 %[[OLD]], %[[VAL]] + // OGCG-NEXT: %[[RES:.+]] = select i1 %[[OLD_LESS]], i32 %[[OLD]], i32 %[[VAL]] + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int c11_atomic_fetch_min(_Atomic(int) *ptr, int value) { + // CIR-LABEL: @c11_atomic_fetch_min + // LLVM-LABEL: @c11_atomic_fetch_min + // OGCG-LABEL: @c11_atomic_fetch_min + + return __c11_atomic_fetch_min(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch min seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw min ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw min ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +float atomic_fetch_min_fp(float *ptr, float value) { + // CIR-LABEL: @atomic_fetch_min_fp + // LLVM-LABEL: @atomic_fetch_min_fp + // OGCG-LABEL: @atomic_fetch_min_fp + + return __atomic_fetch_min(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch min seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!cir.float>, !cir.float) -> !cir.float + + // LLVM: %[[RES:.+]] = atomicrmw fmin ptr %{{.+}}, float %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw fmin ptr %{{.+}}, float %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 +} + +float atomic_min_fetch_fp(float *ptr, float value) { + // CIR-LABEL: @atomic_min_fetch_fp + // LLVM-LABEL: @atomic_min_fetch_fp + // OGCG-LABEL: @atomic_min_fetch_fp + + return __atomic_min_fetch(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch min seq_cst %{{.+}}, %{{.+}} : (!cir.ptr<!cir.float>, !cir.float) -> !cir.float + + // LLVM: %[[OLD:.+]] = atomicrmw fmin ptr %{{.+}}, float %[[VAL:.+]] seq_cst, align 4 + // LLVM-NEXT: %[[RES:.+]] = call float @llvm.minnum.f32(float %[[OLD]], float %[[VAL]]) + // LLVM-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[OLD:.+]] = atomicrmw fmin ptr %{{.+}}, float %[[VAL:.+]] seq_cst, align 4 + // OGCG-NEXT: %[[RES:.+]] = call float @llvm.minnum.f32(float %[[OLD]], float %[[VAL]]) + // OGCG-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 +} + +float c11_atomic_fetch_min_fp(_Atomic(float) *ptr, float value) { + // CIR-LABEL: @c11_atomic_fetch_min_fp + // LLVM-LABEL: @c11_atomic_fetch_min_fp + // OGCG-LABEL: @c11_atomic_fetch_min_fp + + return __c11_atomic_fetch_min(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch min seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!cir.float>, !cir.float) -> !cir.float + + // LLVM: %[[RES:.+]] = atomicrmw fmin ptr %{{.+}}, float %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw fmin ptr %{{.+}}, float %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_fetch_max(int *ptr, int value) { + // CIR-LABEL: @atomic_fetch_max + // LLVM-LABEL: @atomic_fetch_max + // OGCG-LABEL: @atomic_fetch_max + + return __atomic_fetch_max(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch max seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw max ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw max ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_max_fetch(int *ptr, int value) { + // CIR-LABEL: @atomic_max_fetch + // LLVM-LABEL: @atomic_max_fetch + // OGCG-LABEL: @atomic_max_fetch + + return __atomic_max_fetch(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch max seq_cst %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[OLD:.+]] = atomicrmw max ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // LLVM-NEXT: %[[OLD_GREATER:.+]] = icmp sgt i32 %[[OLD]], %[[VAL]] + // LLVM-NEXT: %[[RES:.+]] = select i1 %[[OLD_GREATER]], i32 %[[OLD]], i32 %[[VAL]] + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[OLD:.+]] = atomicrmw max ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // OGCG-NEXT: %[[OLD_GREATER:.+]] = icmp sgt i32 %[[OLD]], %[[VAL]] + // OGCG-NEXT: %[[RES:.+]] = select i1 %[[OLD_GREATER]], i32 %[[OLD]], i32 %[[VAL]] + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int c11_atomic_fetch_max(_Atomic(int) *ptr, int value) { + // CIR-LABEL: @c11_atomic_fetch_max + // LLVM-LABEL: @c11_atomic_fetch_max + // OGCG-LABEL: @c11_atomic_fetch_max + + return __c11_atomic_fetch_max(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch max seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw max ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw max ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +float atomic_fetch_max_fp(float *ptr, float value) { + // CIR-LABEL: @atomic_fetch_max_fp + // LLVM-LABEL: @atomic_fetch_max_fp + // OGCG-LABEL: @atomic_fetch_max_fp + + return __atomic_fetch_max(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch max seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!cir.float>, !cir.float) -> !cir.float + + // LLVM: %[[RES:.+]] = atomicrmw fmax ptr %{{.+}}, float %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw fmax ptr %{{.+}}, float %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 +} + +float atomic_max_fetch_fp(float *ptr, float value) { + // CIR-LABEL: @atomic_max_fetch_fp + // LLVM-LABEL: @atomic_max_fetch_fp + // OGCG-LABEL: @atomic_max_fetch_fp + + return __atomic_max_fetch(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch max seq_cst %{{.+}}, %{{.+}} : (!cir.ptr<!cir.float>, !cir.float) -> !cir.float + + // LLVM: %[[OLD:.+]] = atomicrmw fmax ptr %{{.+}}, float %[[VAL:.+]] seq_cst, align 4 + // LLVM-NEXT: %[[RES:.+]] = call float @llvm.maxnum.f32(float %[[OLD]], float %[[VAL]]) + // LLVM-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[OLD:.+]] = atomicrmw fmax ptr %{{.+}}, float %[[VAL:.+]] seq_cst, align 4 + // OGCG-NEXT: %[[RES:.+]] = call float @llvm.maxnum.f32(float %[[OLD]], float %[[VAL]]) + // OGCG-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 +} + +float c11_atomic_fetch_max_fp(_Atomic(float) *ptr, float value) { + // CIR-LABEL: @c11_atomic_fetch_max_fp + // LLVM-LABEL: @c11_atomic_fetch_max_fp + // OGCG-LABEL: @c11_atomic_fetch_max_fp + + return __c11_atomic_fetch_max(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch max seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!cir.float>, !cir.float) -> !cir.float + + // LLVM: %[[RES:.+]] = atomicrmw fmax ptr %{{.+}}, float %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw fmax ptr %{{.+}}, float %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store float %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_fetch_and(int *ptr, int value) { + // CIR-LABEL: @atomic_fetch_and + // LLVM-LABEL: @atomic_fetch_and + // OGCG-LABEL: @atomic_fetch_and + + return __atomic_fetch_and(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch and seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw and ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw and ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_and_fetch(int *ptr, int value) { + // CIR-LABEL: @atomic_and_fetch + // LLVM-LABEL: @atomic_and_fetch + // OGCG-LABEL: @atomic_and_fetch + + return __atomic_and_fetch(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch and seq_cst %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[OLD:.+]] = atomicrmw and ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // LLVM-NEXT: %[[RES:.+]] = and i32 %[[OLD]], %[[VAL]] + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[OLD:.+]] = atomicrmw and ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // OGCG-NEXT: %[[RES:.+]] = and i32 %[[OLD]], %[[VAL]] + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int c11_atomic_fetch_and(_Atomic(int) *ptr, int value) { + // CIR-LABEL: @c11_atomic_fetch_and + // LLVM-LABEL: @c11_atomic_fetch_and + // OGCG-LABEL: @c11_atomic_fetch_and + + return __c11_atomic_fetch_and(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch and seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw and ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw and ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_fetch_or(int *ptr, int value) { + // CIR-LABEL: @atomic_fetch_or + // LLVM-LABEL: @atomic_fetch_or + // OGCG-LABEL: @atomic_fetch_or + + return __atomic_fetch_or(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch or seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw or ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw or ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_or_fetch(int *ptr, int value) { + // CIR-LABEL: @atomic_or_fetch + // LLVM-LABEL: @atomic_or_fetch + // OGCG-LABEL: @atomic_or_fetch + + return __atomic_or_fetch(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch or seq_cst %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[OLD:.+]] = atomicrmw or ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // LLVM-NEXT: %[[RES:.+]] = or i32 %[[OLD]], %[[VAL]] + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[OLD:.+]] = atomicrmw or ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // OGCG-NEXT: %[[RES:.+]] = or i32 %[[OLD]], %[[VAL]] + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int c11_atomic_fetch_or(_Atomic(int) *ptr, int value) { + // CIR-LABEL: @c11_atomic_fetch_or + // LLVM-LABEL: @c11_atomic_fetch_or + // OGCG-LABEL: @c11_atomic_fetch_or + + return __c11_atomic_fetch_or(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch or seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw or ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw or ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_fetch_xor(int *ptr, int value) { + // CIR-LABEL: @atomic_fetch_xor + // LLVM-LABEL: @atomic_fetch_xor + // OGCG-LABEL: @atomic_fetch_xor + + return __atomic_fetch_xor(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch xor seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw xor ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw xor ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_xor_fetch(int *ptr, int value) { + // CIR-LABEL: @atomic_xor_fetch + // LLVM-LABEL: @atomic_xor_fetch + // OGCG-LABEL: @atomic_xor_fetch + + return __atomic_xor_fetch(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch xor seq_cst %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[OLD:.+]] = atomicrmw xor ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // LLVM-NEXT: %[[RES:.+]] = xor i32 %[[OLD]], %[[VAL]] + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[OLD:.+]] = atomicrmw xor ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // OGCG-NEXT: %[[RES:.+]] = xor i32 %[[OLD]], %[[VAL]] + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int c11_atomic_fetch_xor(_Atomic(int) *ptr, int value) { + // CIR-LABEL: @c11_atomic_fetch_xor + // LLVM-LABEL: @c11_atomic_fetch_xor + // OGCG-LABEL: @c11_atomic_fetch_xor + + return __c11_atomic_fetch_xor(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch xor seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw xor ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw xor ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_fetch_nand(int *ptr, int value) { + // CIR-LABEL: @atomic_fetch_nand + // LLVM-LABEL: @atomic_fetch_nand + // OGCG-LABEL: @atomic_fetch_nand + + return __atomic_fetch_nand(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch nand seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw nand ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw nand ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int atomic_nand_fetch(int *ptr, int value) { + // CIR-LABEL: @atomic_nand_fetch + // LLVM-LABEL: @atomic_nand_fetch + // OGCG-LABEL: @atomic_nand_fetch + + return __atomic_nand_fetch(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch nand seq_cst %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[OLD:.+]] = atomicrmw nand ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // LLVM-NEXT: %[[TMP:.+]] = and i32 %[[OLD]], %[[VAL]] + // LLVM-NEXT: %[[RES:.+]] = xor i32 %[[TMP]], -1 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[OLD:.+]] = atomicrmw nand ptr %{{.+}}, i32 %[[VAL:.+]] seq_cst, align 4 + // OGCG-NEXT: %[[TMP:.+]] = and i32 %[[OLD]], %[[VAL]] + // OGCG-NEXT: %[[RES:.+]] = xor i32 %[[TMP]], -1 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} + +int c11_atomic_fetch_nand(_Atomic(int) *ptr, int value) { + // CIR-LABEL: @c11_atomic_fetch_nand + // LLVM-LABEL: @c11_atomic_fetch_nand + // OGCG-LABEL: @c11_atomic_fetch_nand + + return __c11_atomic_fetch_nand(ptr, value, __ATOMIC_SEQ_CST); + // CIR: %{{.+}} = cir.atomic.fetch nand seq_cst fetch_first %{{.+}}, %{{.+}} : (!cir.ptr<!s32i>, !s32i) -> !s32i + + // LLVM: %[[RES:.+]] = atomicrmw nand ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // LLVM-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 + + // OGCG: %[[RES:.+]] = atomicrmw nand ptr %{{.+}}, i32 %{{.+}} seq_cst, align 4 + // OGCG-NEXT: store i32 %[[RES]], ptr %{{.+}}, align 4 +} diff --git a/clang/test/CIR/CodeGen/binassign.c b/clang/test/CIR/CodeGen/binassign.c index dab9879..44c54b4 100644 --- a/clang/test/CIR/CodeGen/binassign.c +++ b/clang/test/CIR/CodeGen/binassign.c @@ -1,4 +1,4 @@ -// RUN: %clang_cc1 -std=c23 -triple x86_64-unknown-linux-gnu -fclangir -emit-cir %s -o %t.cir +// RUN: %clang_cc1 -std=c23 -triple x86_64-unknown-linux-gnu -fclangir -emit-cir -mmlir --mlir-print-ir-before=cir-lowering-prepare %s -o %t.cir 2> %t-before-lp.cir // RUN: FileCheck --input-file=%t.cir %s -check-prefix=CIR // RUN: %clang_cc1 -std=c23 -triple x86_64-unknown-linux-gnu -Wno-unused-value -fclangir -emit-llvm %s -o %t-cir.ll // RUN: FileCheck --input-file=%t-cir.ll %s -check-prefix=LLVM @@ -54,3 +54,49 @@ void binary_assign(void) { // OGCG: store float 0x40091EB860000000, ptr %[[F_PTR]] // OGCG: store i32 42, ptr %[[I_PTR]] // OGCG: ret void + +struct S { + int a; + float b; +}; + +struct SV { + int a; + volatile float b; +}; + +struct S gs; +struct SV gsv; + +void binary_assign_struct() { + // Test normal struct assignment + struct S ls; + ls = gs; + + // Test assignment of a struct with a volatile member + struct SV lsv; + lsv = gsv; +} + +// CIR: cir.func{{.*}} @binary_assign_struct() +// CIR: %[[LS:.*]] = cir.alloca ![[REC_S:.*]], !cir.ptr<![[REC_S]]>, ["ls"] +// CIR: %[[LSV:.*]] = cir.alloca ![[REC_SV:.*]], !cir.ptr<![[REC_SV]]>, ["lsv"] +// CIR: %[[GS_PTR:.*]] = cir.get_global @gs : !cir.ptr<![[REC_S]]> +// CIR: cir.copy %[[GS_PTR]] to %[[LS]] : !cir.ptr<![[REC_S]]> +// CIR: %[[GSV_PTR:.*]] = cir.get_global @gsv : !cir.ptr<![[REC_SV]]> +// CIR: cir.copy %[[GSV_PTR]] to %[[LSV]] volatile : !cir.ptr<![[REC_SV]]> +// CIR: cir.return + +// LLVM: define {{.*}}void @binary_assign_struct() +// LLVM: %[[LS_PTR:.*]] = alloca %struct.S +// LLVM: %[[LSV_PTR:.*]] = alloca %struct.SV +// LLVM: call void @llvm.memcpy.p0.p0.i32(ptr %[[LS_PTR]], ptr @gs, i32 8, i1 false) +// LLVM: call void @llvm.memcpy.p0.p0.i32(ptr %[[LSV_PTR]], ptr @gsv, i32 8, i1 true) +// LLVM: ret void + +// OGCG: define {{.*}}void @binary_assign_struct() +// OGCG: %[[LS_PTR:.*]] = alloca %struct.S +// OGCG: %[[LSV_PTR:.*]] = alloca %struct.SV +// OGCG: call void @llvm.memcpy.p0.p0.i64(ptr align 4 %[[LS_PTR]], ptr align 4 @gs, i64 8, i1 false) +// OGCG: call void @llvm.memcpy.p0.p0.i64(ptr align 4 %[[LSV_PTR]], ptr align 4 @gsv, i64 8, i1 true) +// OGCG: ret void diff --git a/clang/test/CIR/CodeGen/builtin_inline.c b/clang/test/CIR/CodeGen/builtin_inline.c new file mode 100644 index 0000000..83a3ba6 --- /dev/null +++ b/clang/test/CIR/CodeGen/builtin_inline.c @@ -0,0 +1,91 @@ +// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -fclangir -emit-cir %s -o %t.cir +// RUN: FileCheck --input-file=%t.cir %s -check-prefix=CIR +// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -fclangir -emit-llvm -disable-llvm-passes %s -o %t-cir.ll +// RUN: FileCheck --input-file=%t-cir.ll %s -check-prefix=LLVM +// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -emit-llvm -disable-llvm-passes %s -o %t.ll +// RUN: FileCheck --input-file=%t.ll %s -check-prefix=OGCG + +typedef unsigned long size_t; + +// Normal inline builtin declaration +// When a builtin is redefined with extern inline + always_inline attributes, +// the compiler creates a .inline version to avoid conflicts with the builtin + +extern inline __attribute__((always_inline)) __attribute__((gnu_inline)) +void *memcpy(void *a, const void *b, size_t c) { + return __builtin_memcpy(a, b, c); +} + +void *test_inline_builtin_memcpy(void *a, const void *b, size_t c) { + return memcpy(a, b, c); +} + +// CIR: cir.func internal private{{.*}}@memcpy.inline({{.*}}) -> !cir.ptr<!void> inline(always) + +// CIR-LABEL: @test_inline_builtin_memcpy( +// CIR: cir.call @memcpy.inline( +// CIR: } + +// LLVM: define internal ptr @memcpy.inline(ptr{{.*}}, ptr{{.*}}, i64{{.*}}) #{{[0-9]+}} + +// LLVM-LABEL: @test_inline_builtin_memcpy( +// LLVM: call ptr @memcpy.inline( + +// OGCG-LABEL: @test_inline_builtin_memcpy( +// OGCG: call ptr @memcpy.inline( + +// OGCG: define internal ptr @memcpy.inline(ptr{{.*}} %a, ptr{{.*}} %b, i64{{.*}} %c) #{{[0-9]+}} + +// Shadowing case +// When a non-inline function definition shadows an inline builtin declaration, +// the .inline version should be replaced with the regular function and removed. + +extern inline __attribute__((always_inline)) __attribute__((gnu_inline)) +void *memmove(void *a, const void *b, size_t c) { + return __builtin_memmove(a, b, c); +} + +void *memmove(void *a, const void *b, size_t c) { + char *dst = (char *)a; + const char *src = (const char *)b; + if (dst < src) { + for (size_t i = 0; i < c; i++) { + dst[i] = src[i]; + } + } else { + for (size_t i = c; i > 0; i--) { + dst[i-1] = src[i-1]; + } + } + return a; +} + +void *test_shadowed_memmove(void *a, const void *b, size_t c) { + return memmove(a, b, c); +} + +// CIR: cir.func{{.*}}@memmove({{.*}}) -> !cir.ptr<!void>{{.*}}{ +// CIR-NOT: @memmove.inline + +// CIR-LABEL: @test_shadowed_memmove( +// CIR: cir.call @memmove( +// CIR-NOT: @memmove.inline +// CIR: } + +// LLVM: define dso_local ptr @memmove(ptr{{.*}}, ptr{{.*}}, i64{{.*}}) #{{[0-9]+}} +// LLVM-NOT: @memmove.inline + +// LLVM-LABEL: @test_shadowed_memmove( +// TODO - this deviation from OGCG is expected until we implement the nobuiltin +// attribute. See CIRGenFunction::emitDirectCallee +// LLVM: call ptr @memmove( +// LLVM-NOT: @memmove.inline +// LLVM: } + +// OGCG: define dso_local ptr @memmove(ptr{{.*}} %a, ptr{{.*}} %b, i64{{.*}} %c) #{{[0-9]+}} +// OGCG-NOT: @memmove.inline + +// OGCG-LABEL: @test_shadowed_memmove( +// OGCG: call void @llvm.memmove.p0.p0.i64( +// OGCG-NOT: @memmove.inline +// OGCG: } diff --git a/clang/test/CIR/CodeGen/dtors.cpp b/clang/test/CIR/CodeGen/dtors.cpp index cb3886b..1fe048b7 100644 --- a/clang/test/CIR/CodeGen/dtors.cpp +++ b/clang/test/CIR/CodeGen/dtors.cpp @@ -14,7 +14,7 @@ void test_temporary_dtor() { } // CIR: cir.func dso_local @_Z19test_temporary_dtorv() -// CIR: %[[ALLOCA:.*]] = cir.alloca !rec_A, !cir.ptr<!rec_A>, ["agg.tmp0"] +// CIR: %[[ALLOCA:.*]] = cir.alloca !rec_A, !cir.ptr<!rec_A>, ["agg.tmp.ensured"] // CIR: cir.call @_ZN1AD1Ev(%[[ALLOCA]]) nothrow : (!cir.ptr<!rec_A>) -> () // LLVM: define dso_local void @_Z19test_temporary_dtorv(){{.*}} @@ -35,7 +35,7 @@ bool make_temp(const B &) { return false; } bool test_temp_or() { return make_temp(1) || make_temp(2); } // CIR: cir.func{{.*}} @_Z12test_temp_orv() -// CIR: %[[SCOPE:.*]] = cir.scope { +// CIR: cir.scope { // CIR: %[[REF_TMP0:.*]] = cir.alloca !rec_B, !cir.ptr<!rec_B>, ["ref.tmp0"] // CIR: %[[ONE:.*]] = cir.const #cir.int<1> // CIR: cir.call @_ZN1BC2Ei(%[[REF_TMP0]], %[[ONE]]) @@ -51,9 +51,9 @@ bool test_temp_or() { return make_temp(1) || make_temp(2); } // CIR: cir.call @_ZN1BD2Ev(%[[REF_TMP1]]) // CIR: cir.yield %[[MAKE_TEMP1]] : !cir.bool // CIR: }) +// CIR: cir.store{{.*}} %[[TERNARY]], %[[RETVAL:.*]] // CIR: cir.call @_ZN1BD2Ev(%[[REF_TMP0]]) -// CIR: cir.yield %[[TERNARY]] : !cir.bool -// CIR: } : !cir.bool +// CIR: } // LLVM: define{{.*}} i1 @_Z12test_temp_orv(){{.*}} { // LLVM: %[[REF_TMP0:.*]] = alloca %struct.B @@ -105,7 +105,7 @@ bool test_temp_or() { return make_temp(1) || make_temp(2); } bool test_temp_and() { return make_temp(1) && make_temp(2); } // CIR: cir.func{{.*}} @_Z13test_temp_andv() -// CIR: %[[SCOPE:.*]] = cir.scope { +// CIR: cir.scope { // CIR: %[[REF_TMP0:.*]] = cir.alloca !rec_B, !cir.ptr<!rec_B>, ["ref.tmp0"] // CIR: %[[ONE:.*]] = cir.const #cir.int<1> // CIR: cir.call @_ZN1BC2Ei(%[[REF_TMP0]], %[[ONE]]) @@ -121,9 +121,9 @@ bool test_temp_and() { return make_temp(1) && make_temp(2); } // CIR: %[[FALSE:.*]] = cir.const #false // CIR: cir.yield %[[FALSE]] : !cir.bool // CIR: }) +// CIR: cir.store{{.*}} %[[TERNARY]], %[[RETVAL:.*]] // CIR: cir.call @_ZN1BD2Ev(%[[REF_TMP0]]) -// CIR: cir.yield %[[TERNARY]] : !cir.bool -// CIR: } : !cir.bool +// CIR: } // LLVM: define{{.*}} i1 @_Z13test_temp_andv(){{.*}} { // LLVM: %[[REF_TMP0:.*]] = alloca %struct.B diff --git a/clang/test/CIR/CodeGen/lambda.cpp b/clang/test/CIR/CodeGen/lambda.cpp index 0c32ceb1..91380b9 100644 --- a/clang/test/CIR/CodeGen/lambda.cpp +++ b/clang/test/CIR/CodeGen/lambda.cpp @@ -219,14 +219,13 @@ int f() { // CIR: cir.func dso_local @_Z1fv() -> !s32i {{.*}} { // CIR: %[[RETVAL:.*]] = cir.alloca !s32i, !cir.ptr<!s32i>, ["__retval"] -// CIR: %[[SCOPE_RET:.*]] = cir.scope { +// CIR: cir.scope { // CIR: %[[TMP:.*]] = cir.alloca ![[REC_LAM_G2]], !cir.ptr<![[REC_LAM_G2]]>, ["ref.tmp0"] // CIR: %[[G2:.*]] = cir.call @_Z2g2v() : () -> ![[REC_LAM_G2]] // CIR: cir.store{{.*}} %[[G2]], %[[TMP]] // CIR: %[[RESULT:.*]] = cir.call @_ZZ2g2vENK3$_0clEv(%[[TMP]]) -// CIR: cir.yield %[[RESULT]] +// CIR: cir.store{{.*}} %[[RESULT]], %[[RETVAL]] // CIR: } -// CIR: cir.store{{.*}} %[[SCOPE_RET]], %[[RETVAL]] // CIR: %[[RET:.*]] = cir.load{{.*}} %[[RETVAL]] // CIR: cir.return %[[RET]] @@ -255,10 +254,9 @@ int f() { // LLVM: %[[G2:.*]] = call %[[REC_LAM_G2]] @_Z2g2v() // LLVM: store %[[REC_LAM_G2]] %[[G2]], ptr %[[TMP]] // LLVM: %[[RESULT:.*]] = call i32 @"_ZZ2g2vENK3$_0clEv"(ptr %[[TMP]]) +// LLVM: store i32 %[[RESULT]], ptr %[[RETVAL]] // LLVM: br label %[[RET_BB:.*]] // LLVM: [[RET_BB]]: -// LLVM: %[[RETPHI:.*]] = phi i32 [ %[[RESULT]], %[[SCOPE_BB]] ] -// LLVM: store i32 %[[RETPHI]], ptr %[[RETVAL]] // LLVM: %[[RET:.*]] = load i32, ptr %[[RETVAL]] // LLVM: ret i32 %[[RET]] @@ -333,14 +331,13 @@ struct A { // CIR: %[[RETVAL:.*]] = cir.alloca !s32i, !cir.ptr<!s32i>, ["__retval"] // CIR: cir.store %[[THIS_ARG]], %[[THIS_ADDR]] // CIR: %[[THIS]] = cir.load deref %[[THIS_ADDR]] : !cir.ptr<!cir.ptr<!rec_A>>, !cir.ptr<!rec_A> -// CIR: %[[SCOPE_RET:.*]] = cir.scope { +// CIR: cir.scope { // CIR: %[[LAM_ADDR:.*]] = cir.alloca ![[REC_LAM_A]], !cir.ptr<![[REC_LAM_A]]>, ["ref.tmp0"] // CIR: %[[STRUCT_A:.*]] = cir.get_member %[[LAM_ADDR]][0] {name = "this"} : !cir.ptr<![[REC_LAM_A]]> -> !cir.ptr<!rec_A> // CIR: cir.call @_ZN1AC1ERKS_(%[[STRUCT_A]], %[[THIS]]){{.*}} : (!cir.ptr<!rec_A>, !cir.ptr<!rec_A>){{.*}} -> () // CIR: %[[LAM_RET:.*]] = cir.call @_ZZN1A3fooEvENKUlvE_clEv(%[[LAM_ADDR]]) -// CIR: cir.yield %[[LAM_RET]] +// CIR: cir.store{{.*}} %[[LAM_RET]], %[[RETVAL]] // CIR: } -// CIR: cir.store{{.*}} %[[SCOPE_RET]], %[[RETVAL]] // CIR: %[[RET:.*]] = cir.load{{.*}} %[[RETVAL]] // CIR: cir.return %[[RET]] @@ -355,10 +352,9 @@ struct A { // LLVM: %[[STRUCT_A:.*]] = getelementptr %[[REC_LAM_A]], ptr %[[LAM_ALLOCA]], i32 0, i32 0 // LLVM: call void @_ZN1AC1ERKS_(ptr %[[STRUCT_A]], ptr %[[THIS]]) // LLVM: %[[LAM_RET:.*]] = call i32 @_ZZN1A3fooEvENKUlvE_clEv(ptr %[[LAM_ALLOCA]]) +// LLVM: store i32 %[[LAM_RET]], ptr %[[RETVAL]] // LLVM: br label %[[RET_BB:.*]] // LLVM: [[RET_BB]]: -// LLVM: %[[RETPHI:.*]] = phi i32 [ %[[LAM_RET]], %[[SCOPE_BB]] ] -// LLVM: store i32 %[[RETPHI]], ptr %[[RETVAL]] // LLVM: %[[RET:.*]] = load i32, ptr %[[RETVAL]] // LLVM: ret i32 %[[RET]] @@ -407,14 +403,13 @@ struct A { // CIR: %[[RETVAL:.*]] = cir.alloca !s32i, !cir.ptr<!s32i>, ["__retval"] // CIR: cir.store %[[THIS_ARG]], %[[THIS_ADDR]] // CIR: %[[THIS]] = cir.load %[[THIS_ADDR]] : !cir.ptr<!cir.ptr<!rec_A>>, !cir.ptr<!rec_A> -// CIR: %[[SCOPE_RET:.*]] = cir.scope { +// CIR: cir.scope { // CIR: %[[LAM_ADDR:.*]] = cir.alloca ![[REC_LAM_PTR_A]], !cir.ptr<![[REC_LAM_PTR_A]]>, ["ref.tmp0"] // CIR: %[[A_ADDR_ADDR:.*]] = cir.get_member %[[LAM_ADDR]][0] {name = "this"} : !cir.ptr<![[REC_LAM_PTR_A]]> -> !cir.ptr<!cir.ptr<!rec_A>> // CIR: cir.store{{.*}} %[[THIS]], %[[A_ADDR_ADDR]] // CIR: %[[LAM_RET:.*]] = cir.call @_ZZN1A3barEvENKUlvE_clEv(%[[LAM_ADDR]]) -// CIR: cir.yield %[[LAM_RET]] +// CIR: cir.store{{.*}} %[[LAM_RET]], %[[RETVAL]] // CIR: } -// CIR: cir.store{{.*}} %[[SCOPE_RET]], %[[RETVAL]] // CIR: %[[RET:.*]] = cir.load{{.*}} %[[RETVAL]] // CIR: cir.return %[[RET]] @@ -429,10 +424,9 @@ struct A { // LLVM: %[[A_ADDR_ADDR:.*]] = getelementptr %[[REC_LAM_PTR_A]], ptr %[[LAM_ALLOCA]], i32 0, i32 0 // LLVM: store ptr %[[THIS]], ptr %[[A_ADDR_ADDR]] // LLVM: %[[LAM_RET:.*]] = call i32 @_ZZN1A3barEvENKUlvE_clEv(ptr %[[LAM_ALLOCA]]) +// LLVM: store i32 %[[LAM_RET]], ptr %[[RETVAL]] // LLVM: br label %[[RET_BB:.*]] // LLVM: [[RET_BB]]: -// LLVM: %[[RETPHI:.*]] = phi i32 [ %[[LAM_RET]], %[[SCOPE_BB]] ] -// LLVM: store i32 %[[RETPHI]], ptr %[[RETVAL]] // LLVM: %[[RET:.*]] = load i32, ptr %[[RETVAL]] // LLVM: ret i32 %[[RET]] diff --git a/clang/test/CIR/CodeGen/new.cpp b/clang/test/CIR/CodeGen/new.cpp index 000ea5b..2efad10 100644 --- a/clang/test/CIR/CodeGen/new.cpp +++ b/clang/test/CIR/CodeGen/new.cpp @@ -208,6 +208,127 @@ void t_new_constant_size() { // OGCG: %[[CALL:.*]] = call noalias noundef nonnull ptr @_Znam(i64 noundef 128) // OGCG: store ptr %[[CALL]], ptr %[[P_ADDR]], align 8 +class C { + public: + ~C(); +}; + +void t_constant_size_nontrivial() { + auto p = new C[3]; +} + +// CHECK: cir.func{{.*}} @_Z26t_constant_size_nontrivialv() +// CHECK: %[[P_ADDR:.*]] = cir.alloca !cir.ptr<!rec_C>, !cir.ptr<!cir.ptr<!rec_C>>, ["p", init] {alignment = 8 : i64} +// CHECK: %[[#NUM_ELEMENTS:]] = cir.const #cir.int<3> : !u64i +// CHECK: %[[#SIZE_WITHOUT_COOKIE:]] = cir.const #cir.int<3> : !u64i +// CHECK: %[[#ALLOCATION_SIZE:]] = cir.const #cir.int<11> : !u64i +// CHECK: %[[RAW_PTR:.*]] = cir.call @_Znam(%[[#ALLOCATION_SIZE]]) : (!u64i) -> !cir.ptr<!void> +// CHECK: %[[COOKIE_PTR_BASE:.*]] = cir.cast bitcast %[[RAW_PTR]] : !cir.ptr<!void> -> !cir.ptr<!cir.ptr<!u8i>> +// CHECK: %[[COOKIE_PTR:.*]] = cir.cast bitcast %[[COOKIE_PTR_BASE]] : !cir.ptr<!cir.ptr<!u8i>> -> !cir.ptr<!u64i> +// CHECK: cir.store align(8) %[[#NUM_ELEMENTS]], %[[COOKIE_PTR]] : !u64i, !cir.ptr<!u64i> +// CHECK: %[[#COOKIE_SIZE:]] = cir.const #cir.int<8> : !s32i +// CHECK: %[[DATA_PTR_RAW:.*]] = cir.ptr_stride %[[COOKIE_PTR_BASE]], %[[#COOKIE_SIZE]] : (!cir.ptr<!cir.ptr<!u8i>>, !s32i) -> !cir.ptr<!cir.ptr<!u8i>> +// CHECK: %[[DATA_PTR_VOID:.*]] = cir.cast bitcast %[[DATA_PTR_RAW]] : !cir.ptr<!cir.ptr<!u8i>> -> !cir.ptr<!void> +// CHECK: %[[DATA_PTR:.*]] = cir.cast bitcast %[[DATA_PTR_VOID]] : !cir.ptr<!void> -> !cir.ptr<!rec_C> +// CHECK: cir.store align(8) %[[DATA_PTR]], %[[P_ADDR]] : !cir.ptr<!rec_C>, !cir.ptr<!cir.ptr<!rec_C>> +// CHECK: cir.return +// CHECK: } + +// LLVM: @_Z26t_constant_size_nontrivialv() +// LLVM: %[[ALLOCA:.*]] = alloca ptr, i64 1, align 8 +// LLVM: %[[COOKIE_PTR:.*]] = call ptr @_Znam(i64 11) +// LLVM: store i64 3, ptr %[[COOKIE_PTR]], align 8 +// LLVM: %[[ALLOCATED_PTR:.*]] = getelementptr ptr, ptr %[[COOKIE_PTR]], i64 8 +// LLVM: store ptr %[[ALLOCATED_PTR]], ptr %[[ALLOCA]], align 8 + +// OGCG: @_Z26t_constant_size_nontrivialv() +// OGCG: %[[ALLOCA:.*]] = alloca ptr, align 8 +// OGCG: %[[COOKIE_PTR:.*]] = call noalias noundef nonnull ptr @_Znam(i64 noundef 11) +// OGCG: store i64 3, ptr %[[COOKIE_PTR]], align 8 +// OGCG: %[[ALLOCATED_PTR:.*]] = getelementptr inbounds i8, ptr %[[COOKIE_PTR]], i64 8 +// OGCG: store ptr %[[ALLOCATED_PTR]], ptr %[[ALLOCA]], align 8 + +class D { + public: + int x; + ~D(); +}; + +void t_constant_size_nontrivial2() { + auto p = new D[3]; +} + +// In this test SIZE_WITHOUT_COOKIE isn't used, but it would be if there were +// an initializer. + +// CHECK: cir.func{{.*}} @_Z27t_constant_size_nontrivial2v() +// CHECK: %[[P_ADDR:.*]] = cir.alloca !cir.ptr<!rec_D>, !cir.ptr<!cir.ptr<!rec_D>>, ["p", init] {alignment = 8 : i64} +// CHECK: %[[#NUM_ELEMENTS:]] = cir.const #cir.int<3> : !u64i +// CHECK: %[[#SIZE_WITHOUT_COOKIE:]] = cir.const #cir.int<12> : !u64i +// CHECK: %[[#ALLOCATION_SIZE:]] = cir.const #cir.int<20> : !u64i +// CHECK: %[[RAW_PTR:.*]] = cir.call @_Znam(%[[#ALLOCATION_SIZE]]) : (!u64i) -> !cir.ptr<!void> +// CHECK: %[[COOKIE_PTR_BASE:.*]] = cir.cast bitcast %[[RAW_PTR]] : !cir.ptr<!void> -> !cir.ptr<!cir.ptr<!u8i>> +// CHECK: %[[COOKIE_PTR:.*]] = cir.cast bitcast %[[COOKIE_PTR_BASE]] : !cir.ptr<!cir.ptr<!u8i>> -> !cir.ptr<!u64i> +// CHECK: cir.store align(8) %[[#NUM_ELEMENTS]], %[[COOKIE_PTR]] : !u64i, !cir.ptr<!u64i> +// CHECK: %[[#COOKIE_SIZE:]] = cir.const #cir.int<8> : !s32i +// CHECK: %[[DATA_PTR_RAW:.*]] = cir.ptr_stride %[[COOKIE_PTR_BASE]], %[[#COOKIE_SIZE]] : (!cir.ptr<!cir.ptr<!u8i>>, !s32i) -> !cir.ptr<!cir.ptr<!u8i>> +// CHECK: %[[DATA_PTR_VOID:.*]] = cir.cast bitcast %[[DATA_PTR_RAW]] : !cir.ptr<!cir.ptr<!u8i>> -> !cir.ptr<!void> +// CHECK: %[[DATA_PTR:.*]] = cir.cast bitcast %[[DATA_PTR_VOID]] : !cir.ptr<!void> -> !cir.ptr<!rec_D> +// CHECK: cir.store align(8) %[[DATA_PTR]], %[[P_ADDR]] : !cir.ptr<!rec_D>, !cir.ptr<!cir.ptr<!rec_D>> +// CHECK: cir.return +// CHECK: } + +// LLVM: @_Z27t_constant_size_nontrivial2v() +// LLVM: %[[ALLOCA:.*]] = alloca ptr, i64 1, align 8 +// LLVM: %[[COOKIE_PTR:.*]] = call ptr @_Znam(i64 20) +// LLVM: store i64 3, ptr %[[COOKIE_PTR]], align 8 +// LLVM: %[[ALLOCATED_PTR:.*]] = getelementptr ptr, ptr %[[COOKIE_PTR]], i64 8 +// LLVM: store ptr %[[ALLOCATED_PTR]], ptr %[[ALLOCA]], align 8 + +struct alignas(16) E { + int x; + ~E(); +}; + +void t_align16_nontrivial() { + auto p = new E[2]; +} + +// CHECK: cir.func{{.*}} @_Z20t_align16_nontrivialv() +// CHECK: %[[P_ADDR:.*]] = cir.alloca !cir.ptr<!rec_E>, !cir.ptr<!cir.ptr<!rec_E>>, ["p", init] {alignment = 8 : i64} +// CHECK: %[[#NUM_ELEMENTS:]] = cir.const #cir.int<2> : !u64i +// CHECK: %[[#SIZE_WITHOUT_COOKIE:]] = cir.const #cir.int<32> : !u64i +// CHECK: %[[#ALLOCATION_SIZE:]] = cir.const #cir.int<48> : !u64i +// CHECK: %[[RAW_PTR:.*]] = cir.call @_Znam(%[[#ALLOCATION_SIZE]]) : (!u64i) -> !cir.ptr<!void> +// CHECK: %[[COOKIE_PTR_BASE:.*]] = cir.cast bitcast %[[RAW_PTR]] : !cir.ptr<!void> -> !cir.ptr<!cir.ptr<!u8i>> +// CHECK: %[[COOKIE_OFFSET:.*]] = cir.const #cir.int<8> : !s32i +// CHECK: %[[COOKIE_PTR_RAW:.*]] = cir.ptr_stride %[[COOKIE_PTR_BASE]], %[[COOKIE_OFFSET]] : (!cir.ptr<!cir.ptr<!u8i>>, !s32i) -> !cir.ptr<!cir.ptr<!u8i>> +// CHECK: %[[COOKIE_PTR:.*]] = cir.cast bitcast %[[COOKIE_PTR_RAW]] : !cir.ptr<!cir.ptr<!u8i>> -> !cir.ptr<!u64i> +// CHECK: cir.store align(8) %[[#NUM_ELEMENTS]], %[[COOKIE_PTR]] : !u64i, !cir.ptr<!u64i> +// CHECK: %[[#COOKIE_SIZE:]] = cir.const #cir.int<16> : !s32i +// CHECK: %[[DATA_PTR_RAW:.*]] = cir.ptr_stride %[[COOKIE_PTR_BASE]], %[[#COOKIE_SIZE]] : (!cir.ptr<!cir.ptr<!u8i>>, !s32i) -> !cir.ptr<!cir.ptr<!u8i>> +// CHECK: %[[DATA_PTR_VOID:.*]] = cir.cast bitcast %[[DATA_PTR_RAW]] : !cir.ptr<!cir.ptr<!u8i>> -> !cir.ptr<!void> +// CHECK: %[[DATA_PTR:.*]] = cir.cast bitcast %[[DATA_PTR_VOID]] : !cir.ptr<!void> -> !cir.ptr<!rec_E> +// CHECK: cir.store align(8) %[[DATA_PTR]], %[[P_ADDR]] : !cir.ptr<!rec_E>, !cir.ptr<!cir.ptr<!rec_E>> +// CHECK: cir.return +// CHECK: } + +// LLVM: @_Z20t_align16_nontrivialv() +// LLVM: %[[ALLOCA:.*]] = alloca ptr, i64 1, align 8 +// LLVM: %[[RAW_PTR:.*]] = call ptr @_Znam(i64 48) +// LLVM: %[[COOKIE_PTR:.*]] = getelementptr ptr, ptr %[[RAW_PTR]], i64 8 +// LLVM: store i64 2, ptr %[[COOKIE_PTR]], align 8 +// LLVM: %[[ALLOCATED_PTR:.*]] = getelementptr ptr, ptr %[[RAW_PTR]], i64 16 +// LLVM: store ptr %[[ALLOCATED_PTR]], ptr %[[ALLOCA]], align 8 + +// OGCG: define{{.*}} void @_Z20t_align16_nontrivialv +// OGCG: %[[ALLOCA:.*]] = alloca ptr, align 8 +// OGCG: %[[RAW_PTR:.*]] = call noalias noundef nonnull ptr @_Znam(i64 noundef 48) +// OGCG: %[[COOKIE_PTR:.*]] = getelementptr inbounds i8, ptr %[[RAW_PTR]], i64 8 +// OGCG: store i64 2, ptr %[[COOKIE_PTR]], align 8 +// OGCG: %[[ALLOCATED_PTR:.*]] = getelementptr inbounds i8, ptr %[[RAW_PTR]], i64 16 +// OGCG: store ptr %[[ALLOCATED_PTR]], ptr %[[ALLOCA]], align 8 +// OGCG: ret void void t_new_multidim_constant_size() { auto p = new double[2][3][4]; diff --git a/clang/test/CIR/CodeGen/statement-exprs.c b/clang/test/CIR/CodeGen/statement-exprs.c index f6ec9ec..c784ec9 100644 --- a/clang/test/CIR/CodeGen/statement-exprs.c +++ b/clang/test/CIR/CodeGen/statement-exprs.c @@ -218,7 +218,7 @@ struct S { int x; }; int test3() { return ({ struct S s = {1}; s; }).x; } // CIR: cir.func no_proto dso_local @test3() -> !s32i // CIR: %[[RETVAL:.+]] = cir.alloca !s32i, !cir.ptr<!s32i>, ["__retval"] -// CIR: %[[YIELDVAL:.+]] = cir.scope { +// CIR: cir.scope { // CIR: %[[REF_TMP0:.+]] = cir.alloca !rec_S, !cir.ptr<!rec_S>, ["ref.tmp0"] // CIR: %[[TMP:.+]] = cir.alloca !rec_S, !cir.ptr<!rec_S>, ["tmp"] // CIR: cir.scope { @@ -230,9 +230,8 @@ int test3() { return ({ struct S s = {1}; s; }).x; } // CIR: } // CIR: %[[GEP_X_TMP:.+]] = cir.get_member %[[REF_TMP0]][0] {name = "x"} : !cir.ptr<!rec_S> -> !cir.ptr<!s32i> // CIR: %[[XVAL:.+]] = cir.load {{.*}} %[[GEP_X_TMP]] : !cir.ptr<!s32i>, !s32i -// CIR: cir.yield %[[XVAL]] : !s32i -// CIR: } : !s32i -// CIR: cir.store %[[YIELDVAL]], %[[RETVAL]] : !s32i, !cir.ptr<!s32i> +// CIR: cir.store{{.*}} %[[XVAL]], %[[RETVAL]] : !s32i, !cir.ptr<!s32i> +// CIR: } // CIR: %[[RES:.+]] = cir.load %[[RETVAL]] : !cir.ptr<!s32i>, !s32i // CIR: cir.return %[[RES]] : !s32i @@ -252,10 +251,9 @@ int test3() { return ({ struct S s = {1}; s; }).x; } // LLVM: [[LBL8]]: // LLVM: %[[GEP_VAR1:.+]] = getelementptr %struct.S, ptr %[[VAR1]], i32 0, i32 0 // LLVM: %[[LOAD_X:.+]] = load i32, ptr %[[GEP_VAR1]] +// LLVM: store i32 %[[LOAD_X]], ptr %[[VAR4]] // LLVM: br label %[[LBL11:.+]] // LLVM: [[LBL11]]: -// LLVM: %[[PHI:.+]] = phi i32 [ %[[LOAD_X]], %[[LBL8]] ] -// LLVM: store i32 %[[PHI]], ptr %[[VAR4]] // LLVM: %[[RES:.+]] = load i32, ptr %[[VAR4]] // LLVM: ret i32 %[[RES]] diff --git a/clang/test/CIR/CodeGen/struct-init.cpp b/clang/test/CIR/CodeGen/struct-init.cpp index a47ef53..cb50999 100644 --- a/clang/test/CIR/CodeGen/struct-init.cpp +++ b/clang/test/CIR/CodeGen/struct-init.cpp @@ -9,6 +9,22 @@ struct S { int a, b, c; }; +S partial_init = { 1 }; + +// CIR: cir.global external @partial_init = #cir.const_record<{#cir.int<1> : !s32i, #cir.int<0> : !s32i, #cir.int<0> : !s32i}> : !rec_S +// LLVM: @partial_init = global %struct.S { i32 1, i32 0, i32 0 } +// OGCG: @partial_init = global %struct.S { i32 1, i32 0, i32 0 } + +struct StructWithDefaultInit { + int a = 2; +}; + +StructWithDefaultInit swdi = {}; + +// CIR: cir.global external @swdi = #cir.const_record<{#cir.int<2> : !s32i}> : !rec_StructWithDefaultInit +// LLVM: @swdi = global %struct.StructWithDefaultInit { i32 2 }, align 4 +// OGCG: @swdi = global %struct.StructWithDefaultInit { i32 2 }, align 4 + void init() { S s1 = {1, 2, 3}; S s2 = {4, 5}; diff --git a/clang/test/CIR/CodeGen/struct.cpp b/clang/test/CIR/CodeGen/struct.cpp index 263799f..c8db714 100644 --- a/clang/test/CIR/CodeGen/struct.cpp +++ b/clang/test/CIR/CodeGen/struct.cpp @@ -265,7 +265,7 @@ void bin_comma() { // CIR: cir.func{{.*}} @_Z9bin_commav() // CIR: %[[A_ADDR:.*]] = cir.alloca !rec_CompleteS, !cir.ptr<!rec_CompleteS>, ["a", init] -// CIR: %[[TMP_ADDR:.*]] = cir.alloca !rec_CompleteS, !cir.ptr<!rec_CompleteS>, ["agg.tmp0"] +// CIR: %[[TMP_ADDR:.*]] = cir.alloca !rec_CompleteS, !cir.ptr<!rec_CompleteS>, ["agg.tmp.ensured"] // CIR: %[[ZERO:.*]] = cir.const #cir.zero : !rec_CompleteS // CIR: cir.store{{.*}} %[[ZERO]], %[[TMP_ADDR]] : !rec_CompleteS, !cir.ptr<!rec_CompleteS> // CIR: %[[ZERO:.*]] = cir.const #cir.zero : !rec_CompleteS @@ -280,3 +280,67 @@ void bin_comma() { // OGCG: define{{.*}} void @_Z9bin_commav() // OGCG: %[[A_ADDR:.*]] = alloca %struct.CompleteS, align 4 // OGCG: call void @llvm.memset.p0.i64(ptr align 4 %[[A_ADDR]], i8 0, i64 8, i1 false) + +void compound_literal_expr() { CompleteS a = (CompleteS){}; } + +// CIR: %[[A_ADDR:.*]] = cir.alloca !rec_CompleteS, !cir.ptr<!rec_CompleteS>, ["a", init] +// CIR: %[[A_ELEM_0_PTR:.*]] = cir.get_member %[[A_ADDR]][0] {name = "a"} : !cir.ptr<!rec_CompleteS> -> !cir.ptr<!s32i> +// CIR: %[[CONST_0:.*]] = cir.const #cir.int<0> : !s32i +// CIR: cir.store{{.*}} %[[CONST_0]], %[[A_ELEM_0_PTR]] : !s32i, !cir.ptr<!s32i> +// CIR: %[[A_ELEM_1_PTR:.*]] = cir.get_member %[[A_ADDR]][1] {name = "b"} : !cir.ptr<!rec_CompleteS> -> !cir.ptr<!s8i> +// CIR: %[[CONST_0:.*]] = cir.const #cir.int<0> : !s8i +// CIR: cir.store{{.*}} %[[CONST_0]], %[[A_ELEM_1_PTR]] : !s8i, !cir.ptr<!s8i> + +// TODO(cir): zero-initialize the padding + +// LLVM: %[[A_ADDR:.*]] = alloca %struct.CompleteS, i64 1, align 4 +// LLVM: %[[A_ELEM_0_PTR:.*]] = getelementptr %struct.CompleteS, ptr %[[A_ADDR]], i32 0, i32 0 +// LLVM: store i32 0, ptr %[[A_ELEM_0_PTR]], align 4 +// LLVM: %[[A_ELEM_1_PTR:.*]] = getelementptr %struct.CompleteS, ptr %[[A_ADDR]], i32 0, i32 1 +// LLVM: store i8 0, ptr %[[A_ELEM_1_PTR]], align 4 + +// OGCG: %[[A_ADDR:.*]] = alloca %struct.CompleteS, align 4 +// OGCG: call void @llvm.memset.p0.i64(ptr align 4 %[[A_ADDR]], i8 0, i64 8, i1 false) + +struct StructWithConstMember { + int a : 1; +}; + +void struct_with_const_member_expr() { + int a = (StructWithConstMember){}.a; +} + +// CIR: %[[A_ADDR:.*]] = cir.alloca !s32i, !cir.ptr<!s32i>, ["a", init] +// CIR: %[[RESULT:.*]] = cir.scope { +// CIR: %[[REF_ADDR:.*]] = cir.alloca !rec_StructWithConstMember, !cir.ptr<!rec_StructWithConstMember>, ["ref.tmp0"] +// CIR: %[[ELEM_0_PTR:.*]] = cir.get_member %[[REF_ADDR]][0] {name = "a"} : !cir.ptr<!rec_StructWithConstMember> -> !cir.ptr<!u8i> +// CIR: %[[CONST_0:.*]] = cir.const #cir.int<0> : !s32i +// CIR: %[[SET_BF:.*]] = cir.set_bitfield{{.*}} (#bfi_a, %[[ELEM_0_PTR]] : !cir.ptr<!u8i>, %[[CONST_0]] : !s32i) -> !s32i +// CIR: %[[CONST_0:.*]] = cir.const #cir.int<0> : !s32i +// CIR: cir.yield %[[CONST_0]] : !s32i +// CIR: } : !s32i +// CIR: cir.store{{.*}} %[[RESULT]], %[[A_ADDR]] : !s32i, !cir.ptr<!s32i> + +// TODO(cir): zero-initialize the padding + +// LLVM: %[[REF_ADDR:.*]] = alloca %struct.StructWithConstMember, i64 1, align 4 +// LLVM: %[[A_ADDR:.*]] = alloca i32, i64 1, align 4 +// LLVM: br label %[[BF_LABEL:.*]] +// LLVM: [[BF_LABEL]]: +// LLVM: %[[ELEM_0_PTR:.*]] = getelementptr %struct.StructWithConstMember, ptr %[[REF_ADDR]], i32 0, i32 0 +// LLVM: %[[TMP_REF:.*]] = load i8, ptr %[[ELEM_0_PTR]], align 4 +// LLVM: %[[BF_CLEAR:.*]] = and i8 %[[TMP_REF]], -2 +// LLVM: %[[BF_SET:.*]] = or i8 %[[BF_CLEAR]], 0 +// LLVM: store i8 %[[BF_SET]], ptr %[[ELEM_0_PTR]], align 4 +// LLVM: br label %[[RESULT_LABEL:.*]] +// LLVM: [[RESULT_LABEL]]: +// LLVM: %[[RESULT:.*]] = phi i32 [ 0, %[[BF_LABEL]] ] +// LLVM: store i32 %[[RESULT]], ptr %[[A_ADDR]], align 4 + +// OGCG: %[[A_ADDR:.*]] = alloca i32, align 4 +// OGCG: %[[REF_ADDR:.*]] = alloca %struct.StructWithConstMember, align 4 +// OGCG: %[[TMP_REF:.*]] = load i8, ptr %[[REF_ADDR]], align 4 +// OGCG: %[[BF_CLEAR:.*]] = and i8 %[[TMP_REF]], -2 +// OGCG: %[[BF_SET:.*]] = or i8 %[[BF_CLEAR]], 0 +// OGCG: store i8 %[[BF_SET]], ptr %[[REF_ADDR]], align 4 +// OGCG: store i32 0, ptr %[[A_ADDR]], align 4 diff --git a/clang/test/CIR/CodeGen/try-catch.cpp b/clang/test/CIR/CodeGen/try-catch.cpp index 8f0b3c4..5a50310 100644 --- a/clang/test/CIR/CodeGen/try-catch.cpp +++ b/clang/test/CIR/CodeGen/try-catch.cpp @@ -30,3 +30,90 @@ void empty_try_block_with_catch_with_int_exception() { // OGCG: define{{.*}} void @_Z45empty_try_block_with_catch_with_int_exceptionv() // OGCG: ret void + +void try_catch_with_empty_catch_all() { + int a = 1; + try { + return; + ++a; + } catch (...) { + } +} + +// CIR: %[[A_ADDR:.*]] = cir.alloca !s32i, !cir.ptr<!s32i>, ["a", init] +// CIR: %[[CONST_1:.*]] = cir.const #cir.int<1> : !s32i +// CIR: cir.store{{.*}} %[[CONST_1]], %[[A_ADDR]] : !s32i, !cir.ptr<!s32i +// CIR: cir.scope { +// CIR: cir.try { +// CIR: cir.return +// CIR: ^bb1: // no predecessors +// CIR: %[[TMP_A:.*]] = cir.load{{.*}} %[[A_ADDR]] : !cir.ptr<!s32i>, !s32i +// CIR: %[[RESULT:.*]] = cir.unary(inc, %[[TMP_A]]) nsw : !s32i, !s32i +// CIR: cir.store{{.*}} %[[RESULT]], %[[A_ADDR]] : !s32i, !cir.ptr<!s32i> +// CIR: cir.yield +// CIR: } +// CIR: } + +// LLVM: %[[A_ADDR:.*]] = alloca i32, i64 1, align 4 +// LLVM: store i32 1, ptr %[[A_ADDR]], align 4 +// LLVM: br label %[[BB_2:.*]] +// LLVM: [[BB_2]]: +// LLVM: br label %[[BB_3:.*]] +// LLVM: [[BB_3]]: +// LLVM: ret void +// LLVM: [[BB_4:.*]]: +// LLVM: %[[TMP_A:.*]] = load i32, ptr %[[A_ADDR]], align 4 +// LLVM: %[[RESULT:.*]] = add nsw i32 %[[TMP_A]], 1 +// LLVM: store i32 %[[RESULT]], ptr %[[A_ADDR]], align 4 +// LLVM: br label %[[BB_7:.*]] +// LLVM: [[BB_7]]: +// LLVM: br label %[[BB_8:.*]] +// LLVM: [[BB_8]]: +// LLVM: ret void + +// OGCG: %[[A_ADDR:.*]] = alloca i32, align 4 +// OGCG: store i32 1, ptr %[[A_ADDR]], align 4 +// OGCG: ret void + +void try_catch_with_empty_catch_all_2() { + int a = 1; + try { + ++a; + return; + } catch (...) { + } +} + +// CIR: %[[A_ADDR:.*]] = cir.alloca !s32i, !cir.ptr<!s32i>, ["a", init] +// CIR: %[[CONST_1:.*]] = cir.const #cir.int<1> : !s32i +// CIR: cir.store{{.*}} %[[CONST_1]], %[[A_ADDR]] : !s32i, !cir.ptr<!s32i> +// CIR: cir.scope { +// CIR: cir.try { +// CIR: %[[TMP_A:.*]] = cir.load{{.*}} %[[A_ADDR]] : !cir.ptr<!s32i>, !s32i +// CIR: %[[RESULT:.*]] = cir.unary(inc, %[[TMP_A]]) nsw : !s32i, !s32i +// CIR: cir.store{{.*}} %[[RESULT]], %[[A_ADDR]] : !s32i, !cir.ptr<!s32i> +// CIR: cir.return +// CIR: } +// CIR: } + +// LLVM: %[[A_ADDR]] = alloca i32, i64 1, align 4 +// LLVM: store i32 1, ptr %[[A_ADDR]], align 4 +// LLVM: br label %[[BB_2:.*]] +// LLVM: [[BB_2]]: +// LLVM: br label %[[BB_3:.*]] +// LLVM: [[BB_3]]: +// LLVM: %[[TMP_A:.*]] = load i32, ptr %[[A_ADDR]], align 4 +// LLVM: %[[RESULT:.*]] = add nsw i32 %[[TMP_A:.*]], 1 +// LLVM: store i32 %[[RESULT]], ptr %[[A_ADDR]], align 4 +// LLVM: ret void +// LLVM: [[BB_6:.*]]: +// LLVM: br label %[[BB_7:.*]] +// LLVM: [[BB_7]]: +// LLVM: ret void + +// OGCG: %[[A_ADDR:.*]] = alloca i32, align 4 +// OGCG: store i32 1, ptr %[[A_ADDR]], align 4 +// OGCG: %[[TMP_A:.*]] = load i32, ptr %[[A_ADDR]], align 4 +// OGCG: %[[RESULT:.*]] = add nsw i32 %[[TMP_A]], 1 +// OGCG: store i32 %[[RESULT]], ptr %[[A_ADDR]], align 4 +// OGCG: ret void diff --git a/clang/test/CIR/CodeGen/vla.c b/clang/test/CIR/CodeGen/vla.c index b22c704..0af4f83 100644 --- a/clang/test/CIR/CodeGen/vla.c +++ b/clang/test/CIR/CodeGen/vla.c @@ -282,4 +282,61 @@ void f3(unsigned len) { // break; // } // } -
\ No newline at end of file + +int f5(unsigned long len) { + int arr[len]; + return arr[2]; +} + +// CIR: cir.func{{.*}} @f5(%[[LEN_ARG:.*]]: !u64i {{.*}}) -> !s32i +// CIR: %[[LEN_ADDR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["len", init] +// CIR: %[[RET_ADDR:.*]] = cir.alloca !s32i, !cir.ptr<!s32i>, ["__retval"] +// CIR: %[[SAVED_STACK:.*]] = cir.alloca !cir.ptr<!u8i>, !cir.ptr<!cir.ptr<!u8i>>, ["saved_stack"] +// CIR: cir.store{{.*}} %[[LEN_ARG]], %[[LEN_ADDR]] +// CIR: %[[LEN:.*]] = cir.load{{.*}} %[[LEN_ADDR]] +// CIR: %[[STACK_PTR:.*]] = cir.stacksave +// CIR: cir.store{{.*}} %[[STACK_PTR]], %[[SAVED_STACK]] +// CIR: %[[ARR:.*]] = cir.alloca !s32i, !cir.ptr<!s32i>, %[[LEN]] : !u64i, ["arr"] +// CIR: %[[TWO:.*]] = cir.const #cir.int<2> : !s32i +// CIR: %[[ARR_2:.*]] = cir.ptr_stride %[[ARR]], %[[TWO]] +// CIR: %[[ARR_VAL:.*]] = cir.load{{.*}} %[[ARR_2]] : !cir.ptr<!s32i>, !s32i +// CIR: cir.store{{.*}} %[[ARR_VAL]], %[[RET_ADDR]] : !s32i, !cir.ptr<!s32i> +// CIR: %[[STACK_RESTORE_PTR:.*]] = cir.load{{.*}} %[[SAVED_STACK]] +// CIR: cir.stackrestore %[[STACK_RESTORE_PTR]] +// CIR: %[[RET_VAL:.*]] = cir.load{{.*}} %[[RET_ADDR]] +// CIR: cir.return %[[RET_VAL]] : !s32i + +// LLVM: define{{.*}} i32 @f5(i64 %[[LEN_ARG:.*]]) +// LLVM: %[[LEN_ADDR:.*]] = alloca i64 +// LLVM: %[[RET_ADDR:.*]] = alloca i32 +// LLVM: %[[SAVED_STACK:.*]] = alloca ptr +// LLVM: store i64 %[[LEN_ARG]], ptr %[[LEN_ADDR]] +// LLVM: %[[LEN:.*]] = load i64, ptr %[[LEN_ADDR]] +// LLVM: %[[STACK_PTR:.*]] = call ptr @llvm.stacksave.p0() +// LLVM: store ptr %[[STACK_PTR]], ptr %[[SAVED_STACK]] +// LLVM: %[[ARR:.*]] = alloca i32, i64 %[[LEN]] +// LLVM: %[[ARR_2:.*]] = getelementptr i32, ptr %[[ARR]], i64 2 +// LLVM: %[[ARR_VAL:.*]] = load i32, ptr %[[ARR_2]] +// LLVM: store i32 %[[ARR_VAL]], ptr %[[RET_ADDR]] +// LLVM: %[[STACK_RESTORE_PTR:.*]] = load ptr, ptr %[[SAVED_STACK]] +// LLVM: call void @llvm.stackrestore.p0(ptr %[[STACK_RESTORE_PTR]]) +// LLVM: %[[RET_VAL:.*]] = load i32, ptr %[[RET_ADDR]] +// LLVM: ret i32 %[[RET_VAL]] + +// Note: VLA_EXPR0 below is emitted to capture debug info. + +// OGCG: define{{.*}} i32 @f5(i64 {{.*}} %[[LEN_ARG:.*]]) +// OGCG: %[[LEN_ADDR:.*]] = alloca i64 +// OGCG: %[[SAVED_STACK:.*]] = alloca ptr +// OGCG: %[[VLA_EXPR0:.*]] = alloca i64 +// OGCG: store i64 %[[LEN_ARG]], ptr %[[LEN_ADDR]] +// OGCG: %[[LEN:.*]] = load i64, ptr %[[LEN_ADDR]] +// OGCG: %[[STACK_PTR:.*]] = call ptr @llvm.stacksave.p0() +// OGCG: store ptr %[[STACK_PTR]], ptr %[[SAVED_STACK]] +// OGCG: %[[ARR:.*]] = alloca i32, i64 %[[LEN]] +// OGCG: store i64 %[[LEN]], ptr %[[VLA_EXPR0]] +// OGCG: %[[ARR_2:.*]] = getelementptr inbounds i32, ptr %[[ARR]], i64 2 +// OGCG: %[[ARR_VAL:.*]] = load i32, ptr %[[ARR_2]] +// OGCG: %[[STACK_RESTORE_PTR:.*]] = load ptr, ptr %[[SAVED_STACK]] +// OGCG: call void @llvm.stackrestore.p0(ptr %[[STACK_RESTORE_PTR]]) +// OGCG: ret i32 %[[ARR_VAL]] diff --git a/clang/test/CIR/CodeGenOpenACC/atomic-read.cpp b/clang/test/CIR/CodeGenOpenACC/atomic-read.cpp new file mode 100644 index 0000000..9882f05 --- /dev/null +++ b/clang/test/CIR/CodeGenOpenACC/atomic-read.cpp @@ -0,0 +1,24 @@ +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s + +void use(int x, unsigned int y, float f) { + // CHECK: cir.func{{.*}}(%[[X_ARG:.*]]: !s32i{{.*}}, %[[Y_ARG:.*]]: !u32i{{.*}}, %[[F_ARG:.*]]: !cir.float{{.*}}){{.*}}{ + // CHECK-NEXT: %[[X_ALLOC:.*]] = cir.alloca !s32i, !cir.ptr<!s32i>, ["x", init] + // CHECK-NEXT: %[[Y_ALLOC:.*]] = cir.alloca !u32i, !cir.ptr<!u32i>, ["y", init] + // CHECK-NEXT: %[[F_ALLOC:.*]] = cir.alloca !cir.float, !cir.ptr<!cir.float>, ["f", init] + // CHECK-NEXT: cir.store %[[X_ARG]], %[[X_ALLOC]] : !s32i, !cir.ptr<!s32i> + // CHECK-NEXT: cir.store %[[Y_ARG]], %[[Y_ALLOC]] : !u32i, !cir.ptr<!u32i> + // CHECK-NEXT: cir.store %[[F_ARG]], %[[F_ALLOC]] : !cir.float, !cir.ptr<!cir.float> + + // CHECK-NEXT: acc.atomic.read %[[X_ALLOC]] = %[[Y_ALLOC]] : !cir.ptr<!s32i>, !cir.ptr<!u32i>, !s32i +#pragma acc atomic read + x = y; + + // CHECK-NEXT: %[[X_LOAD:.*]] = cir.load{{.*}} %[[X_ALLOC]] : !cir.ptr<!s32i>, !s32i + // CHECK-NEXT: %[[X_CAST:.*]] = cir.cast integral %[[X_LOAD]] : !s32i -> !u32i + // CHECK-NEXT: %[[Y_LOAD:.*]] = cir.load{{.*}} %[[Y_ALLOC]] : !cir.ptr<!u32i>, !u32i + // CHECK-NEXT: %[[CMP:.*]] = cir.cmp(eq, %[[X_CAST]], %[[Y_LOAD]]) : !u32i, !cir.bool + // CHECK-NEXT: %[[CMP_CAST:.*]] = builtin.unrealized_conversion_cast %[[CMP]] : !cir.bool to i1 + // CHECK-NEXT: acc.atomic.read if(%[[CMP_CAST]]) %[[F_ALLOC]] = %[[Y_ALLOC]] : !cir.ptr<!cir.float>, !cir.ptr<!u32i>, !cir.float +#pragma acc atomic read if (x == y) + f = y; +} diff --git a/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-default-ops.cpp b/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-default-ops.cpp index c1c2e4b..53eba7b 100644 --- a/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-default-ops.cpp +++ b/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-default-ops.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s struct DefaultOperators { int i; @@ -480,7 +480,77 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_DefaultOperators> // CHECK-NEXT: } for(int i = 0; i < 5; ++i); @@ -507,7 +577,77 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_DefaultOperators> // CHECK-NEXT: } for(int i = 0; i < 5; ++i); @@ -1532,7 +1672,101 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } for(int i = 0; i < 5; ++i); @@ -1576,7 +1810,101 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } for(int i = 0; i < 5; ++i); @@ -2398,6 +2726,104 @@ void acc_combined() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } for(int i = 0; i < 5; ++i); @@ -2446,6 +2872,104 @@ void acc_combined() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } for(int i = 0; i < 5; ++i); diff --git a/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-float.cpp b/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-float.cpp index 853f345..63d6952 100644 --- a/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-float.cpp +++ b/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-float.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s template<typename T> void acc_combined() { T someVar; @@ -92,7 +92,18 @@ void acc_combined() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !cir.float, !cir.ptr<!cir.float> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.float> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -106,7 +117,18 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !cir.float, !cir.ptr<!cir.float> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.float> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -371,7 +393,41 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -401,7 +457,41 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -708,7 +798,45 @@ void acc_combined() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -744,7 +872,45 @@ void acc_combined() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); diff --git a/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-inline-ops.cpp b/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-inline-ops.cpp index 67e8460..9c1b161 100644 --- a/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-inline-ops.cpp +++ b/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-inline-ops.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s struct HasOperatorsInline { int i; @@ -14,8 +14,8 @@ struct HasOperatorsInline { HasOperatorsInline &operator&=(HasOperatorsInline& other); HasOperatorsInline &operator|=(HasOperatorsInline& other); HasOperatorsInline &operator^=(HasOperatorsInline& other); - bool &operator&&(HasOperatorsInline& other); - bool &operator||(HasOperatorsInline& other); + HasOperatorsInline &operator&&(HasOperatorsInline& other); + HasOperatorsInline &operator||(HasOperatorsInline& other); // For min/max bool operator<(HasOperatorsInline& other); HasOperatorsInline &operator=(HasOperatorsInline& other); @@ -277,7 +277,8 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineaaERS_(%[[LHSARG]], %[[RHSARG]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHSARG]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_HasOperatorsInline> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}, %[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}): @@ -286,7 +287,7 @@ void acc_combined() { // CHECK-NEXT: } for(int i=0;i < 5; ++i); #pragma acc parallel loop reduction(||:someVar) -// CHECK-NEXT: acc.reduction.recipe @reduction_lor__ZTS18HasOperatorsInline : !cir.ptr<!rec_HasOperatorsInline> reduction_operator <lor> init { +// CHECK: acc.reduction.recipe @reduction_lor__ZTS18HasOperatorsInline : !cir.ptr<!rec_HasOperatorsInline> reduction_operator <lor> init { // CHECK-NEXT: ^bb0(%[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsInline>{{.*}}) // CHECK-NEXT: %[[ALLOCA:.*]] = cir.alloca !rec_HasOperatorsInline, !cir.ptr<!rec_HasOperatorsInline>, ["openacc.reduction.init", init] // CHECK-NEXT: %[[GET_I:.*]] = cir.get_member %[[ALLOCA]][0] {name = "i"} : !cir.ptr<!rec_HasOperatorsInline> -> !cir.ptr<!s32i> @@ -308,7 +309,8 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineooERS_(%[[LHSARG]], %[[RHSARG]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHSARG]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_HasOperatorsInline> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}, %[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}): @@ -318,7 +320,7 @@ void acc_combined() { for(int i=0;i < 5; ++i); #pragma acc parallel loop reduction(+:someVarArr) -// CHECK-NEXT: acc.reduction.recipe @reduction_add__ZTSA5_18HasOperatorsInline : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> reduction_operator <add> init { +// CHECK: acc.reduction.recipe @reduction_add__ZTSA5_18HasOperatorsInline : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> reduction_operator <add> init { // CHECK-NEXT: ^bb0(%[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>>{{.*}}) // CHECK-NEXT: %[[ALLOCA:.*]] = cir.alloca !cir.array<!rec_HasOperatorsInline x 5>, !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>>, ["openacc.reduction.init", init] // CHECK-NEXT: %[[TEMP_ITR:.*]] = cir.alloca !cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!cir.ptr<!rec_HasOperatorsInline>>, ["arrayinit.temp"] @@ -1254,7 +1256,31 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !s64i) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !s64i) -> !cir.ptr<!rec_HasOperatorsInline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineaaERS_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}): @@ -1318,7 +1344,31 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !s64i) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !s64i) -> !cir.ptr<!rec_HasOperatorsInline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineooERS_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}): @@ -2121,6 +2171,35 @@ void acc_combined() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !u64i) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !u64i) -> !cir.ptr<!rec_HasOperatorsInline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineaaERS_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}}): @@ -2198,6 +2277,35 @@ void acc_combined() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !u64i) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !u64i) -> !cir.ptr<!rec_HasOperatorsInline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineooERS_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}}): diff --git a/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-int.cpp b/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-int.cpp index d74de82..78b43dd 100644 --- a/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-int.cpp +++ b/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-int.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s template<typename T> void acc_combined() { @@ -145,7 +145,18 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!s32i> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!s32i> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !s32i, !cir.ptr<!s32i> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!s32i> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -159,7 +170,18 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!s32i> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!s32i> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !s32i, !cir.ptr<!s32i> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!s32i> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -587,7 +609,41 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -617,7 +673,41 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -1116,7 +1206,45 @@ void acc_combined() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -1152,7 +1280,45 @@ void acc_combined() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); diff --git a/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-outline-ops.cpp b/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-outline-ops.cpp index a6df6c0..5b37071 100644 --- a/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-outline-ops.cpp +++ b/clang/test/CIR/CodeGenOpenACC/combined-reduction-clause-outline-ops.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s struct HasOperatorsOutline { int i; unsigned u; @@ -15,8 +15,8 @@ HasOperatorsOutline &operator*=(HasOperatorsOutline &, HasOperatorsOutline &); HasOperatorsOutline &operator&=(HasOperatorsOutline &, HasOperatorsOutline &); HasOperatorsOutline &operator|=(HasOperatorsOutline &, HasOperatorsOutline &); HasOperatorsOutline &operator^=(HasOperatorsOutline &, HasOperatorsOutline &); -bool &operator&&(HasOperatorsOutline &, HasOperatorsOutline &); -bool &operator||(HasOperatorsOutline &, HasOperatorsOutline &); +HasOperatorsOutline &operator&&(HasOperatorsOutline &, HasOperatorsOutline &); +HasOperatorsOutline &operator||(HasOperatorsOutline &, HasOperatorsOutline &); // For min/max bool operator<(HasOperatorsOutline &, HasOperatorsOutline &); @@ -276,7 +276,8 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZaaR19HasOperatorsOutlineS0_(%[[LHSARG]], %[[RHSARG]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHSARG]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_HasOperatorsOutline> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}, %[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}): @@ -285,7 +286,7 @@ void acc_combined() { // CHECK-NEXT: } for(int i=0;i < 5; ++i); #pragma acc parallel loop reduction(||:someVar) -// CHECK-NEXT: acc.reduction.recipe @reduction_lor__ZTS19HasOperatorsOutline : !cir.ptr<!rec_HasOperatorsOutline> reduction_operator <lor> init { +// CHECK: acc.reduction.recipe @reduction_lor__ZTS19HasOperatorsOutline : !cir.ptr<!rec_HasOperatorsOutline> reduction_operator <lor> init { // CHECK-NEXT: ^bb0(%[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline>{{.*}}) // CHECK-NEXT: %[[ALLOCA:.*]] = cir.alloca !rec_HasOperatorsOutline, !cir.ptr<!rec_HasOperatorsOutline>, ["openacc.reduction.init", init] // CHECK-NEXT: %[[GET_I:.*]] = cir.get_member %[[ALLOCA]][0] {name = "i"} : !cir.ptr<!rec_HasOperatorsOutline> -> !cir.ptr<!s32i> @@ -307,7 +308,8 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZooR19HasOperatorsOutlineS0_(%[[LHSARG]], %[[RHSARG]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHSARG]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_HasOperatorsOutline> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}, %[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}): @@ -317,7 +319,7 @@ void acc_combined() { for(int i=0;i < 5; ++i); #pragma acc parallel loop reduction(+:someVarArr) -// CHECK-NEXT: acc.reduction.recipe @reduction_add__ZTSA5_19HasOperatorsOutline : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> reduction_operator <add> init { +// CHECK: acc.reduction.recipe @reduction_add__ZTSA5_19HasOperatorsOutline : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> reduction_operator <add> init { // CHECK-NEXT: ^bb0(%[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>>{{.*}}) // CHECK-NEXT: %[[ALLOCA:.*]] = cir.alloca !cir.array<!rec_HasOperatorsOutline x 5>, !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>>, ["openacc.reduction.init", init] // CHECK-NEXT: %[[TEMP_ITR:.*]] = cir.alloca !cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!cir.ptr<!rec_HasOperatorsOutline>>, ["arrayinit.temp"] @@ -1253,7 +1255,31 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !s64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !s64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZaaR19HasOperatorsOutlineS0_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}): @@ -1317,7 +1343,31 @@ void acc_combined() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !s64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !s64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZooR19HasOperatorsOutlineS0_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}): @@ -2120,6 +2170,35 @@ void acc_combined() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !u64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !u64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZaaR19HasOperatorsOutlineS0_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}}): @@ -2197,6 +2276,35 @@ void acc_combined() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !u64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !u64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZooR19HasOperatorsOutlineS0_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}}): diff --git a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-default-ops.c b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-default-ops.c index d65d5d4..6ec1c43 100644 --- a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-default-ops.c +++ b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-default-ops.c @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -std=c23 -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -std=c23 -triple x86_64-linux-pc %s -o - | FileCheck %s struct DefaultOperators { int i; @@ -485,7 +485,80 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast integral %[[RES_TO_INT]] : !s32i -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_DefaultOperators> // CHECK-NEXT: } ; @@ -512,7 +585,80 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast integral %[[RES_TO_INT]] : !s32i -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_DefaultOperators> // CHECK-NEXT: } ; @@ -1516,7 +1662,104 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast integral %[[RES_TO_INT]] : !s32i -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } ; @@ -1546,7 +1789,104 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast integral %[[RES_TO_INT]] : !s32i -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } ; @@ -2376,6 +2716,107 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast integral %[[RES_TO_INT]] : !s32i -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } ; @@ -2424,6 +2865,107 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast integral %[[RES_TO_INT]] : !s32i -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } ; diff --git a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-default-ops.cpp b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-default-ops.cpp index f32fa2d..7bd6f67 100644 --- a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-default-ops.cpp +++ b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-default-ops.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s struct DefaultOperators { int i; @@ -480,7 +480,77 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_DefaultOperators> // CHECK-NEXT: } ; @@ -507,7 +577,77 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_DefaultOperators> // CHECK-NEXT: } ; @@ -1532,7 +1672,101 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } ; @@ -1576,7 +1810,101 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } ; @@ -2398,6 +2726,104 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } ; @@ -2446,6 +2872,104 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } ; diff --git a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-float.c b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-float.c index 9f73367..13c335b 100644 --- a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-float.c +++ b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-float.c @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s void acc_compute() { float someVar; @@ -92,7 +92,19 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !cir.float, !cir.ptr<!cir.float> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.float> // CHECK-NEXT: } ; @@ -106,7 +118,19 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !cir.float, !cir.ptr<!cir.float> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.float> // CHECK-NEXT: } ; @@ -371,7 +395,42 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } ; @@ -401,7 +460,42 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } ; @@ -708,7 +802,46 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } ; @@ -744,7 +877,46 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast int_to_float %[[RES_TO_INT]] : !s32i -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } ; diff --git a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-float.cpp b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-float.cpp index ffd2631..6737821 100644 --- a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-float.cpp +++ b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-float.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s template<typename T> void acc_compute() { @@ -93,7 +93,18 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !cir.float, !cir.ptr<!cir.float> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.float> // CHECK-NEXT: } ; @@ -107,7 +118,18 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !cir.float, !cir.ptr<!cir.float> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.float> // CHECK-NEXT: } ; @@ -372,7 +394,41 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } ; @@ -402,7 +458,41 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } ; @@ -709,7 +799,45 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } ; @@ -745,7 +873,45 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } ; diff --git a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-inline-ops.cpp b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-inline-ops.cpp index 1e367ee..262fe98 100644 --- a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-inline-ops.cpp +++ b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-inline-ops.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s struct HasOperatorsInline { int i; @@ -14,8 +14,8 @@ struct HasOperatorsInline { HasOperatorsInline &operator&=(HasOperatorsInline& other); HasOperatorsInline &operator|=(HasOperatorsInline& other); HasOperatorsInline &operator^=(HasOperatorsInline& other); - bool &operator&&(HasOperatorsInline& other); - bool &operator||(HasOperatorsInline& other); + HasOperatorsInline &operator&&(HasOperatorsInline& other); + HasOperatorsInline &operator||(HasOperatorsInline& other); // For min/max bool operator<(HasOperatorsInline& other); HasOperatorsInline &operator=(HasOperatorsInline& other); @@ -277,7 +277,8 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineaaERS_(%[[LHSARG]], %[[RHSARG]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHSARG]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_HasOperatorsInline> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}, %[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}): @@ -286,7 +287,7 @@ void acc_compute() { // CHECK-NEXT: } ; #pragma acc parallel reduction(||:someVar) -// CHECK-NEXT: acc.reduction.recipe @reduction_lor__ZTS18HasOperatorsInline : !cir.ptr<!rec_HasOperatorsInline> reduction_operator <lor> init { +// CHECK: acc.reduction.recipe @reduction_lor__ZTS18HasOperatorsInline : !cir.ptr<!rec_HasOperatorsInline> reduction_operator <lor> init { // CHECK-NEXT: ^bb0(%[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsInline>{{.*}}) // CHECK-NEXT: %[[ALLOCA:.*]] = cir.alloca !rec_HasOperatorsInline, !cir.ptr<!rec_HasOperatorsInline>, ["openacc.reduction.init", init] // CHECK-NEXT: %[[GET_I:.*]] = cir.get_member %[[ALLOCA]][0] {name = "i"} : !cir.ptr<!rec_HasOperatorsInline> -> !cir.ptr<!s32i> @@ -308,7 +309,8 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineooERS_(%[[LHSARG]], %[[RHSARG]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHSARG]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_HasOperatorsInline> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}, %[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}): @@ -318,7 +320,7 @@ void acc_compute() { ; #pragma acc parallel reduction(+:someVarArr) -// CHECK-NEXT: acc.reduction.recipe @reduction_add__ZTSA5_18HasOperatorsInline : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> reduction_operator <add> init { +// CHECK: acc.reduction.recipe @reduction_add__ZTSA5_18HasOperatorsInline : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> reduction_operator <add> init { // CHECK-NEXT: ^bb0(%[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>>{{.*}}) // CHECK-NEXT: %[[ALLOCA:.*]] = cir.alloca !cir.array<!rec_HasOperatorsInline x 5>, !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>>, ["openacc.reduction.init", init] // CHECK-NEXT: %[[TEMP_ITR:.*]] = cir.alloca !cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!cir.ptr<!rec_HasOperatorsInline>>, ["arrayinit.temp"] @@ -1254,7 +1256,31 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !s64i) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !s64i) -> !cir.ptr<!rec_HasOperatorsInline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineaaERS_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}): @@ -1318,7 +1344,31 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !s64i) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !s64i) -> !cir.ptr<!rec_HasOperatorsInline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineooERS_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}): @@ -2121,6 +2171,35 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !u64i) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !u64i) -> !cir.ptr<!rec_HasOperatorsInline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineaaERS_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}}): @@ -2198,6 +2277,35 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !u64i) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !u64i) -> !cir.ptr<!rec_HasOperatorsInline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineooERS_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}}): diff --git a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-int.c b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-int.c index 2f42a5c..be7b123 100644 --- a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-int.c +++ b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-int.c @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s void acc_compute() { int someVar; @@ -144,7 +144,18 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!s32i> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!s32i> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !s32i, !cir.ptr<!s32i> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!s32i> // CHECK-NEXT: } ; @@ -158,7 +169,18 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!s32i> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!s32i> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !s32i, !cir.ptr<!s32i> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!s32i> // CHECK-NEXT: } ; @@ -586,7 +608,41 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } ; @@ -616,7 +672,41 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } ; @@ -1115,7 +1205,45 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } ; @@ -1151,7 +1279,45 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } ; diff --git a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-int.cpp b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-int.cpp index af7bcf3..fb6984f 100644 --- a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-int.cpp +++ b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-int.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s template<typename T> void acc_compute() { @@ -145,7 +145,18 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!s32i> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!s32i> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !s32i, !cir.ptr<!s32i> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!s32i> // CHECK-NEXT: } ; @@ -159,7 +170,18 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!s32i> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!s32i> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !s32i, !cir.ptr<!s32i> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!s32i> // CHECK-NEXT: } ; @@ -587,7 +609,41 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } ; @@ -617,7 +673,41 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } ; @@ -1116,7 +1206,45 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } ; @@ -1152,7 +1280,45 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } ; diff --git a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-outline-ops.cpp b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-outline-ops.cpp index ec890e2..3a80ed5 100644 --- a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-outline-ops.cpp +++ b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-outline-ops.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s struct HasOperatorsOutline { int i; unsigned u; @@ -15,8 +15,8 @@ HasOperatorsOutline &operator*=(HasOperatorsOutline &, HasOperatorsOutline &); HasOperatorsOutline &operator&=(HasOperatorsOutline &, HasOperatorsOutline &); HasOperatorsOutline &operator|=(HasOperatorsOutline &, HasOperatorsOutline &); HasOperatorsOutline &operator^=(HasOperatorsOutline &, HasOperatorsOutline &); -bool &operator&&(HasOperatorsOutline &, HasOperatorsOutline &); -bool &operator||(HasOperatorsOutline &, HasOperatorsOutline &); +HasOperatorsOutline &operator&&(HasOperatorsOutline &, HasOperatorsOutline &); +HasOperatorsOutline &operator||(HasOperatorsOutline &, HasOperatorsOutline &); // For min/max bool operator<(HasOperatorsOutline &, HasOperatorsOutline &); @@ -276,7 +276,8 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZaaR19HasOperatorsOutlineS0_(%[[LHSARG]], %[[RHSARG]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHSARG]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_HasOperatorsOutline> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}, %[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}): @@ -285,7 +286,7 @@ void acc_compute() { // CHECK-NEXT: } ; #pragma acc parallel reduction(||:someVar) -// CHECK-NEXT: acc.reduction.recipe @reduction_lor__ZTS19HasOperatorsOutline : !cir.ptr<!rec_HasOperatorsOutline> reduction_operator <lor> init { +// CHECK: acc.reduction.recipe @reduction_lor__ZTS19HasOperatorsOutline : !cir.ptr<!rec_HasOperatorsOutline> reduction_operator <lor> init { // CHECK-NEXT: ^bb0(%[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline>{{.*}}) // CHECK-NEXT: %[[ALLOCA:.*]] = cir.alloca !rec_HasOperatorsOutline, !cir.ptr<!rec_HasOperatorsOutline>, ["openacc.reduction.init", init] // CHECK-NEXT: %[[GET_I:.*]] = cir.get_member %[[ALLOCA]][0] {name = "i"} : !cir.ptr<!rec_HasOperatorsOutline> -> !cir.ptr<!s32i> @@ -307,7 +308,8 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZooR19HasOperatorsOutlineS0_(%[[LHSARG]], %[[RHSARG]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHSARG]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_HasOperatorsOutline> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}, %[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}): @@ -317,7 +319,7 @@ void acc_compute() { ; #pragma acc parallel reduction(+:someVarArr) -// CHECK-NEXT: acc.reduction.recipe @reduction_add__ZTSA5_19HasOperatorsOutline : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> reduction_operator <add> init { +// CHECK: acc.reduction.recipe @reduction_add__ZTSA5_19HasOperatorsOutline : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> reduction_operator <add> init { // CHECK-NEXT: ^bb0(%[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>>{{.*}}) // CHECK-NEXT: %[[ALLOCA:.*]] = cir.alloca !cir.array<!rec_HasOperatorsOutline x 5>, !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>>, ["openacc.reduction.init", init] // CHECK-NEXT: %[[TEMP_ITR:.*]] = cir.alloca !cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!cir.ptr<!rec_HasOperatorsOutline>>, ["arrayinit.temp"] @@ -1253,7 +1255,31 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !s64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !s64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZaaR19HasOperatorsOutlineS0_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}): @@ -1317,7 +1343,31 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !s64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !s64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZooR19HasOperatorsOutlineS0_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}): @@ -2120,6 +2170,35 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !u64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !u64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZaaR19HasOperatorsOutlineS0_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}}): @@ -2197,6 +2276,35 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !u64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !u64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZooR19HasOperatorsOutlineS0_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}}): diff --git a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-unsigned-int.c b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-unsigned-int.c index 08daa70..9b10a29 100644 --- a/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-unsigned-int.c +++ b/clang/test/CIR/CodeGenOpenACC/compute-reduction-clause-unsigned-int.c @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s void acc_compute() { unsigned int someVar; @@ -143,7 +143,19 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!u32i> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!u32i> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_SINT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast integral %[[RES_TO_SINT]] : !s32i -> !u32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !u32i, !cir.ptr<!u32i> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!u32i> // CHECK-NEXT: } ; @@ -157,7 +169,19 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!u32i> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!u32i> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_SINT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast integral %[[RES_TO_SINT]] : !s32i -> !u32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !u32i, !cir.ptr<!u32i> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!u32i> // CHECK-NEXT: } ; @@ -585,7 +609,42 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!u32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!u32i x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!u32i x 5>> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!u32i>, !s64i) -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!u32i x 5>> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!u32i>, !s64i) -> !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_SINT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast integral %[[RES_TO_SINT]] : !s32i -> !u32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!u32i x 5>> // CHECK-NEXT: } ; @@ -615,7 +674,42 @@ void acc_compute() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!u32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!u32i x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!u32i x 5>> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!u32i>, !s64i) -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!u32i x 5>> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!u32i>, !s64i) -> !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_SINT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast integral %[[RES_TO_SINT]] : !s32i -> !u32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!u32i x 5>> // CHECK-NEXT: } ; @@ -1114,7 +1208,46 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!u32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!u32i x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!u32i x 5>> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!u32i>, !u64i) -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!u32i x 5>> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!u32i>, !u64i) -> !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_SINT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast integral %[[RES_TO_SINT]] : !s32i -> !u32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!u32i x 5>> // CHECK-NEXT: } ; @@ -1150,7 +1283,46 @@ void acc_compute() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!u32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!u32i x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!u32i x 5>> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!u32i>, !u64i) -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!u32i x 5>> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!u32i>, !u64i) -> !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_SINT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast integral %[[RES_TO_SINT]] : !s32i -> !u32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!u32i x 5>> // CHECK-NEXT: } ; diff --git a/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-default-ops.cpp b/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-default-ops.cpp index 1a77c0f..11ebd7b 100644 --- a/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-default-ops.cpp +++ b/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-default-ops.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s struct DefaultOperators { int i; @@ -480,7 +480,77 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_DefaultOperators> // CHECK-NEXT: } for(int i = 0; i < 5; ++i); @@ -507,7 +577,77 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_DefaultOperators> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHSARG]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHSARG]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHSARG]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHSARG]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHSARG]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_DefaultOperators> // CHECK-NEXT: } for(int i = 0; i < 5; ++i); @@ -1532,7 +1672,101 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } for(int i = 0; i < 5; ++i); @@ -1576,7 +1810,101 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !s64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } for(int i = 0; i < 5; ++i); @@ -2398,6 +2726,104 @@ void acc_loop() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } for(int i = 0; i < 5; ++i); @@ -2446,6 +2872,104 @@ void acc_loop() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> -> !cir.ptr<!rec_DefaultOperators> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_DefaultOperators>, !u64i) -> !cir.ptr<!rec_DefaultOperators> +// +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_GET_U:.*]] = cir.get_member %[[LHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[RHS_GET_U:.*]] = cir.get_member %[[RHS_STRIDE]][1] {name = "u"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!u32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_U]] : !cir.ptr<!u32i>, !u32i +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !u32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !u32i +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_U]] : !u32i, !cir.ptr<!u32i> +// +// CHECK-NEXT: %[[LHS_GET_F:.*]] = cir.get_member %[[LHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_GET_F:.*]] = cir.get_member %[[RHS_STRIDE]][2] {name = "f"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_F]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_F]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_GET_D:.*]] = cir.get_member %[[LHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[RHS_GET_D:.*]] = cir.get_member %[[RHS_STRIDE]][3] {name = "d"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.double> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[LHS_CAST:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_D]] : !cir.ptr<!cir.double>, !cir.double +// CHECK-NEXT: %[[RHS_CAST:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.double -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.double +// CHECK-NEXT: cir.store {{.*}} %[[RES_TO_VAL]], %[[LHS_GET_D]] : !cir.double, !cir.ptr<!cir.double> +// +// CHECK-NEXT: %[[LHS_GET_B:.*]] = cir.get_member %[[LHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[RHS_GET_B:.*]] = cir.get_member %[[RHS_STRIDE]][4] {name = "b"} : !cir.ptr<!rec_DefaultOperators> -> !cir.ptr<!cir.bool> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_LOAD]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_B]] : !cir.ptr<!cir.bool>, !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_LOAD]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: cir.store {{.*}} %[[TERNARY]], %[[LHS_GET_B]] : !cir.bool, !cir.ptr<!cir.bool> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_DefaultOperators x 5>> // CHECK-NEXT: } for(int i = 0; i < 5; ++i); diff --git a/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-float.cpp b/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-float.cpp index 7faef71..57cc1af 100644 --- a/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-float.cpp +++ b/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-float.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s template<typename T> void acc_loop() { @@ -93,7 +93,18 @@ void acc_loop() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !cir.float, !cir.ptr<!cir.float> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.float> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -107,7 +118,18 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.float> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !cir.float, !cir.ptr<!cir.float> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.float> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -372,7 +394,41 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -402,7 +458,41 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !s64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -709,7 +799,45 @@ void acc_loop() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -745,7 +873,45 @@ void acc_loop() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!cir.float x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> -> !cir.ptr<!cir.float> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!cir.float>, !u64i) -> !cir.ptr<!cir.float> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[LHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!cir.float>, !cir.float +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast float_to_bool %[[RHS_LOAD]] : !cir.float -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_float %[[TERNARY]] : !cir.bool -> !cir.float +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !cir.float, !cir.ptr<!cir.float> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!cir.float x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); diff --git a/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-inline-ops.cpp b/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-inline-ops.cpp index 43c9fbbc..8a5bf3e 100644 --- a/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-inline-ops.cpp +++ b/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-inline-ops.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s struct HasOperatorsInline { int i; @@ -14,8 +14,8 @@ struct HasOperatorsInline { HasOperatorsInline &operator&=(HasOperatorsInline& other); HasOperatorsInline &operator|=(HasOperatorsInline& other); HasOperatorsInline &operator^=(HasOperatorsInline& other); - bool &operator&&(HasOperatorsInline& other); - bool &operator||(HasOperatorsInline& other); + HasOperatorsInline &operator&&(HasOperatorsInline& other); + HasOperatorsInline &operator||(HasOperatorsInline& other); // For min/max bool operator<(HasOperatorsInline& other); HasOperatorsInline &operator=(HasOperatorsInline& other); @@ -277,7 +277,8 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineaaERS_(%[[LHSARG]], %[[RHSARG]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHSARG]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_HasOperatorsInline> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}, %[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}): @@ -286,7 +287,7 @@ void acc_loop() { // CHECK-NEXT: } for(int i=0;i < 5; ++i); #pragma acc loop reduction(||:someVar) -// CHECK-NEXT: acc.reduction.recipe @reduction_lor__ZTS18HasOperatorsInline : !cir.ptr<!rec_HasOperatorsInline> reduction_operator <lor> init { +// CHECK: acc.reduction.recipe @reduction_lor__ZTS18HasOperatorsInline : !cir.ptr<!rec_HasOperatorsInline> reduction_operator <lor> init { // CHECK-NEXT: ^bb0(%[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsInline>{{.*}}) // CHECK-NEXT: %[[ALLOCA:.*]] = cir.alloca !rec_HasOperatorsInline, !cir.ptr<!rec_HasOperatorsInline>, ["openacc.reduction.init", init] // CHECK-NEXT: %[[GET_I:.*]] = cir.get_member %[[ALLOCA]][0] {name = "i"} : !cir.ptr<!rec_HasOperatorsInline> -> !cir.ptr<!s32i> @@ -308,7 +309,8 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineooERS_(%[[LHSARG]], %[[RHSARG]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHSARG]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_HasOperatorsInline> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}, %[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsInline> {{.*}}): @@ -318,7 +320,7 @@ void acc_loop() { for(int i=0;i < 5; ++i); #pragma acc loop reduction(+:someVarArr) -// CHECK-NEXT: acc.reduction.recipe @reduction_add__ZTSA5_18HasOperatorsInline : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> reduction_operator <add> init { +// CHECK: acc.reduction.recipe @reduction_add__ZTSA5_18HasOperatorsInline : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> reduction_operator <add> init { // CHECK-NEXT: ^bb0(%[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>>{{.*}}) // CHECK-NEXT: %[[ALLOCA:.*]] = cir.alloca !cir.array<!rec_HasOperatorsInline x 5>, !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>>, ["openacc.reduction.init", init] // CHECK-NEXT: %[[TEMP_ITR:.*]] = cir.alloca !cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!cir.ptr<!rec_HasOperatorsInline>>, ["arrayinit.temp"] @@ -1254,7 +1256,31 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !s64i) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !s64i) -> !cir.ptr<!rec_HasOperatorsInline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineaaERS_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}): @@ -1318,7 +1344,31 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !s64i) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !s64i) -> !cir.ptr<!rec_HasOperatorsInline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineooERS_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}): @@ -2121,6 +2171,35 @@ void acc_loop() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !u64i) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !u64i) -> !cir.ptr<!rec_HasOperatorsInline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineaaERS_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}}): @@ -2198,6 +2277,35 @@ void acc_loop() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !u64i) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsInline>, !u64i) -> !cir.ptr<!rec_HasOperatorsInline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZN18HasOperatorsInlineooERS_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline> +// CHECK-NEXT: @_ZN18HasOperatorsInlineaSERS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsInline>, !cir.ptr<!rec_HasOperatorsInline>) -> !cir.ptr<!rec_HasOperatorsInline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsInline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}}): diff --git a/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-int.cpp b/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-int.cpp index 5353218..f60dff9 100644 --- a/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-int.cpp +++ b/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-int.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s template<typename T> void acc_loop() { @@ -145,7 +145,18 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!s32i> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!s32i> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !s32i, !cir.ptr<!s32i> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!s32i> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -159,7 +170,18 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!s32i> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!s32i> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHSARG]] : !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHSARG]] : !s32i, !cir.ptr<!s32i> // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!s32i> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -587,7 +609,41 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -617,7 +673,41 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !s64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -1116,7 +1206,45 @@ void acc_loop() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[FALSE:.*]] = cir.const #false +// CHECK-NEXT: cir.yield %[[FALSE]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); @@ -1152,7 +1280,45 @@ void acc_loop() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!s32i x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!s32i>, !u64i) -> !cir.ptr<!s32i> +// +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load {{.*}} %[[LHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_TO_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load {{.*}} %[[RHS_STRIDE]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_TO_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_TO_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_VAL:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_VAL]], %[[LHS_STRIDE]] : !s32i, !cir.ptr<!s32i> +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!s32i x 5>> // CHECK-NEXT: } for(int i=0;i < 5; ++i); diff --git a/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-outline-ops.cpp b/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-outline-ops.cpp index e193cfa..8613bc8a 100644 --- a/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-outline-ops.cpp +++ b/clang/test/CIR/CodeGenOpenACC/loop-reduction-clause-outline-ops.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s struct HasOperatorsOutline { int i; unsigned u; @@ -15,8 +15,8 @@ HasOperatorsOutline &operator*=(HasOperatorsOutline &, HasOperatorsOutline &); HasOperatorsOutline &operator&=(HasOperatorsOutline &, HasOperatorsOutline &); HasOperatorsOutline &operator|=(HasOperatorsOutline &, HasOperatorsOutline &); HasOperatorsOutline &operator^=(HasOperatorsOutline &, HasOperatorsOutline &); -bool &operator&&(HasOperatorsOutline &, HasOperatorsOutline &); -bool &operator||(HasOperatorsOutline &, HasOperatorsOutline &); +HasOperatorsOutline &operator&&(HasOperatorsOutline &, HasOperatorsOutline &); +HasOperatorsOutline &operator||(HasOperatorsOutline &, HasOperatorsOutline &); // For min/max bool operator<(HasOperatorsOutline &, HasOperatorsOutline &); @@ -276,7 +276,8 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZaaR19HasOperatorsOutlineS0_(%[[LHSARG]], %[[RHSARG]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHSARG]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_HasOperatorsOutline> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}, %[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}): @@ -285,7 +286,7 @@ void acc_loop() { // CHECK-NEXT: } for(int i=0;i < 5; ++i); #pragma acc loop reduction(||:someVar) -// CHECK-NEXT: acc.reduction.recipe @reduction_lor__ZTS19HasOperatorsOutline : !cir.ptr<!rec_HasOperatorsOutline> reduction_operator <lor> init { +// CHECK: acc.reduction.recipe @reduction_lor__ZTS19HasOperatorsOutline : !cir.ptr<!rec_HasOperatorsOutline> reduction_operator <lor> init { // CHECK-NEXT: ^bb0(%[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline>{{.*}}) // CHECK-NEXT: %[[ALLOCA:.*]] = cir.alloca !rec_HasOperatorsOutline, !cir.ptr<!rec_HasOperatorsOutline>, ["openacc.reduction.init", init] // CHECK-NEXT: %[[GET_I:.*]] = cir.get_member %[[ALLOCA]][0] {name = "i"} : !cir.ptr<!rec_HasOperatorsOutline> -> !cir.ptr<!s32i> @@ -307,7 +308,8 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZooR19HasOperatorsOutlineS0_(%[[LHSARG]], %[[RHSARG]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHSARG]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!rec_HasOperatorsOutline> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}, %[[ARG:.*]]: !cir.ptr<!rec_HasOperatorsOutline> {{.*}}): @@ -317,7 +319,7 @@ void acc_loop() { for(int i=0;i < 5; ++i); #pragma acc loop reduction(+:someVarArr) -// CHECK-NEXT: acc.reduction.recipe @reduction_add__ZTSA5_19HasOperatorsOutline : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> reduction_operator <add> init { +// CHECK: acc.reduction.recipe @reduction_add__ZTSA5_19HasOperatorsOutline : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> reduction_operator <add> init { // CHECK-NEXT: ^bb0(%[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>>{{.*}}) // CHECK-NEXT: %[[ALLOCA:.*]] = cir.alloca !cir.array<!rec_HasOperatorsOutline x 5>, !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>>, ["openacc.reduction.init", init] // CHECK-NEXT: %[[TEMP_ITR:.*]] = cir.alloca !cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!cir.ptr<!rec_HasOperatorsOutline>>, ["arrayinit.temp"] @@ -1253,7 +1255,31 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !s64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !s64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZaaR19HasOperatorsOutlineS0_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}): @@ -1317,7 +1343,31 @@ void acc_loop() { // // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}) -// TODO OpenACC: Expecting combination operation here +// CHECK-NEXT: %[[ZERO:.*]] = cir.const #cir.int<0> : !s64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !s64i, !cir.ptr<!s64i>, ["itr"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[ZERO]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[END_VAL:.*]] = cir.const #cir.int<5> : !s64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[END_VAL]]) : !s64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !s64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !s64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZooR19HasOperatorsOutlineS0_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!s64i>, !s64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !s64i, !s64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !s64i, !cir.ptr<!s64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}): @@ -2120,6 +2170,35 @@ void acc_loop() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !u64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !u64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZaaR19HasOperatorsOutlineS0_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}}): @@ -2197,6 +2276,35 @@ void acc_loop() { // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { // CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}})) +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB]] : index to !u64i +// CHECK-NEXT: %[[UB:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB]] : index to !u64i +// CHECK-NEXT: %[[ITR:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB_CAST]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR_LOAD]], %[[UB_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR_LOAD:.*]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !u64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_DECAY]], %[[ITR_LOAD]] : (!cir.ptr<!rec_HasOperatorsOutline>, !u64i) -> !cir.ptr<!rec_HasOperatorsOutline> +// +// CHECK-NEXT: %[[OP_RES:.*]] = cir.call @_ZooR19HasOperatorsOutlineS0_(%[[LHS_STRIDE]], %[[RHS_STRIDE]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline> +// CHECK-NEXT: @_ZN19HasOperatorsOutlineaSERKS_(%[[LHS_STRIDE]], %[[OP_RES]]) : (!cir.ptr<!rec_HasOperatorsOutline>, !cir.ptr<!rec_HasOperatorsOutline>) -> !cir.ptr<!rec_HasOperatorsOutline +// +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR_LOAD]] = cir.load %[[ITR]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } // CHECK-NEXT: acc.yield %[[LHSARG]] : !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[ORIG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[ARG:.*]]: !cir.ptr<!cir.array<!rec_HasOperatorsOutline x 5>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty{{.*}}): diff --git a/clang/test/CIR/CodeGenOpenACC/reduction-clause-recipes.cpp b/clang/test/CIR/CodeGenOpenACC/reduction-clause-recipes.cpp index 20ad7a3..29f1b5f 100644 --- a/clang/test/CIR/CodeGenOpenACC/reduction-clause-recipes.cpp +++ b/clang/test/CIR/CodeGenOpenACC/reduction-clause-recipes.cpp @@ -1,4 +1,4 @@ -// RUN: not %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s +// RUN: %clang_cc1 -fopenacc -triple x86_64-linux-gnu -Wno-openacc-self-if-potential-conflict -emit-cir -fclangir -triple x86_64-linux-pc %s -o - | FileCheck %s // Note: unlike the 'private' recipe checks, this is just for spot-checking, // so this test isn't as comprehensive. The same code paths are used for @@ -753,8 +753,97 @@ void do_things(unsigned A, unsigned B) { // CHECK-NEXT: } // CHECK-NEXT: acc.yield // CHECK-NEXT: } combiner { -// CHECK-NEXT: ^bb0(%[[REF:.*]]: !cir.ptr<!cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>>> {{.*}}, %[[PRIVATE:.*]]: !cir.ptr<!cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty {{.*}}, %[[BOUND2:.*]]: !acc.data_bounds_ty {{.*}}, %[[BOUND3:.*]]: !acc.data_bounds_ty {{.*}}): -// CHECK-NEXT: acc.yield +// CHECK-NEXT: ^bb0(%[[LHSARG:.*]]: !cir.ptr<!cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>>> {{.*}}, %[[RHSARG:.*]]: !cir.ptr<!cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty {{.*}}, %[[BOUND2:.*]]: !acc.data_bounds_ty {{.*}}, %[[BOUND3:.*]]: !acc.data_bounds_ty {{.*}}): +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB3:.*]] = acc.get_lowerbound %[[BOUND3]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB3_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB3]] : index to !u64i +// CHECK-NEXT: %[[UB3:.*]] = acc.get_upperbound %[[BOUND3]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB3_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB3]] : index to !u64i +// CHECK-NEXT: %[[ITR3:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB3_CAST]], %[[ITR3]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR3_LOAD:.*]] = cir.load %[[ITR3]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[CMP:.*]] = cir.cmp(lt, %[[ITR3_LOAD]], %[[UB3_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[CMP]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR3_LOAD:.*]] = cir.load %[[ITR3]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_TLA_LOAD:.*]] = cir.load %[[LHSARG]] : !cir.ptr<!cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>>>, !cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>> +// CHECK-NEXT: %[[LHS_BOUND3_STRIDE:.*]] = cir.ptr_stride %[[LHS_TLA_LOAD]], %[[ITR3_LOAD]] : (!cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>>, !u64i) -> !cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>> +// CHECK-NEXT: %[[RHS_TLA_LOAD:.*]] = cir.load %[[RHSARG]] : !cir.ptr<!cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>>>, !cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>> +// CHECK-NEXT: %[[RHS_BOUND3_STRIDE:.*]] = cir.ptr_stride %[[RHS_TLA_LOAD]], %[[ITR3_LOAD]] : (!cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>>, !u64i) -> !cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>> +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB2:.*]] = acc.get_lowerbound %[[BOUND2]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB2_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB2]] : index to !u64i +// CHECK-NEXT: %[[UB2:.*]] = acc.get_upperbound %[[BOUND2]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB2_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB2]] : index to !u64i +// CHECK-NEXT: %[[ITR2:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB2_CAST]], %[[ITR2]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR2_LOAD:.*]] = cir.load %[[ITR2]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR2_LOAD]], %[[UB2_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR2_LOAD:.*]] = cir.load %[[ITR2]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_BOUND3_STRIDE_DECAY:.*]] = cir.cast array_to_ptrdecay %[[LHS_BOUND3_STRIDE]] : !cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>> -> !cir.ptr<!cir.ptr<!rec_NoOps>> +// CHECK-NEXT: %[[LHS_BOUND2_STRIDE:.*]] = cir.ptr_stride %[[LHS_BOUND3_STRIDE_DECAY]], %[[ITR2_LOAD]] : (!cir.ptr<!cir.ptr<!rec_NoOps>>, !u64i) -> !cir.ptr<!cir.ptr<!rec_NoOps>> +// CHECK-NEXT: %[[RHS_BOUND3_STRIDE_DECAY:.*]] = cir.cast array_to_ptrdecay %[[RHS_BOUND3_STRIDE]] : !cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>> -> !cir.ptr<!cir.ptr<!rec_NoOps>> +// CHECK-NEXT: %[[RHS_BOUND2_STRIDE:.*]] = cir.ptr_stride %[[RHS_BOUND3_STRIDE_DECAY]], %[[ITR2_LOAD]] : (!cir.ptr<!cir.ptr<!rec_NoOps>>, !u64i) -> !cir.ptr<!cir.ptr<!rec_NoOps>> +// CHECK-NEXT: cir.scope { +// CHECK-NEXT: %[[LB1:.*]] = acc.get_lowerbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[LB1_CAST:.*]] = builtin.unrealized_conversion_cast %[[LB1]] : index to !u64i +// CHECK-NEXT: %[[UB1:.*]] = acc.get_upperbound %[[BOUND1]] : (!acc.data_bounds_ty) -> index +// CHECK-NEXT: %[[UB1_CAST:.*]] = builtin.unrealized_conversion_cast %[[UB1]] : index to !u64i +// CHECK-NEXT: %[[ITR1:.*]] = cir.alloca !u64i, !cir.ptr<!u64i>, ["iter"] {alignment = 8 : i64} +// CHECK-NEXT: cir.store %[[LB1_CAST]], %[[ITR1]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.for : cond { +// CHECK-NEXT: %[[ITR1_LOAD:.*]] = cir.load %[[ITR1]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[COND:.*]] = cir.cmp(lt, %[[ITR1_LOAD]], %[[UB1_CAST]]) : !u64i, !cir.bool +// CHECK-NEXT: cir.condition(%[[COND]]) +// CHECK-NEXT: } body { +// CHECK-NEXT: %[[ITR1_LOAD:.*]] = cir.load %[[ITR1]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[LHS_BOUND2_STRIDE_LOAD:.*]] = cir.load %[[LHS_BOUND2_STRIDE]] : !cir.ptr<!cir.ptr<!rec_NoOps>>, !cir.ptr<!rec_NoOps> +// CHECK-NEXT: %[[LHS_STRIDE:.*]] = cir.ptr_stride %[[LHS_BOUND2_STRIDE_LOAD]], %[[ITR1_LOAD]] : (!cir.ptr<!rec_NoOps>, !u64i) -> !cir.ptr<!rec_NoOps> +// CHECK-NEXT: %[[RHS_BOUND2_STRIDE_LOAD:.*]] = cir.load %[[RHS_BOUND2_STRIDE]] : !cir.ptr<!cir.ptr<!rec_NoOps>>, !cir.ptr<!rec_NoOps> +// CHECK-NEXT: %[[RHS_STRIDE:.*]] = cir.ptr_stride %[[RHS_BOUND2_STRIDE_LOAD]], %[[ITR1_LOAD]] : (!cir.ptr<!rec_NoOps>, !u64i) -> !cir.ptr<!rec_NoOps> +// CHECK-NEXT: %[[LHS_GET_I:.*]] = cir.get_member %[[LHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_NoOps> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[RHS_GET_I:.*]] = cir.get_member %[[RHS_STRIDE]][0] {name = "i"} : !cir.ptr<!rec_NoOps> -> !cir.ptr<!s32i> +// CHECK-NEXT: %[[LHS_LOAD:.*]] = cir.load{{.*}} %[[LHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[LHS_CAST_BOOL:.*]] = cir.cast int_to_bool %[[LHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: %[[TERNARY:.*]] = cir.ternary(%[[LHS_CAST_BOOL]], true { +// CHECK-NEXT: %[[TRUE:.*]] = cir.const #true +// CHECK-NEXT: cir.yield %[[TRUE]] : !cir.bool +// CHECK-NEXT: }, false { +// CHECK-NEXT: %[[RHS_LOAD:.*]] = cir.load{{.*}} %[[RHS_GET_I]] : !cir.ptr<!s32i>, !s32i +// CHECK-NEXT: %[[RHS_CAST_BOOL:.*]] = cir.cast int_to_bool %[[RHS_LOAD]] : !s32i -> !cir.bool +// CHECK-NEXT: cir.yield %[[RHS_CAST_BOOL]] : !cir.bool +// CHECK-NEXT: }) : (!cir.bool) -> !cir.bool +// CHECK-NEXT: %[[RES_TO_INT:.*]] = cir.cast bool_to_int %[[TERNARY]] : !cir.bool -> !s32i +// CHECK-NEXT: cir.store{{.*}} %[[RES_TO_INT]], %[[LHS_GET_I]] : !s32i, !cir.ptr<!s32i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR1_LOAD]] = cir.load %[[ITR1]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR1_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR1]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR2_LOAD]] = cir.load %[[ITR2]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR2_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR2]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } step { +// CHECK-NEXT: %[[ITR3_LOAD]] = cir.load %[[ITR3]] : !cir.ptr<!u64i>, !u64i +// CHECK-NEXT: %[[INC:.*]] = cir.unary(inc, %[[ITR3_LOAD]]) : !u64i, !u64i +// CHECK-NEXT: cir.store %[[INC]], %[[ITR3]] : !u64i, !cir.ptr<!u64i> +// CHECK-NEXT: cir.yield +// CHECK-NEXT: } +// CHECK-NEXT: } +// CHECK-NEXT: acc.yield %[[LHSARG]] // CHECK-NEXT: } destroy { // CHECK-NEXT: ^bb0(%[[REF:.*]]: !cir.ptr<!cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>>> {{.*}}, %[[PRIVATE:.*]]: !cir.ptr<!cir.ptr<!cir.array<!cir.ptr<!rec_NoOps> x 5>>> {{.*}}, %[[BOUND1:.*]]: !acc.data_bounds_ty {{.*}}, %[[BOUND2:.*]]: !acc.data_bounds_ty {{.*}}, %[[BOUND3:.*]]: !acc.data_bounds_ty {{.*}}): // CHECK-NEXT: cir.scope { diff --git a/clang/test/CIR/IR/invalid-atomic.cir b/clang/test/CIR/IR/invalid-atomic.cir new file mode 100644 index 0000000..a124e43 --- /dev/null +++ b/clang/test/CIR/IR/invalid-atomic.cir @@ -0,0 +1,7 @@ +// RUN: cir-opt %s -verify-diagnostics -split-input-file + +cir.func @f1(%arg0: !cir.ptr<!cir.float>, %arg1: !cir.float) { + // expected-error @below {{only atomic add, sub, max, and min operation could operate on floating-point values}} + %0 = cir.atomic.fetch and seq_cst %arg0, %arg1 : (!cir.ptr<!cir.float>, !cir.float) -> !cir.float + cir.return +} diff --git a/clang/test/CodeGen/AArch64/ABI-align-packed.c b/clang/test/CodeGen/AArch64/ABI-align-packed.c index 09f9180..d88a338 100644 --- a/clang/test/CodeGen/AArch64/ABI-align-packed.c +++ b/clang/test/CodeGen/AArch64/ABI-align-packed.c @@ -1,3 +1,4 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 // REQUIRES: aarch64-registered-target // RUN: %clang_cc1 -triple aarch64 -target-feature +neon -emit-llvm -O2 -o - %s | FileCheck %s #include <stdarg.h> @@ -58,8 +59,8 @@ struct non_packed_struct gs_non_packed_struct; // CHECK-SAME: (double [[D0:%.*]], double [[D1:%.*]], double [[D2:%.*]], double [[D3:%.*]], double [[D4:%.*]], double [[D5:%.*]], double [[D6:%.*]], double [[D7:%.*]], double noundef [[D8:%.*]], [1 x <8 x i16>] alignstack(16) [[S_NON_PACKED_STRUCT_COERCE:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: entry: // CHECK-NEXT: [[S_NON_PACKED_STRUCT_COERCE_FCA_0_EXTRACT:%.*]] = extractvalue [1 x <8 x i16>] [[S_NON_PACKED_STRUCT_COERCE]], 0 -// CHECK-NEXT: store double [[D8]], ptr @gd, align 8, !tbaa [[TBAA2:![0-9]+]] -// CHECK-NEXT: store <8 x i16> [[S_NON_PACKED_STRUCT_COERCE_FCA_0_EXTRACT]], ptr @gs_non_packed_struct, align 16, !tbaa [[TBAA6:![0-9]+]] +// CHECK-NEXT: store double [[D8]], ptr @gd, align 8, !tbaa [[TBAA6:![0-9]+]] +// CHECK-NEXT: store <8 x i16> [[S_NON_PACKED_STRUCT_COERCE_FCA_0_EXTRACT]], ptr @gs_non_packed_struct, align 16, !tbaa [[TBAA8:![0-9]+]] // CHECK-NEXT: ret void __attribute__((noinline)) void named_arg_non_packed_struct(double d0, double d1, double d2, double d3, double d4, double d5, double d6, double d7, @@ -113,8 +114,8 @@ struct packed_struct gs_packed_struct; // CHECK-SAME: (double [[D0:%.*]], double [[D1:%.*]], double [[D2:%.*]], double [[D3:%.*]], double [[D4:%.*]], double [[D5:%.*]], double [[D6:%.*]], double [[D7:%.*]], double noundef [[D8:%.*]], [1 x <8 x i16>] alignstack(8) [[S_PACKED_STRUCT_COERCE:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: entry: // CHECK-NEXT: [[S_PACKED_STRUCT_COERCE_FCA_0_EXTRACT:%.*]] = extractvalue [1 x <8 x i16>] [[S_PACKED_STRUCT_COERCE]], 0 -// CHECK-NEXT: store double [[D8]], ptr @gd, align 8, !tbaa [[TBAA2]] -// CHECK-NEXT: store <8 x i16> [[S_PACKED_STRUCT_COERCE_FCA_0_EXTRACT]], ptr @gs_packed_struct, align 1, !tbaa [[TBAA6]] +// CHECK-NEXT: store double [[D8]], ptr @gd, align 8, !tbaa [[TBAA6]] +// CHECK-NEXT: store <8 x i16> [[S_PACKED_STRUCT_COERCE_FCA_0_EXTRACT]], ptr @gs_packed_struct, align 1, !tbaa [[TBAA8]] // CHECK-NEXT: ret void __attribute__((noinline)) void named_arg_packed_struct(double d0, double d1, double d2, double d3, double d4, double d5, double d6, double d7, @@ -168,8 +169,8 @@ struct packed_member gs_packed_member; // CHECK-SAME: (double [[D0:%.*]], double [[D1:%.*]], double [[D2:%.*]], double [[D3:%.*]], double [[D4:%.*]], double [[D5:%.*]], double [[D6:%.*]], double [[D7:%.*]], double noundef [[D8:%.*]], [1 x <8 x i16>] alignstack(8) [[S_PACKED_MEMBER_COERCE:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: entry: // CHECK-NEXT: [[S_PACKED_MEMBER_COERCE_FCA_0_EXTRACT:%.*]] = extractvalue [1 x <8 x i16>] [[S_PACKED_MEMBER_COERCE]], 0 -// CHECK-NEXT: store double [[D8]], ptr @gd, align 8, !tbaa [[TBAA2]] -// CHECK-NEXT: store <8 x i16> [[S_PACKED_MEMBER_COERCE_FCA_0_EXTRACT]], ptr @gs_packed_member, align 1, !tbaa [[TBAA6]] +// CHECK-NEXT: store double [[D8]], ptr @gd, align 8, !tbaa [[TBAA6]] +// CHECK-NEXT: store <8 x i16> [[S_PACKED_MEMBER_COERCE_FCA_0_EXTRACT]], ptr @gs_packed_member, align 1, !tbaa [[TBAA8]] // CHECK-NEXT: ret void __attribute__((noinline)) void named_arg_packed_member(double d0, double d1, double d2, double d3, double d4, double d5, double d6, double d7, @@ -223,8 +224,8 @@ struct aligned_struct_8 gs_aligned_struct_8; // CHECK-SAME: (double [[D0:%.*]], double [[D1:%.*]], double [[D2:%.*]], double [[D3:%.*]], double [[D4:%.*]], double [[D5:%.*]], double [[D6:%.*]], double [[D7:%.*]], double noundef [[D8:%.*]], [1 x <8 x i16>] alignstack(16) [[S_ALIGNED_STRUCT_8_COERCE:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: entry: // CHECK-NEXT: [[S_ALIGNED_STRUCT_8_COERCE_FCA_0_EXTRACT:%.*]] = extractvalue [1 x <8 x i16>] [[S_ALIGNED_STRUCT_8_COERCE]], 0 -// CHECK-NEXT: store double [[D8]], ptr @gd, align 8, !tbaa [[TBAA2]] -// CHECK-NEXT: store <8 x i16> [[S_ALIGNED_STRUCT_8_COERCE_FCA_0_EXTRACT]], ptr @gs_aligned_struct_8, align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: store double [[D8]], ptr @gd, align 8, !tbaa [[TBAA6]] +// CHECK-NEXT: store <8 x i16> [[S_ALIGNED_STRUCT_8_COERCE_FCA_0_EXTRACT]], ptr @gs_aligned_struct_8, align 16, !tbaa [[TBAA8]] // CHECK-NEXT: ret void __attribute__((noinline)) void named_arg_aligned_struct_8(double d0, double d1, double d2, double d3, double d4, double d5, double d6, double d7, @@ -278,8 +279,8 @@ struct aligned_member_8 gs_aligned_member_8; // CHECK-SAME: (double [[D0:%.*]], double [[D1:%.*]], double [[D2:%.*]], double [[D3:%.*]], double [[D4:%.*]], double [[D5:%.*]], double [[D6:%.*]], double [[D7:%.*]], double noundef [[D8:%.*]], [1 x <8 x i16>] alignstack(16) [[S_ALIGNED_MEMBER_8_COERCE:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: entry: // CHECK-NEXT: [[S_ALIGNED_MEMBER_8_COERCE_FCA_0_EXTRACT:%.*]] = extractvalue [1 x <8 x i16>] [[S_ALIGNED_MEMBER_8_COERCE]], 0 -// CHECK-NEXT: store double [[D8]], ptr @gd, align 8, !tbaa [[TBAA2]] -// CHECK-NEXT: store <8 x i16> [[S_ALIGNED_MEMBER_8_COERCE_FCA_0_EXTRACT]], ptr @gs_aligned_member_8, align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: store double [[D8]], ptr @gd, align 8, !tbaa [[TBAA6]] +// CHECK-NEXT: store <8 x i16> [[S_ALIGNED_MEMBER_8_COERCE_FCA_0_EXTRACT]], ptr @gs_aligned_member_8, align 16, !tbaa [[TBAA8]] // CHECK-NEXT: ret void __attribute__((noinline)) void named_arg_aligned_member_8(double d0, double d1, double d2, double d3, double d4, double d5, double d6, double d7, @@ -333,8 +334,8 @@ struct pragma_packed_struct_8 gs_pragma_packed_struct_8; // CHECK-SAME: (double [[D0:%.*]], double [[D1:%.*]], double [[D2:%.*]], double [[D3:%.*]], double [[D4:%.*]], double [[D5:%.*]], double [[D6:%.*]], double [[D7:%.*]], double noundef [[D8:%.*]], [1 x <8 x i16>] alignstack(8) [[S_PRAGMA_PACKED_STRUCT_8_COERCE:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: entry: // CHECK-NEXT: [[S_PRAGMA_PACKED_STRUCT_8_COERCE_FCA_0_EXTRACT:%.*]] = extractvalue [1 x <8 x i16>] [[S_PRAGMA_PACKED_STRUCT_8_COERCE]], 0 -// CHECK-NEXT: store double [[D8]], ptr @gd, align 8, !tbaa [[TBAA2]] -// CHECK-NEXT: store <8 x i16> [[S_PRAGMA_PACKED_STRUCT_8_COERCE_FCA_0_EXTRACT]], ptr @gs_pragma_packed_struct_8, align 8, !tbaa [[TBAA6]] +// CHECK-NEXT: store double [[D8]], ptr @gd, align 8, !tbaa [[TBAA6]] +// CHECK-NEXT: store <8 x i16> [[S_PRAGMA_PACKED_STRUCT_8_COERCE_FCA_0_EXTRACT]], ptr @gs_pragma_packed_struct_8, align 8, !tbaa [[TBAA8]] // CHECK-NEXT: ret void __attribute__((noinline)) void named_arg_pragma_packed_struct_8(double d0, double d1, double d2, double d3, double d4, double d5, double d6, double d7, @@ -388,8 +389,8 @@ struct pragma_packed_struct_4 gs_pragma_packed_struct_4; // CHECK-SAME: (double [[D0:%.*]], double [[D1:%.*]], double [[D2:%.*]], double [[D3:%.*]], double [[D4:%.*]], double [[D5:%.*]], double [[D6:%.*]], double [[D7:%.*]], double noundef [[D8:%.*]], [1 x <8 x i16>] alignstack(8) [[S_PRAGMA_PACKED_STRUCT_4_COERCE:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: entry: // CHECK-NEXT: [[S_PRAGMA_PACKED_STRUCT_4_COERCE_FCA_0_EXTRACT:%.*]] = extractvalue [1 x <8 x i16>] [[S_PRAGMA_PACKED_STRUCT_4_COERCE]], 0 -// CHECK-NEXT: store double [[D8]], ptr @gd, align 8, !tbaa [[TBAA2]] -// CHECK-NEXT: store <8 x i16> [[S_PRAGMA_PACKED_STRUCT_4_COERCE_FCA_0_EXTRACT]], ptr @gs_pragma_packed_struct_4, align 4, !tbaa [[TBAA6]] +// CHECK-NEXT: store double [[D8]], ptr @gd, align 8, !tbaa [[TBAA6]] +// CHECK-NEXT: store <8 x i16> [[S_PRAGMA_PACKED_STRUCT_4_COERCE_FCA_0_EXTRACT]], ptr @gs_pragma_packed_struct_4, align 4, !tbaa [[TBAA8]] // CHECK-NEXT: ret void __attribute__((noinline)) void named_arg_pragma_packed_struct_4(double d0, double d1, double d2, double d3, double d4, double d5, double d6, double d7, @@ -437,9 +438,9 @@ void test_pragma_packed_struct_4() { variadic_pragma_packed_struct_4(1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 2.0, s_pragma_packed_struct_4); } //. -// CHECK: [[TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"double", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} -// CHECK: [[TBAA6]] = !{[[META4]], [[META4]], i64 0} +// CHECK: [[TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK: [[META7]] = !{!"double", [[META4]], i64 0} +// CHECK: [[TBAA8]] = !{[[META4]], [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/AArch64/fp8-init-list.c b/clang/test/CodeGen/AArch64/fp8-init-list.c index 7c0f6278..f461977 100644 --- a/clang/test/CodeGen/AArch64/fp8-init-list.c +++ b/clang/test/CodeGen/AArch64/fp8-init-list.c @@ -34,26 +34,26 @@ struct S s; // CHECK-LABEL: define dso_local void @f( // CHECK-SAME: <1 x i8> [[X:%.*]]) local_unnamed_addr #[[ATTR1:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: store <1 x i8> [[X]], ptr @s, align 1, !tbaa [[__MFP8_TBAA2:![0-9]+]] +// CHECK-NEXT: store <1 x i8> [[X]], ptr @s, align 1, !tbaa [[__MFP8_TBAA6:![0-9]+]] // CHECK-NEXT: ret void // // CHECK-CXX-LABEL: define dso_local void @_Z1fu6__mfp8( // CHECK-CXX-SAME: <1 x i8> [[X:%.*]]) local_unnamed_addr #[[ATTR1:[0-9]+]] { // CHECK-CXX-NEXT: [[ENTRY:.*:]] -// CHECK-CXX-NEXT: store <1 x i8> [[X]], ptr @s, align 1, !tbaa [[__MFP8_TBAA2:![0-9]+]] +// CHECK-CXX-NEXT: store <1 x i8> [[X]], ptr @s, align 1, !tbaa [[__MFP8_TBAA6:![0-9]+]] // CHECK-CXX-NEXT: ret void // void f(__mfp8 x) { s = (struct S){x}; } //. -// CHECK: [[__MFP8_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"__mfp8", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[__MFP8_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK: [[META7]] = !{!"__mfp8", [[META4]], i64 0} //. -// CHECK-CXX: [[__MFP8_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK-CXX: [[META3]] = !{!"__mfp8", [[META4:![0-9]+]], i64 0} -// CHECK-CXX: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK-CXX: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK-CXX: [[META5]] = !{!"Simple C++ TBAA"} +// CHECK-CXX: [[__MFP8_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK-CXX: [[META7]] = !{!"__mfp8", [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/AArch64/ls64-inline-asm.c b/clang/test/CodeGen/AArch64/ls64-inline-asm.c index 1d217eb..04e2207 100644 --- a/clang/test/CodeGen/AArch64/ls64-inline-asm.c +++ b/clang/test/CodeGen/AArch64/ls64-inline-asm.c @@ -6,7 +6,7 @@ struct foo { unsigned long long x[8]; }; // CHECK-LABEL: define dso_local void @load( // CHECK-SAME: ptr noundef writeonly captures(none) initializes((0, 64)) [[OUTPUT:%.*]], ptr noundef [[ADDR:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = tail call i512 asm sideeffect "ld64b $0,[$1]", "=r,r,~{memory}"(ptr [[ADDR]]) #[[ATTR1:[0-9]+]], !srcloc [[META2:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = tail call i512 asm sideeffect "ld64b $0,[$1]", "=r,r,~{memory}"(ptr [[ADDR]]) #[[ATTR1:[0-9]+]], !srcloc [[META6:![0-9]+]] // CHECK-NEXT: store i512 [[TMP0]], ptr [[OUTPUT]], align 8 // CHECK-NEXT: ret void // @@ -19,7 +19,7 @@ void load(struct foo *output, void *addr) // CHECK-SAME: ptr noundef readonly captures(none) [[INPUT:%.*]], ptr noundef [[ADDR:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = load i512, ptr [[INPUT]], align 8 -// CHECK-NEXT: tail call void asm sideeffect "st64b $0,[$1]", "r,r,~{memory}"(i512 [[TMP0]], ptr [[ADDR]]) #[[ATTR1]], !srcloc [[META3:![0-9]+]] +// CHECK-NEXT: tail call void asm sideeffect "st64b $0,[$1]", "r,r,~{memory}"(i512 [[TMP0]], ptr [[ADDR]]) #[[ATTR1]], !srcloc [[META7:![0-9]+]] // CHECK-NEXT: ret void // void store(const struct foo *input, void *addr) @@ -30,28 +30,28 @@ void store(const struct foo *input, void *addr) // CHECK-LABEL: define dso_local void @store2( // CHECK-SAME: ptr noundef readonly captures(none) [[IN:%.*]], ptr noundef [[ADDR:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[IN]], align 4, !tbaa [[INT_TBAA4:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[IN]], align 4, !tbaa [[INT_TBAA2:![0-9]+]] // CHECK-NEXT: [[CONV:%.*]] = sext i32 [[TMP0]] to i64 // CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i8, ptr [[IN]], i64 4 -// CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4, !tbaa [[INT_TBAA4]] +// CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4, !tbaa [[INT_TBAA2]] // CHECK-NEXT: [[CONV2:%.*]] = sext i32 [[TMP1]] to i64 // CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw i8, ptr [[IN]], i64 16 -// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4, !tbaa [[INT_TBAA4]] +// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4, !tbaa [[INT_TBAA2]] // CHECK-NEXT: [[CONV5:%.*]] = sext i32 [[TMP2]] to i64 // CHECK-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds nuw i8, ptr [[IN]], i64 64 -// CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARRAYIDX7]], align 4, !tbaa [[INT_TBAA4]] +// CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARRAYIDX7]], align 4, !tbaa [[INT_TBAA2]] // CHECK-NEXT: [[CONV8:%.*]] = sext i32 [[TMP3]] to i64 // CHECK-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds nuw i8, ptr [[IN]], i64 100 -// CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX10]], align 4, !tbaa [[INT_TBAA4]] +// CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX10]], align 4, !tbaa [[INT_TBAA2]] // CHECK-NEXT: [[CONV11:%.*]] = sext i32 [[TMP4]] to i64 // CHECK-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds nuw i8, ptr [[IN]], i64 144 -// CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX13]], align 4, !tbaa [[INT_TBAA4]] +// CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX13]], align 4, !tbaa [[INT_TBAA2]] // CHECK-NEXT: [[CONV14:%.*]] = sext i32 [[TMP5]] to i64 // CHECK-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds nuw i8, ptr [[IN]], i64 196 -// CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX16]], align 4, !tbaa [[INT_TBAA4]] +// CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX16]], align 4, !tbaa [[INT_TBAA2]] // CHECK-NEXT: [[CONV17:%.*]] = sext i32 [[TMP6]] to i64 // CHECK-NEXT: [[ARRAYIDX19:%.*]] = getelementptr inbounds nuw i8, ptr [[IN]], i64 256 -// CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX19]], align 4, !tbaa [[INT_TBAA4]] +// CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX19]], align 4, !tbaa [[INT_TBAA2]] // CHECK-NEXT: [[CONV20:%.*]] = sext i32 [[TMP7]] to i64 // CHECK-NEXT: [[S_SROA_10_0_INSERT_EXT:%.*]] = zext i64 [[CONV20]] to i512 // CHECK-NEXT: [[S_SROA_10_0_INSERT_SHIFT:%.*]] = shl nuw i512 [[S_SROA_10_0_INSERT_EXT]], 448 @@ -84,11 +84,11 @@ void store2(int *in, void *addr) __asm__ volatile ("st64b %0,[%1]" : : "r" (s), "r" (addr) : "memory" ); } //. -// CHECK: [[META2]] = !{i64 789} -// CHECK: [[META3]] = !{i64 1368} -// CHECK: [[INT_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} -// CHECK: [[META5]] = !{!"int", [[META6:![0-9]+]], i64 0} -// CHECK: [[META6]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} -// CHECK: [[META7]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[INT_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// CHECK: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META6]] = !{i64 789} +// CHECK: [[META7]] = !{i64 1368} // CHECK: [[META8]] = !{i64 5992} //. diff --git a/clang/test/CodeGen/AArch64/sign-return-address.c b/clang/test/CodeGen/AArch64/sign-return-address.c index 11dd683..2b505de 100644 --- a/clang/test/CodeGen/AArch64/sign-return-address.c +++ b/clang/test/CodeGen/AArch64/sign-return-address.c @@ -28,17 +28,17 @@ // NONE-NOT: !"branch-target-enforcement" // ALL-NOT: !"branch-target-enforcement" // PART-NOT: !"branch-target-enforcement" -// BTE: !{i32 8, !"branch-target-enforcement", i32 1} +// BTE: !{i32 8, !"branch-target-enforcement", i32 2} // B-KEY-NOT: !"branch-target-enforcement" // NONE-NOT: !"sign-return-address" -// ALL: !{i32 8, !"sign-return-address", i32 1} -// PART: !{i32 8, !"sign-return-address", i32 1} +// ALL: !{i32 8, !"sign-return-address", i32 2} +// PART: !{i32 8, !"sign-return-address", i32 2} // BTE-NOT: !"sign-return-address" -// B-KEY: !{i32 8, !"sign-return-address", i32 1} +// B-KEY: !{i32 8, !"sign-return-address", i32 2} // NONE-NOT: !"sign-return-address-all" -// ALL: !{i32 8, !"sign-return-address-all", i32 1} +// ALL: !{i32 8, !"sign-return-address-all", i32 2} // PART-NOT: !"sign-return-address-all" // BTE-NOT: !"sign-return-address-all" // B-KEY-NOT: !"sign-return-address-all" @@ -47,6 +47,6 @@ // ALL-NOT: !"sign-return-address-with-bkey" // PART-NOT: !"sign-return-address-with-bkey" // BTE-NOT: !"sign-return-address-with-bkey" -// B-KEY: !{i32 8, !"sign-return-address-with-bkey", i32 1} +// B-KEY: !{i32 8, !"sign-return-address-with-bkey", i32 2} void foo() {} diff --git a/clang/test/CodeGen/LoongArch/inline-asm-operand-modifiers.c b/clang/test/CodeGen/LoongArch/inline-asm-operand-modifiers.c index b36fe7a..bbb9962 100644 --- a/clang/test/CodeGen/LoongArch/inline-asm-operand-modifiers.c +++ b/clang/test/CodeGen/LoongArch/inline-asm-operand-modifiers.c @@ -6,7 +6,7 @@ // CHECK-LABEL: @test_z_zero( // CHECK-NEXT: entry: -// CHECK-NEXT: [[TMP0:%.*]] = tail call i32 asm sideeffect "add.w $0, $1, ${2:z}", "=r,r,ri"(i32 [[A:%.*]], i32 0) #[[ATTR1:[0-9]+]], !srcloc !2 +// CHECK-NEXT: [[TMP0:%.*]] = tail call i32 asm sideeffect "add.w $0, $1, ${2:z}", "=r,r,ri"(i32 [[A:%.*]], i32 0) #[[ATTR1:[0-9]+]], !srcloc [[META6:![0-9]+]] // CHECK-NEXT: ret void // void test_z_zero(int a) { @@ -16,7 +16,7 @@ void test_z_zero(int a) { // CHECK-LABEL: @test_z_nonzero( // CHECK-NEXT: entry: -// CHECK-NEXT: [[TMP0:%.*]] = tail call i32 asm sideeffect "add.w $0, $1, ${2:z}", "=r,r,ri"(i32 [[A:%.*]], i32 1) #[[ATTR1]], !srcloc !3 +// CHECK-NEXT: [[TMP0:%.*]] = tail call i32 asm sideeffect "add.w $0, $1, ${2:z}", "=r,r,ri"(i32 [[A:%.*]], i32 1) #[[ATTR1]], !srcloc [[META7:![0-9]+]] // CHECK-NEXT: ret void // void test_z_nonzero(int a) { diff --git a/clang/test/CodeGen/LoongArch/lasx/builtin-alias.c b/clang/test/CodeGen/LoongArch/lasx/builtin-alias.c index dd094e5..03a746c 100644 --- a/clang/test/CodeGen/LoongArch/lasx/builtin-alias.c +++ b/clang/test/CodeGen/LoongArch/lasx/builtin-alias.c @@ -6,2808 +6,2808 @@ // CHECK-LABEL: define dso_local void @xvsll_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2:![0-9]+]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6:![0-9]+]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsll.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsll_b(v32i8 _1, v32i8 _2) { return __lasx_xvsll_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsll_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsll.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsll_h(v16i16 _1, v16i16 _2) { return __lasx_xvsll_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsll_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsll.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsll_w(v8i32 _1, v8i32 _2) { return __lasx_xvsll_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsll_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsll.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsll_d(v4i64 _1, v4i64 _2) { return __lasx_xvsll_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvslli_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvslli.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvslli_b(v32i8 _1) { return __lasx_xvslli_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvslli_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvslli.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvslli_h(v16i16 _1) { return __lasx_xvslli_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvslli_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvslli.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvslli_w(v8i32 _1) { return __lasx_xvslli_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvslli_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvslli.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvslli_d(v4i64 _1) { return __lasx_xvslli_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvsra_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsra.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsra_b(v32i8 _1, v32i8 _2) { return __lasx_xvsra_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsra_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsra.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsra_h(v16i16 _1, v16i16 _2) { return __lasx_xvsra_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsra_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsra.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsra_w(v8i32 _1, v8i32 _2) { return __lasx_xvsra_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsra_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsra.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsra_d(v4i64 _1, v4i64 _2) { return __lasx_xvsra_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrai_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrai.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrai_b(v32i8 _1) { return __lasx_xvsrai_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrai_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrai.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrai_h(v16i16 _1) { return __lasx_xvsrai_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrai_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrai.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrai_w(v8i32 _1) { return __lasx_xvsrai_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrai_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrai.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrai_d(v4i64 _1) { return __lasx_xvsrai_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrar_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrar.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrar_b(v32i8 _1, v32i8 _2) { return __lasx_xvsrar_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrar_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrar.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrar_h(v16i16 _1, v16i16 _2) { return __lasx_xvsrar_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrar_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrar.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrar_w(v8i32 _1, v8i32 _2) { return __lasx_xvsrar_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrar_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrar.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrar_d(v4i64 _1, v4i64 _2) { return __lasx_xvsrar_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrari_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrari.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrari_b(v32i8 _1) { return __lasx_xvsrari_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrari_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrari.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrari_h(v16i16 _1) { return __lasx_xvsrari_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrari_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrari.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrari_w(v8i32 _1) { return __lasx_xvsrari_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrari_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrari.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrari_d(v4i64 _1) { return __lasx_xvsrari_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrl_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrl.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrl_b(v32i8 _1, v32i8 _2) { return __lasx_xvsrl_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrl_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrl.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrl_h(v16i16 _1, v16i16 _2) { return __lasx_xvsrl_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrl_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrl.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrl_w(v8i32 _1, v8i32 _2) { return __lasx_xvsrl_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrl_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrl.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrl_d(v4i64 _1, v4i64 _2) { return __lasx_xvsrl_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrli_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrli.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrli_b(v32i8 _1) { return __lasx_xvsrli_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrli_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrli.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrli_h(v16i16 _1) { return __lasx_xvsrli_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrli_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrli.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrli_w(v8i32 _1) { return __lasx_xvsrli_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrli_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrli.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrli_d(v4i64 _1) { return __lasx_xvsrli_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrlr_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrlr.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrlr_b(v32i8 _1, v32i8 _2) { return __lasx_xvsrlr_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrlr_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrlr.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrlr_h(v16i16 _1, v16i16 _2) { return __lasx_xvsrlr_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrlr_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrlr.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrlr_w(v8i32 _1, v8i32 _2) { return __lasx_xvsrlr_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrlr_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrlr.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrlr_d(v4i64 _1, v4i64 _2) { return __lasx_xvsrlr_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrlri_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrlri.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrlri_b(v32i8 _1) { return __lasx_xvsrlri_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrlri_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrlri.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrlri_h(v16i16 _1) { return __lasx_xvsrlri_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrlri_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrlri.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrlri_w(v8i32 _1) { return __lasx_xvsrlri_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrlri_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrlri.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrlri_d(v4i64 _1) { return __lasx_xvsrlri_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitclr_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitclr.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitclr_b(v32u8 _1, v32u8 _2) { return __lasx_xvbitclr_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitclr_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvbitclr.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvbitclr_h(v16u16 _1, v16u16 _2) { return __lasx_xvbitclr_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitclr_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvbitclr.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvbitclr_w(v8u32 _1, v8u32 _2) { return __lasx_xvbitclr_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitclr_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvbitclr.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvbitclr_d(v4u64 _1, v4u64 _2) { return __lasx_xvbitclr_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitclri_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitclri.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitclri_b(v32u8 _1) { return __lasx_xvbitclri_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitclri_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvbitclri.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvbitclri_h(v16u16 _1) { return __lasx_xvbitclri_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitclri_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvbitclri.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvbitclri_w(v8u32 _1) { return __lasx_xvbitclri_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitclri_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvbitclri.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvbitclri_d(v4u64 _1) { return __lasx_xvbitclri_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitset_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitset.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitset_b(v32u8 _1, v32u8 _2) { return __lasx_xvbitset_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitset_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvbitset.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvbitset_h(v16u16 _1, v16u16 _2) { return __lasx_xvbitset_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitset_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvbitset.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvbitset_w(v8u32 _1, v8u32 _2) { return __lasx_xvbitset_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitset_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvbitset.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvbitset_d(v4u64 _1, v4u64 _2) { return __lasx_xvbitset_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitseti_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitseti.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitseti_b(v32u8 _1) { return __lasx_xvbitseti_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitseti_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvbitseti.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvbitseti_h(v16u16 _1) { return __lasx_xvbitseti_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitseti_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvbitseti.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvbitseti_w(v8u32 _1) { return __lasx_xvbitseti_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitseti_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvbitseti.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvbitseti_d(v4u64 _1) { return __lasx_xvbitseti_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitrev_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitrev.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitrev_b(v32u8 _1, v32u8 _2) { return __lasx_xvbitrev_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitrev_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvbitrev.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvbitrev_h(v16u16 _1, v16u16 _2) { return __lasx_xvbitrev_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitrev_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvbitrev.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvbitrev_w(v8u32 _1, v8u32 _2) { return __lasx_xvbitrev_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitrev_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvbitrev.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvbitrev_d(v4u64 _1, v4u64 _2) { return __lasx_xvbitrev_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitrevi_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitrevi.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitrevi_b(v32u8 _1) { return __lasx_xvbitrevi_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitrevi_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvbitrevi.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvbitrevi_h(v16u16 _1) { return __lasx_xvbitrevi_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitrevi_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvbitrevi.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvbitrevi_w(v8u32 _1) { return __lasx_xvbitrevi_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitrevi_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvbitrevi.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvbitrevi_d(v4u64 _1) { return __lasx_xvbitrevi_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvadd_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvadd.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvadd_b(v32i8 _1, v32i8 _2) { return __lasx_xvadd_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvadd_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvadd.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvadd_h(v16i16 _1, v16i16 _2) { return __lasx_xvadd_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvadd_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvadd.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvadd_w(v8i32 _1, v8i32 _2) { return __lasx_xvadd_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvadd_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvadd.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvadd_d(v4i64 _1, v4i64 _2) { return __lasx_xvadd_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddi_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvaddi.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvaddi_bu(v32i8 _1) { return __lasx_xvaddi_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvaddi_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvaddi.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvaddi_hu(v16i16 _1) { return __lasx_xvaddi_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvaddi_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvaddi.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvaddi_wu(v8i32 _1) { return __lasx_xvaddi_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvaddi_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddi.du(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddi_du(v4i64 _1) { return __lasx_xvaddi_du(_1, 1); } // CHECK-LABEL: define dso_local void @xvsub_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsub.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsub_b(v32i8 _1, v32i8 _2) { return __lasx_xvsub_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsub_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsub.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsub_h(v16i16 _1, v16i16 _2) { return __lasx_xvsub_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsub_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsub.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsub_w(v8i32 _1, v8i32 _2) { return __lasx_xvsub_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsub_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsub.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsub_d(v4i64 _1, v4i64 _2) { return __lasx_xvsub_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubi_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsubi.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsubi_bu(v32i8 _1) { return __lasx_xvsubi_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsubi_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsubi.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsubi_hu(v16i16 _1) { return __lasx_xvsubi_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsubi_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsubi.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsubi_wu(v8i32 _1) { return __lasx_xvsubi_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsubi_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubi.du(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubi_du(v4i64 _1) { return __lasx_xvsubi_du(_1, 1); } // CHECK-LABEL: define dso_local void @xvmax_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmax.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmax_b(v32i8 _1, v32i8 _2) { return __lasx_xvmax_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmax_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmax.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmax_h(v16i16 _1, v16i16 _2) { return __lasx_xvmax_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmax_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmax.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmax_w(v8i32 _1, v8i32 _2) { return __lasx_xvmax_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmax_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmax.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmax_d(v4i64 _1, v4i64 _2) { return __lasx_xvmax_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmaxi_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmaxi.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmaxi_b(v32i8 _1) { return __lasx_xvmaxi_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvmaxi_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaxi.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmaxi_h(v16i16 _1) { return __lasx_xvmaxi_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvmaxi_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaxi.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmaxi_w(v8i32 _1) { return __lasx_xvmaxi_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvmaxi_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaxi.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaxi_d(v4i64 _1) { return __lasx_xvmaxi_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvmax_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmax.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvmax_bu(v32u8 _1, v32u8 _2) { return __lasx_xvmax_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmax_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmax.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmax_hu(v16u16 _1, v16u16 _2) { return __lasx_xvmax_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmax_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmax.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmax_wu(v8u32 _1, v8u32 _2) { return __lasx_xvmax_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmax_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmax.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmax_du(v4u64 _1, v4u64 _2) { return __lasx_xvmax_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvmaxi_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmaxi.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvmaxi_bu(v32u8 _1) { return __lasx_xvmaxi_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvmaxi_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaxi.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmaxi_hu(v16u16 _1) { return __lasx_xvmaxi_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvmaxi_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaxi.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmaxi_wu(v8u32 _1) { return __lasx_xvmaxi_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvmaxi_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaxi.du(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmaxi_du(v4u64 _1) { return __lasx_xvmaxi_du(_1, 1); } // CHECK-LABEL: define dso_local void @xvmin_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmin.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmin_b(v32i8 _1, v32i8 _2) { return __lasx_xvmin_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmin_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmin.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmin_h(v16i16 _1, v16i16 _2) { return __lasx_xvmin_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmin_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmin.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmin_w(v8i32 _1, v8i32 _2) { return __lasx_xvmin_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmin_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmin.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmin_d(v4i64 _1, v4i64 _2) { return __lasx_xvmin_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmini_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmini.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmini_b(v32i8 _1) { return __lasx_xvmini_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvmini_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmini.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmini_h(v16i16 _1) { return __lasx_xvmini_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvmini_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmini.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmini_w(v8i32 _1) { return __lasx_xvmini_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvmini_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmini.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmini_d(v4i64 _1) { return __lasx_xvmini_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvmin_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmin.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvmin_bu(v32u8 _1, v32u8 _2) { return __lasx_xvmin_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmin_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmin.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmin_hu(v16u16 _1, v16u16 _2) { return __lasx_xvmin_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmin_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmin.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmin_wu(v8u32 _1, v8u32 _2) { return __lasx_xvmin_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmin_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmin.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmin_du(v4u64 _1, v4u64 _2) { return __lasx_xvmin_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvmini_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmini.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvmini_bu(v32u8 _1) { return __lasx_xvmini_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvmini_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmini.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmini_hu(v16u16 _1) { return __lasx_xvmini_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvmini_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmini.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmini_wu(v8u32 _1) { return __lasx_xvmini_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvmini_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmini.du(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmini_du(v4u64 _1) { return __lasx_xvmini_du(_1, 1); } // CHECK-LABEL: define dso_local void @xvseq_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvseq.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvseq_b(v32i8 _1, v32i8 _2) { return __lasx_xvseq_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvseq_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvseq.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvseq_h(v16i16 _1, v16i16 _2) { return __lasx_xvseq_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvseq_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvseq.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvseq_w(v8i32 _1, v8i32 _2) { return __lasx_xvseq_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvseq_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvseq.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvseq_d(v4i64 _1, v4i64 _2) { return __lasx_xvseq_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvseqi_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvseqi.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvseqi_b(v32i8 _1) { return __lasx_xvseqi_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvseqi_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvseqi.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvseqi_h(v16i16 _1) { return __lasx_xvseqi_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvseqi_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvseqi.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvseqi_w(v8i32 _1) { return __lasx_xvseqi_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvseqi_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvseqi.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvseqi_d(v4i64 _1) { return __lasx_xvseqi_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvslt_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvslt.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvslt_b(v32i8 _1, v32i8 _2) { return __lasx_xvslt_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvslt_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvslt.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvslt_h(v16i16 _1, v16i16 _2) { return __lasx_xvslt_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvslt_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvslt.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvslt_w(v8i32 _1, v8i32 _2) { return __lasx_xvslt_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvslt_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvslt.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvslt_d(v4i64 _1, v4i64 _2) { return __lasx_xvslt_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvslti_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvslti.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvslti_b(v32i8 _1) { return __lasx_xvslti_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvslti_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvslti.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvslti_h(v16i16 _1) { return __lasx_xvslti_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvslti_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvslti.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvslti_w(v8i32 _1) { return __lasx_xvslti_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvslti_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvslti.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvslti_d(v4i64 _1) { return __lasx_xvslti_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvslt_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvslt.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvslt_bu(v32u8 _1, v32u8 _2) { return __lasx_xvslt_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvslt_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvslt.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvslt_hu(v16u16 _1, v16u16 _2) { return __lasx_xvslt_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvslt_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvslt.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvslt_wu(v8u32 _1, v8u32 _2) { return __lasx_xvslt_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvslt_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvslt.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvslt_du(v4u64 _1, v4u64 _2) { return __lasx_xvslt_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvslti_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvslti.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvslti_bu(v32u8 _1) { return __lasx_xvslti_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvslti_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvslti.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvslti_hu(v16u16 _1) { return __lasx_xvslti_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvslti_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvslti.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvslti_wu(v8u32 _1) { return __lasx_xvslti_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvslti_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvslti.du(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvslti_du(v4u64 _1) { return __lasx_xvslti_du(_1, 1); } // CHECK-LABEL: define dso_local void @xvsle_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsle.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsle_b(v32i8 _1, v32i8 _2) { return __lasx_xvsle_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsle_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsle.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsle_h(v16i16 _1, v16i16 _2) { return __lasx_xvsle_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsle_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsle.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsle_w(v8i32 _1, v8i32 _2) { return __lasx_xvsle_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsle_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsle.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsle_d(v4i64 _1, v4i64 _2) { return __lasx_xvsle_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvslei_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvslei.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvslei_b(v32i8 _1) { return __lasx_xvslei_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvslei_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvslei.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvslei_h(v16i16 _1) { return __lasx_xvslei_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvslei_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvslei.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvslei_w(v8i32 _1) { return __lasx_xvslei_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvslei_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvslei.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvslei_d(v4i64 _1) { return __lasx_xvslei_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvsle_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsle.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsle_bu(v32u8 _1, v32u8 _2) { return __lasx_xvsle_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsle_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsle.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsle_hu(v16u16 _1, v16u16 _2) { return __lasx_xvsle_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsle_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsle.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsle_wu(v8u32 _1, v8u32 _2) { return __lasx_xvsle_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsle_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsle.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsle_du(v4u64 _1, v4u64 _2) { return __lasx_xvsle_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvslei_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvslei.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvslei_bu(v32u8 _1) { return __lasx_xvslei_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvslei_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvslei.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvslei_hu(v16u16 _1) { return __lasx_xvslei_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvslei_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvslei.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvslei_wu(v8u32 _1) { return __lasx_xvslei_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvslei_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvslei.du(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvslei_du(v4u64 _1) { return __lasx_xvslei_du(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsat.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsat_b(v32i8 _1) { return __lasx_xvsat_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsat.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsat_h(v16i16 _1) { return __lasx_xvsat_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsat.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsat_w(v8i32 _1) { return __lasx_xvsat_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsat.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsat_d(v4i64 _1) { return __lasx_xvsat_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsat.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvsat_bu(v32u8 _1) { return __lasx_xvsat_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsat.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvsat_hu(v16u16 _1) { return __lasx_xvsat_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsat.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvsat_wu(v8u32 _1) { return __lasx_xvsat_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsat.du(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvsat_du(v4u64 _1) { return __lasx_xvsat_du(_1, 1); } // CHECK-LABEL: define dso_local void @xvadda_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvadda.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvadda_b(v32i8 _1, v32i8 _2) { return __lasx_xvadda_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvadda_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvadda.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvadda_h(v16i16 _1, v16i16 _2) { return __lasx_xvadda_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvadda_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvadda.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvadda_w(v8i32 _1, v8i32 _2) { return __lasx_xvadda_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvadda_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvadda.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvadda_d(v4i64 _1, v4i64 _2) { return __lasx_xvadda_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsadd.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsadd_b(v32i8 _1, v32i8 _2) { return __lasx_xvsadd_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsadd.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsadd_h(v16i16 _1, v16i16 _2) { return __lasx_xvsadd_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsadd.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsadd_w(v8i32 _1, v8i32 _2) { return __lasx_xvsadd_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsadd.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsadd_d(v4i64 _1, v4i64 _2) { return __lasx_xvsadd_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsadd.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvsadd_bu(v32u8 _1, v32u8 _2) { return __lasx_xvsadd_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsadd.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvsadd_hu(v16u16 _1, v16u16 _2) { return __lasx_xvsadd_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsadd.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvsadd_wu(v8u32 _1, v8u32 _2) { return __lasx_xvsadd_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsadd.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvsadd_du(v4u64 _1, v4u64 _2) { return __lasx_xvsadd_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvavg.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvavg_b(v32i8 _1, v32i8 _2) { return __lasx_xvavg_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvavg.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvavg_h(v16i16 _1, v16i16 _2) { return __lasx_xvavg_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvavg.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvavg_w(v8i32 _1, v8i32 _2) { return __lasx_xvavg_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvavg.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvavg_d(v4i64 _1, v4i64 _2) { return __lasx_xvavg_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvavg.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvavg_bu(v32u8 _1, v32u8 _2) { return __lasx_xvavg_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvavg.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvavg_hu(v16u16 _1, v16u16 _2) { return __lasx_xvavg_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvavg.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvavg_wu(v8u32 _1, v8u32 _2) { return __lasx_xvavg_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvavg.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvavg_du(v4u64 _1, v4u64 _2) { return __lasx_xvavg_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvavgr.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvavgr_b(v32i8 _1, v32i8 _2) { return __lasx_xvavgr_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvavgr.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvavgr_h(v16i16 _1, v16i16 _2) { return __lasx_xvavgr_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvavgr.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvavgr_w(v8i32 _1, v8i32 _2) { return __lasx_xvavgr_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvavgr.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvavgr_d(v4i64 _1, v4i64 _2) { return __lasx_xvavgr_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvavgr.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvavgr_bu(v32u8 _1, v32u8 _2) { return __lasx_xvavgr_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvavgr.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvavgr_hu(v16u16 _1, v16u16 _2) { return __lasx_xvavgr_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvavgr.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvavgr_wu(v8u32 _1, v8u32 _2) { return __lasx_xvavgr_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvavgr.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvavgr_du(v4u64 _1, v4u64 _2) { return __lasx_xvavgr_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssub.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssub_b(v32i8 _1, v32i8 _2) { return __lasx_xvssub_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssub.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssub_h(v16i16 _1, v16i16 _2) { return __lasx_xvssub_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssub.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssub_w(v8i32 _1, v8i32 _2) { return __lasx_xvssub_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssub.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvssub_d(v4i64 _1, v4i64 _2) { return __lasx_xvssub_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssub.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssub_bu(v32u8 _1, v32u8 _2) { return __lasx_xvssub_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssub.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssub_hu(v16u16 _1, v16u16 _2) { return __lasx_xvssub_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssub.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssub_wu(v8u32 _1, v8u32 _2) { return __lasx_xvssub_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssub.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvssub_du(v4u64 _1, v4u64 _2) { return __lasx_xvssub_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvabsd.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvabsd_b(v32i8 _1, v32i8 _2) { return __lasx_xvabsd_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvabsd.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvabsd_h(v16i16 _1, v16i16 _2) { return __lasx_xvabsd_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvabsd.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvabsd_w(v8i32 _1, v8i32 _2) { return __lasx_xvabsd_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvabsd.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvabsd_d(v4i64 _1, v4i64 _2) { return __lasx_xvabsd_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvabsd.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvabsd_bu(v32u8 _1, v32u8 _2) { return __lasx_xvabsd_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvabsd.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvabsd_hu(v16u16 _1, v16u16 _2) { return __lasx_xvabsd_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvabsd.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvabsd_wu(v8u32 _1, v8u32 _2) { return __lasx_xvabsd_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvabsd.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvabsd_du(v4u64 _1, v4u64 _2) { return __lasx_xvabsd_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvmul_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmul.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmul_b(v32i8 _1, v32i8 _2) { return __lasx_xvmul_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmul_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmul.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmul_h(v16i16 _1, v16i16 _2) { return __lasx_xvmul_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmul_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmul.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmul_w(v8i32 _1, v8i32 _2) { return __lasx_xvmul_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmul_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmul.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmul_d(v4i64 _1, v4i64 _2) { return __lasx_xvmul_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmadd_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmadd.b(<32 x i8> [[_136]], <32 x i8> [[_247]], <32 x i8> [[_358]]) -// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmadd_b(v32i8 _1, v32i8 _2, v32i8 _3) { return __lasx_xvmadd_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmadd_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmadd.h(<16 x i16> [[_136]], <16 x i16> [[_247]], <16 x i16> [[_358]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmadd_h(v16i16 _1, v16i16 _2, v16i16 _3) { return __lasx_xvmadd_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmadd_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmadd.w(<8 x i32> [[_136]], <8 x i32> [[_247]], <8 x i32> [[_358]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmadd_w(v8i32 _1, v8i32 _2, v8i32 _3) { return __lasx_xvmadd_w(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmadd_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmadd.d(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmadd_d(v4i64 _1, v4i64 _2, v4i64 _3) { return __lasx_xvmadd_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmsub_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmsub.b(<32 x i8> [[_136]], <32 x i8> [[_247]], <32 x i8> [[_358]]) -// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmsub_b(v32i8 _1, v32i8 _2, v32i8 _3) { return __lasx_xvmsub_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmsub_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmsub.h(<16 x i16> [[_136]], <16 x i16> [[_247]], <16 x i16> [[_358]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmsub_h(v16i16 _1, v16i16 _2, v16i16 _3) { return __lasx_xvmsub_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmsub_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmsub.w(<8 x i32> [[_136]], <8 x i32> [[_247]], <8 x i32> [[_358]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmsub_w(v8i32 _1, v8i32 _2, v8i32 _3) { return __lasx_xvmsub_w(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmsub_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmsub.d(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmsub_d(v4i64 _1, v4i64 _2, v4i64 _3) { return __lasx_xvmsub_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvdiv_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvdiv.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvdiv_b(v32i8 _1, v32i8 _2) { return __lasx_xvdiv_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvdiv_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvdiv.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvdiv_h(v16i16 _1, v16i16 _2) { return __lasx_xvdiv_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvdiv_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvdiv.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvdiv_w(v8i32 _1, v8i32 _2) { return __lasx_xvdiv_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvdiv_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvdiv.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvdiv_d(v4i64 _1, v4i64 _2) { return __lasx_xvdiv_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvdiv_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvdiv.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvdiv_bu(v32u8 _1, v32u8 _2) { return __lasx_xvdiv_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvdiv_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvdiv.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvdiv_hu(v16u16 _1, v16u16 _2) { return __lasx_xvdiv_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvdiv_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvdiv.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvdiv_wu(v8u32 _1, v8u32 _2) { return __lasx_xvdiv_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvdiv_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvdiv.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvdiv_du(v4u64 _1, v4u64 _2) { return __lasx_xvdiv_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvhaddw.h.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvhaddw_h_b(v32i8 _1, v32i8 _2) { return __lasx_xvhaddw_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvhaddw.w.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvhaddw_w_h(v16i16 _1, v16i16 _2) { return __lasx_xvhaddw_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhaddw.d.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvhaddw_d_w(v8i32 _1, v8i32 _2) { return __lasx_xvhaddw_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_hu_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvhaddw.hu.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvhaddw_hu_bu(v32u8 _1, v32u8 _2) { return __lasx_xvhaddw_hu_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_wu_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvhaddw.wu.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvhaddw_wu_hu(v16u16 _1, v16u16 _2) { return __lasx_xvhaddw_wu_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_du_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhaddw.du.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvhaddw_du_wu(v8u32 _1, v8u32 _2) { return __lasx_xvhaddw_du_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvhsubw.h.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvhsubw_h_b(v32i8 _1, v32i8 _2) { return __lasx_xvhsubw_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvhsubw.w.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvhsubw_w_h(v16i16 _1, v16i16 _2) { return __lasx_xvhsubw_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhsubw.d.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvhsubw_d_w(v8i32 _1, v8i32 _2) { return __lasx_xvhsubw_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_hu_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvhsubw.hu.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvhsubw_hu_bu(v32u8 _1, v32u8 _2) { return __lasx_xvhsubw_hu_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_wu_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvhsubw.wu.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvhsubw_wu_hu(v16u16 _1, v16u16 _2) { return __lasx_xvhsubw_wu_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_du_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhsubw.du.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvhsubw_du_wu(v8u32 _1, v8u32 _2) { return __lasx_xvhsubw_du_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmod.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmod_b(v32i8 _1, v32i8 _2) { return __lasx_xvmod_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmod.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmod_h(v16i16 _1, v16i16 _2) { return __lasx_xvmod_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmod.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmod_w(v8i32 _1, v8i32 _2) { return __lasx_xvmod_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmod.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmod_d(v4i64 _1, v4i64 _2) { return __lasx_xvmod_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmod.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvmod_bu(v32u8 _1, v32u8 _2) { return __lasx_xvmod_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmod.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmod_hu(v16u16 _1, v16u16 _2) { return __lasx_xvmod_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmod.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmod_wu(v8u32 _1, v8u32 _2) { return __lasx_xvmod_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmod.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmod_du(v4u64 _1, v4u64 _2) { return __lasx_xvmod_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvrepl128vei_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvrepl128vei.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvrepl128vei_b(v32i8 _1) { return __lasx_xvrepl128vei_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvrepl128vei_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvrepl128vei.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvrepl128vei_h(v16i16 _1) { return __lasx_xvrepl128vei_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvrepl128vei_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvrepl128vei.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvrepl128vei_w(v8i32 _1) { return __lasx_xvrepl128vei_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvrepl128vei_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvrepl128vei.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvrepl128vei_d(v4i64 _1) { return __lasx_xvrepl128vei_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvpickev_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvpickev.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvpickev_b(v32i8 _1, v32i8 _2) { return __lasx_xvpickev_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickev_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvpickev.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvpickev_h(v16i16 _1, v16i16 _2) { return __lasx_xvpickev_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickev_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvpickev.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvpickev_w(v8i32 _1, v8i32 _2) { return __lasx_xvpickev_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickev_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvpickev.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvpickev_d(v4i64 _1, v4i64 _2) { return __lasx_xvpickev_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickod_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvpickod.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvpickod_b(v32i8 _1, v32i8 _2) { return __lasx_xvpickod_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickod_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvpickod.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvpickod_h(v16i16 _1, v16i16 _2) { return __lasx_xvpickod_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickod_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvpickod.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvpickod_w(v8i32 _1, v8i32 _2) { return __lasx_xvpickod_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickod_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvpickod.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvpickod_d(v4i64 _1, v4i64 _2) { return __lasx_xvpickod_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvh_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvilvh.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvilvh_b(v32i8 _1, v32i8 _2) { return __lasx_xvilvh_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvh_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvilvh.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvilvh_h(v16i16 _1, v16i16 _2) { return __lasx_xvilvh_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvh_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvilvh.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvilvh_w(v8i32 _1, v8i32 _2) { return __lasx_xvilvh_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvh_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvilvh.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvilvh_d(v4i64 _1, v4i64 _2) { return __lasx_xvilvh_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvl_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvilvl.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvilvl_b(v32i8 _1, v32i8 _2) { return __lasx_xvilvl_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvl_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvilvl.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvilvl_h(v16i16 _1, v16i16 _2) { return __lasx_xvilvl_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvl_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvilvl.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvilvl_w(v8i32 _1, v8i32 _2) { return __lasx_xvilvl_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvl_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvilvl.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvilvl_d(v4i64 _1, v4i64 _2) { return __lasx_xvilvl_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackev_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvpackev.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvpackev_b(v32i8 _1, v32i8 _2) { return __lasx_xvpackev_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackev_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvpackev.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvpackev_h(v16i16 _1, v16i16 _2) { return __lasx_xvpackev_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackev_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvpackev.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvpackev_w(v8i32 _1, v8i32 _2) { return __lasx_xvpackev_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackev_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvpackev.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvpackev_d(v4i64 _1, v4i64 _2) { return __lasx_xvpackev_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackod_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvpackod.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvpackod_b(v32i8 _1, v32i8 _2) { return __lasx_xvpackod_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackod_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvpackod.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvpackod_h(v16i16 _1, v16i16 _2) { return __lasx_xvpackod_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackod_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvpackod.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvpackod_w(v8i32 _1, v8i32 _2) { return __lasx_xvpackod_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackod_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvpackod.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvpackod_d(v4i64 _1, v4i64 _2) { return __lasx_xvpackod_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvshuf_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvshuf.b(<32 x i8> [[_136]], <32 x i8> [[_247]], <32 x i8> [[_358]]) -// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvshuf_b(v32i8 _1, v32i8 _2, v32i8 _3) { return __lasx_xvshuf_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvshuf_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvshuf.h(<16 x i16> [[_136]], <16 x i16> [[_247]], <16 x i16> [[_358]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvshuf_h(v16i16 _1, v16i16 _2, v16i16 _3) { return __lasx_xvshuf_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvshuf_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvshuf.w(<8 x i32> [[_136]], <8 x i32> [[_247]], <8 x i32> [[_358]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvshuf_w(v8i32 _1, v8i32 _2, v8i32 _3) { return __lasx_xvshuf_w(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvshuf_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvshuf.d(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvshuf_d(v4i64 _1, v4i64 _2, v4i64 _3) { return __lasx_xvshuf_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvand_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvand.v(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvand_v(v32u8 _1, v32u8 _2) { return __lasx_xvand_v(_1, _2); } // CHECK-LABEL: define dso_local void @xvandi_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvandi.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvandi_b(v32u8 _1) { return __lasx_xvandi_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvor_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvor.v(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvor_v(v32u8 _1, v32u8 _2) { return __lasx_xvor_v(_1, _2); } // CHECK-LABEL: define dso_local void @xvori_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvori.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvori_b(v32u8 _1) { return __lasx_xvori_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvnor_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvnor.v(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvnor_v(v32u8 _1, v32u8 _2) { return __lasx_xvnor_v(_1, _2); } // CHECK-LABEL: define dso_local void @xvnori_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvnori.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvnori_b(v32u8 _1) { return __lasx_xvnori_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvxor_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvxor.v(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvxor_v(v32u8 _1, v32u8 _2) { return __lasx_xvxor_v(_1, _2); } // CHECK-LABEL: define dso_local void @xvxori_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvxori.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvxori_b(v32u8 _1) { return __lasx_xvxori_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitsel_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitsel.v(<32 x i8> [[_136]], <32 x i8> [[_247]], <32 x i8> [[_358]]) -// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitsel_v(v32u8 _1, v32u8 _2, v32u8 _3) { return __lasx_xvbitsel_v(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvbitseli_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitseli.b(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitseli_b(v32u8 _1, v32u8 _2) { return __lasx_xvbitseli_b(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvshuf4i_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvshuf4i.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvshuf4i_b(v32i8 _1) { return __lasx_xvshuf4i_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvshuf4i_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvshuf4i.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvshuf4i_h(v16i16 _1) { return __lasx_xvshuf4i_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvshuf4i_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvshuf4i.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvshuf4i_w(v8i32 _1) { return __lasx_xvshuf4i_w(_1, 1); } @@ -2815,7 +2815,7 @@ v8i32 xvshuf4i_w(v8i32 _1) { return __lasx_xvshuf4i_w(_1, 1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], i32 noundef signext [[_1:%.*]]) local_unnamed_addr #[[ATTR2:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvreplgr2vr.b(i32 [[_1]]) -// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvreplgr2vr_b(int _1) { return __lasx_xvreplgr2vr_b(_1); } @@ -2823,7 +2823,7 @@ v32i8 xvreplgr2vr_b(int _1) { return __lasx_xvreplgr2vr_b(_1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], i32 noundef signext [[_1:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvreplgr2vr.h(i32 [[_1]]) -// CHECK-NEXT: store <16 x i16> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvreplgr2vr_h(int _1) { return __lasx_xvreplgr2vr_h(_1); } @@ -2831,7 +2831,7 @@ v16i16 xvreplgr2vr_h(int _1) { return __lasx_xvreplgr2vr_h(_1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], i32 noundef signext [[_1:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvreplgr2vr.w(i32 [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvreplgr2vr_w(int _1) { return __lasx_xvreplgr2vr_w(_1); } @@ -2840,1641 +2840,1641 @@ v8i32 xvreplgr2vr_w(int _1) { return __lasx_xvreplgr2vr_w(_1); } // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[CONV:%.*]] = sext i32 [[_1]] to i64 // CHECK-NEXT: [[TMP0:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvreplgr2vr.d(i64 [[CONV]]) -// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvreplgr2vr_d(int _1) { return __lasx_xvreplgr2vr_d(_1); } // CHECK-LABEL: define dso_local void @xvpcnt_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvpcnt.b(<32 x i8> [[_112]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvpcnt_b(v32i8 _1) { return __lasx_xvpcnt_b(_1); } // CHECK-LABEL: define dso_local void @xvpcnt_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvpcnt.h(<16 x i16> [[_112]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvpcnt_h(v16i16 _1) { return __lasx_xvpcnt_h(_1); } // CHECK-LABEL: define dso_local void @xvpcnt_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvpcnt.w(<8 x i32> [[_112]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvpcnt_w(v8i32 _1) { return __lasx_xvpcnt_w(_1); } // CHECK-LABEL: define dso_local void @xvpcnt_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvpcnt.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvpcnt_d(v4i64 _1) { return __lasx_xvpcnt_d(_1); } // CHECK-LABEL: define dso_local void @xvclo_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvclo.b(<32 x i8> [[_112]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvclo_b(v32i8 _1) { return __lasx_xvclo_b(_1); } // CHECK-LABEL: define dso_local void @xvclo_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvclo.h(<16 x i16> [[_112]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvclo_h(v16i16 _1) { return __lasx_xvclo_h(_1); } // CHECK-LABEL: define dso_local void @xvclo_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvclo.w(<8 x i32> [[_112]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvclo_w(v8i32 _1) { return __lasx_xvclo_w(_1); } // CHECK-LABEL: define dso_local void @xvclo_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvclo.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvclo_d(v4i64 _1) { return __lasx_xvclo_d(_1); } // CHECK-LABEL: define dso_local void @xvclz_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvclz.b(<32 x i8> [[_112]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvclz_b(v32i8 _1) { return __lasx_xvclz_b(_1); } // CHECK-LABEL: define dso_local void @xvclz_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvclz.h(<16 x i16> [[_112]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvclz_h(v16i16 _1) { return __lasx_xvclz_h(_1); } // CHECK-LABEL: define dso_local void @xvclz_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvclz.w(<8 x i32> [[_112]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvclz_w(v8i32 _1) { return __lasx_xvclz_w(_1); } // CHECK-LABEL: define dso_local void @xvclz_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvclz.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvclz_d(v4i64 _1) { return __lasx_xvclz_d(_1); } // CHECK-LABEL: define dso_local void @xvfadd_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfadd.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfadd_s(v8f32 _1, v8f32 _2) { return __lasx_xvfadd_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfadd_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfadd.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfadd_d(v4f64 _1, v4f64 _2) { return __lasx_xvfadd_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfsub_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfsub.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfsub_s(v8f32 _1, v8f32 _2) { return __lasx_xvfsub_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfsub_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfsub.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfsub_d(v4f64 _1, v4f64 _2) { return __lasx_xvfsub_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmul_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfmul.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfmul_s(v8f32 _1, v8f32 _2) { return __lasx_xvfmul_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmul_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfmul.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfmul_d(v4f64 _1, v4f64 _2) { return __lasx_xvfmul_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfdiv_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfdiv.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfdiv_s(v8f32 _1, v8f32 _2) { return __lasx_xvfdiv_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfdiv_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfdiv.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfdiv_d(v4f64 _1, v4f64 _2) { return __lasx_xvfdiv_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcvt_h_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvfcvt.h.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvfcvt_h_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcvt_h_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcvt_s_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfcvt.s.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfcvt_s_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcvt_s_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmin_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfmin.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfmin_s(v8f32 _1, v8f32 _2) { return __lasx_xvfmin_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmin_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfmin.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfmin_d(v4f64 _1, v4f64 _2) { return __lasx_xvfmin_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmina_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfmina.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfmina_s(v8f32 _1, v8f32 _2) { return __lasx_xvfmina_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmina_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfmina.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfmina_d(v4f64 _1, v4f64 _2) { return __lasx_xvfmina_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmax_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfmax.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfmax_s(v8f32 _1, v8f32 _2) { return __lasx_xvfmax_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmax_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfmax.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfmax_d(v4f64 _1, v4f64 _2) { return __lasx_xvfmax_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmaxa_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfmaxa.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfmaxa_s(v8f32 _1, v8f32 _2) { return __lasx_xvfmaxa_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmaxa_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfmaxa.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfmaxa_d(v4f64 _1, v4f64 _2) { return __lasx_xvfmaxa_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfclass_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfclass.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfclass_s(v8f32 _1) { return __lasx_xvfclass_s(_1); } // CHECK-LABEL: define dso_local void @xvfclass_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfclass.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfclass_d(v4f64 _1) { return __lasx_xvfclass_d(_1); } // CHECK-LABEL: define dso_local void @xvfsqrt_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfsqrt.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfsqrt_s(v8f32 _1) { return __lasx_xvfsqrt_s(_1); } // CHECK-LABEL: define dso_local void @xvfsqrt_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfsqrt.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfsqrt_d(v4f64 _1) { return __lasx_xvfsqrt_d(_1); } // CHECK-LABEL: define dso_local void @xvfrecip_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrecip.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfrecip_s(v8f32 _1) { return __lasx_xvfrecip_s(_1); } // CHECK-LABEL: define dso_local void @xvfrecip_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrecip.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfrecip_d(v4f64 _1) { return __lasx_xvfrecip_d(_1); } // CHECK-LABEL: define dso_local void @xvfrint_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrint.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfrint_s(v8f32 _1) { return __lasx_xvfrint_s(_1); } // CHECK-LABEL: define dso_local void @xvfrint_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrint.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfrint_d(v4f64 _1) { return __lasx_xvfrint_d(_1); } // CHECK-LABEL: define dso_local void @xvfrsqrt_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrsqrt.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfrsqrt_s(v8f32 _1) { return __lasx_xvfrsqrt_s(_1); } // CHECK-LABEL: define dso_local void @xvfrsqrt_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrsqrt.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfrsqrt_d(v4f64 _1) { return __lasx_xvfrsqrt_d(_1); } // CHECK-LABEL: define dso_local void @xvflogb_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvflogb.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvflogb_s(v8f32 _1) { return __lasx_xvflogb_s(_1); } // CHECK-LABEL: define dso_local void @xvflogb_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvflogb.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvflogb_d(v4f64 _1) { return __lasx_xvflogb_d(_1); } // CHECK-LABEL: define dso_local void @xvfcvth_s_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfcvth.s.h(<16 x i16> [[_112]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfcvth_s_h(v16i16 _1) { return __lasx_xvfcvth_s_h(_1); } // CHECK-LABEL: define dso_local void @xvfcvth_d_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfcvth.d.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfcvth_d_s(v8f32 _1) { return __lasx_xvfcvth_d_s(_1); } // CHECK-LABEL: define dso_local void @xvfcvtl_s_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfcvtl.s.h(<16 x i16> [[_112]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfcvtl_s_h(v16i16 _1) { return __lasx_xvfcvtl_s_h(_1); } // CHECK-LABEL: define dso_local void @xvfcvtl_d_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfcvtl.d.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfcvtl_d_s(v8f32 _1) { return __lasx_xvfcvtl_d_s(_1); } // CHECK-LABEL: define dso_local void @xvftint_w_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftint.w.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftint_w_s(v8f32 _1) { return __lasx_xvftint_w_s(_1); } // CHECK-LABEL: define dso_local void @xvftint_l_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftint.l.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftint_l_d(v4f64 _1) { return __lasx_xvftint_l_d(_1); } // CHECK-LABEL: define dso_local void @xvftint_wu_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftint.wu.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvftint_wu_s(v8f32 _1) { return __lasx_xvftint_wu_s(_1); } // CHECK-LABEL: define dso_local void @xvftint_lu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftint.lu.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvftint_lu_d(v4f64 _1) { return __lasx_xvftint_lu_d(_1); } // CHECK-LABEL: define dso_local void @xvftintrz_w_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrz.w.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrz_w_s(v8f32 _1) { return __lasx_xvftintrz_w_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrz_l_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrz.l.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrz_l_d(v4f64 _1) { return __lasx_xvftintrz_l_d(_1); } // CHECK-LABEL: define dso_local void @xvftintrz_wu_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrz.wu.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvftintrz_wu_s(v8f32 _1) { return __lasx_xvftintrz_wu_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrz_lu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrz.lu.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvftintrz_lu_d(v4f64 _1) { return __lasx_xvftintrz_lu_d(_1); } // CHECK-LABEL: define dso_local void @xvffint_s_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvffint.s.w(<8 x i32> [[_112]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvffint_s_w(v8i32 _1) { return __lasx_xvffint_s_w(_1); } // CHECK-LABEL: define dso_local void @xvffint_d_l( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvffint.d.l(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvffint_d_l(v4i64 _1) { return __lasx_xvffint_d_l(_1); } // CHECK-LABEL: define dso_local void @xvffint_s_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvffint.s.wu(<8 x i32> [[_112]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvffint_s_wu(v8u32 _1) { return __lasx_xvffint_s_wu(_1); } // CHECK-LABEL: define dso_local void @xvffint_d_lu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvffint.d.lu(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvffint_d_lu(v4u64 _1) { return __lasx_xvffint_d_lu(_1); } // CHECK-LABEL: define dso_local void @xvreplve_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], i32 noundef signext [[_2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvreplve.b(<32 x i8> [[_112]], i32 [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvreplve_b(v32i8 _1, int _2) { return __lasx_xvreplve_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvreplve_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], i32 noundef signext [[_2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvreplve.h(<16 x i16> [[_112]], i32 [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvreplve_h(v16i16 _1, int _2) { return __lasx_xvreplve_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvreplve_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], i32 noundef signext [[_2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvreplve.w(<8 x i32> [[_112]], i32 [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvreplve_w(v8i32 _1, int _2) { return __lasx_xvreplve_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvreplve_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], i32 noundef signext [[_2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvreplve.d(<4 x i64> [[_1]], i32 [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvreplve_d(v4i64 _1, int _2) { return __lasx_xvreplve_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvpermi_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvpermi.w(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvpermi_w(v8i32 _1, v8i32 _2) { return __lasx_xvpermi_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvandn_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvandn.v(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvandn_v(v32u8 _1, v32u8 _2) { return __lasx_xvandn_v(_1, _2); } // CHECK-LABEL: define dso_local void @xvneg_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvneg.b(<32 x i8> [[_112]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvneg_b(v32i8 _1) { return __lasx_xvneg_b(_1); } // CHECK-LABEL: define dso_local void @xvneg_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvneg.h(<16 x i16> [[_112]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvneg_h(v16i16 _1) { return __lasx_xvneg_h(_1); } // CHECK-LABEL: define dso_local void @xvneg_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvneg.w(<8 x i32> [[_112]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvneg_w(v8i32 _1) { return __lasx_xvneg_w(_1); } // CHECK-LABEL: define dso_local void @xvneg_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvneg.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvneg_d(v4i64 _1) { return __lasx_xvneg_d(_1); } // CHECK-LABEL: define dso_local void @xvmuh_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmuh.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmuh_b(v32i8 _1, v32i8 _2) { return __lasx_xvmuh_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmuh_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmuh.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmuh_h(v16i16 _1, v16i16 _2) { return __lasx_xvmuh_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmuh_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmuh.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmuh_w(v8i32 _1, v8i32 _2) { return __lasx_xvmuh_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmuh_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmuh.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmuh_d(v4i64 _1, v4i64 _2) { return __lasx_xvmuh_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmuh_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmuh.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvmuh_bu(v32u8 _1, v32u8 _2) { return __lasx_xvmuh_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmuh_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmuh.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmuh_hu(v16u16 _1, v16u16 _2) { return __lasx_xvmuh_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmuh_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmuh.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmuh_wu(v8u32 _1, v8u32 _2) { return __lasx_xvmuh_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmuh_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmuh.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmuh_du(v4u64 _1, v4u64 _2) { return __lasx_xvmuh_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvsllwil_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsllwil.h.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsllwil_h_b(v32i8 _1) { return __lasx_xvsllwil_h_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvsllwil_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsllwil.w.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsllwil_w_h(v16i16 _1) { return __lasx_xvsllwil_w_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvsllwil_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsllwil.d.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsllwil_d_w(v8i32 _1) { return __lasx_xvsllwil_d_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvsllwil_hu_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsllwil.hu.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvsllwil_hu_bu(v32u8 _1) { return __lasx_xvsllwil_hu_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsllwil_wu_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsllwil.wu.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvsllwil_wu_hu(v16u16 _1) { return __lasx_xvsllwil_wu_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsllwil_du_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsllwil.du.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvsllwil_du_wu(v8u32 _1) { return __lasx_xvsllwil_du_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsran_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsran.b.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsran_b_h(v16i16 _1, v16i16 _2) { return __lasx_xvsran_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsran_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsran.h.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsran_h_w(v8i32 _1, v8i32 _2) { return __lasx_xvsran_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsran_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsran.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsran_w_d(v4i64 _1, v4i64 _2) { return __lasx_xvsran_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssran_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssran.b.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssran_b_h(v16i16 _1, v16i16 _2) { return __lasx_xvssran_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssran_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssran.h.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssran_h_w(v8i32 _1, v8i32 _2) { return __lasx_xvssran_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssran_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssran.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssran_w_d(v4i64 _1, v4i64 _2) { return __lasx_xvssran_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssran_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssran.bu.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssran_bu_h(v16u16 _1, v16u16 _2) { return __lasx_xvssran_bu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssran_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssran.hu.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssran_hu_w(v8u32 _1, v8u32 _2) { return __lasx_xvssran_hu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssran_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssran.wu.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssran_wu_d(v4u64 _1, v4u64 _2) { return __lasx_xvssran_wu_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrarn_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrarn.b.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrarn_b_h(v16i16 _1, v16i16 _2) { return __lasx_xvsrarn_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrarn_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrarn.h.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrarn_h_w(v8i32 _1, v8i32 _2) { return __lasx_xvsrarn_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrarn_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrarn.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrarn_w_d(v4i64 _1, v4i64 _2) { return __lasx_xvsrarn_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrarn_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrarn.b.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssrarn_b_h(v16i16 _1, v16i16 _2) { return __lasx_xvssrarn_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrarn_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrarn.h.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssrarn_h_w(v8i32 _1, v8i32 _2) { return __lasx_xvssrarn_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrarn_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrarn.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssrarn_w_d(v4i64 _1, v4i64 _2) { return __lasx_xvssrarn_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrarn_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrarn.bu.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssrarn_bu_h(v16u16 _1, v16u16 _2) { return __lasx_xvssrarn_bu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrarn_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrarn.hu.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssrarn_hu_w(v8u32 _1, v8u32 _2) { return __lasx_xvssrarn_hu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrarn_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrarn.wu.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssrarn_wu_d(v4u64 _1, v4u64 _2) { return __lasx_xvssrarn_wu_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrln_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrln.b.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrln_b_h(v16i16 _1, v16i16 _2) { return __lasx_xvsrln_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrln_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrln.h.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrln_h_w(v8i32 _1, v8i32 _2) { return __lasx_xvsrln_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrln_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrln.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrln_w_d(v4i64 _1, v4i64 _2) { return __lasx_xvsrln_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrln_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrln.bu.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssrln_bu_h(v16u16 _1, v16u16 _2) { return __lasx_xvssrln_bu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrln_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrln.hu.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssrln_hu_w(v8u32 _1, v8u32 _2) { return __lasx_xvssrln_hu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrln_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrln.wu.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssrln_wu_d(v4u64 _1, v4u64 _2) { return __lasx_xvssrln_wu_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrlrn_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrlrn.b.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrlrn_b_h(v16i16 _1, v16i16 _2) { return __lasx_xvsrlrn_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrlrn_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrlrn.h.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrlrn_h_w(v8i32 _1, v8i32 _2) { return __lasx_xvsrlrn_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrlrn_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrlrn.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrlrn_w_d(v4i64 _1, v4i64 _2) { return __lasx_xvsrlrn_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrlrn_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrlrn.bu.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssrlrn_bu_h(v16u16 _1, v16u16 _2) { return __lasx_xvssrlrn_bu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrlrn_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrlrn.hu.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssrlrn_hu_w(v8u32 _1, v8u32 _2) { return __lasx_xvssrlrn_hu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrlrn_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrlrn.wu.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssrlrn_wu_d(v4u64 _1, v4u64 _2) { return __lasx_xvssrlrn_wu_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfrstpi_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvfrstpi.b(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvfrstpi_b(v32i8 _1, v32i8 _2) { return __lasx_xvfrstpi_b(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvfrstpi_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvfrstpi.h(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvfrstpi_h(v16i16 _1, v16i16 _2) { return __lasx_xvfrstpi_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvfrstp_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvfrstp.b(<32 x i8> [[_136]], <32 x i8> [[_247]], <32 x i8> [[_358]]) -// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvfrstp_b(v32i8 _1, v32i8 _2, v32i8 _3) { return __lasx_xvfrstp_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfrstp_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvfrstp.h(<16 x i16> [[_136]], <16 x i16> [[_247]], <16 x i16> [[_358]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvfrstp_h(v16i16 _1, v16i16 _2, v16i16 _3) { return __lasx_xvfrstp_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvshuf4i_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvshuf4i.d(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvshuf4i_d(v4i64 _1, v4i64 _2) { return __lasx_xvshuf4i_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvbsrl_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbsrl.v(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvbsrl_v(v32i8 _1) { return __lasx_xvbsrl_v(_1, 1); } // CHECK-LABEL: define dso_local void @xvbsll_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbsll.v(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvbsll_v(v32i8 _1) { return __lasx_xvbsll_v(_1, 1); } // CHECK-LABEL: define dso_local void @xvextrins_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvextrins.b(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvextrins_b(v32i8 _1, v32i8 _2) { return __lasx_xvextrins_b(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvextrins_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvextrins.h(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvextrins_h(v16i16 _1, v16i16 _2) { return __lasx_xvextrins_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvextrins_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvextrins.w(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvextrins_w(v8i32 _1, v8i32 _2) { return __lasx_xvextrins_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvextrins_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvextrins.d(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvextrins_d(v4i64 _1, v4i64 _2) { return __lasx_xvextrins_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvmskltz_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmskltz.b(<32 x i8> [[_112]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmskltz_b(v32i8 _1) { return __lasx_xvmskltz_b(_1); } // CHECK-LABEL: define dso_local void @xvmskltz_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmskltz.h(<16 x i16> [[_112]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmskltz_h(v16i16 _1) { return __lasx_xvmskltz_h(_1); } // CHECK-LABEL: define dso_local void @xvmskltz_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmskltz.w(<8 x i32> [[_112]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmskltz_w(v8i32 _1) { return __lasx_xvmskltz_w(_1); } // CHECK-LABEL: define dso_local void @xvmskltz_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmskltz.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmskltz_d(v4i64 _1) { return __lasx_xvmskltz_d(_1); } // CHECK-LABEL: define dso_local void @xvsigncov_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsigncov.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsigncov_b(v32i8 _1, v32i8 _2) { return __lasx_xvsigncov_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsigncov_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsigncov.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsigncov_h(v16i16 _1, v16i16 _2) { return __lasx_xvsigncov_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsigncov_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsigncov.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsigncov_w(v8i32 _1, v8i32 _2) { return __lasx_xvsigncov_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsigncov_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsigncov.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsigncov_d(v4i64 _1, v4i64 _2) { return __lasx_xvsigncov_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmadd_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfmadd.s(<8 x float> [[_1]], <8 x float> [[_2]], <8 x float> [[_3]]) -// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfmadd_s(v8f32 _1, v8f32 _2, v8f32 _3) { return __lasx_xvfmadd_s(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfmadd_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfmadd.d(<4 x double> [[_1]], <4 x double> [[_2]], <4 x double> [[_3]]) -// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfmadd_d(v4f64 _1, v4f64 _2, v4f64 _3) { return __lasx_xvfmadd_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfmsub_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfmsub.s(<8 x float> [[_1]], <8 x float> [[_2]], <8 x float> [[_3]]) -// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfmsub_s(v8f32 _1, v8f32 _2, v8f32 _3) { return __lasx_xvfmsub_s(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfmsub_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfmsub.d(<4 x double> [[_1]], <4 x double> [[_2]], <4 x double> [[_3]]) -// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfmsub_d(v4f64 _1, v4f64 _2, v4f64 _3) { return __lasx_xvfmsub_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfnmadd_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfnmadd.s(<8 x float> [[_1]], <8 x float> [[_2]], <8 x float> [[_3]]) -// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfnmadd_s(v8f32 _1, v8f32 _2, v8f32 _3) { return __lasx_xvfnmadd_s(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfnmadd_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfnmadd.d(<4 x double> [[_1]], <4 x double> [[_2]], <4 x double> [[_3]]) -// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfnmadd_d(v4f64 _1, v4f64 _2, v4f64 _3) { return __lasx_xvfnmadd_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfnmsub_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfnmsub.s(<8 x float> [[_1]], <8 x float> [[_2]], <8 x float> [[_3]]) -// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfnmsub_s(v8f32 _1, v8f32 _2, v8f32 _3) { return __lasx_xvfnmsub_s(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfnmsub_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfnmsub.d(<4 x double> [[_1]], <4 x double> [[_2]], <4 x double> [[_3]]) -// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfnmsub_d(v4f64 _1, v4f64 _2, v4f64 _3) { return __lasx_xvfnmsub_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvftintrne_w_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrne.w.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrne_w_s(v8f32 _1) { return __lasx_xvftintrne_w_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrne_l_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrne.l.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrne_l_d(v4f64 _1) { return __lasx_xvftintrne_l_d(_1); } // CHECK-LABEL: define dso_local void @xvftintrp_w_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrp.w.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrp_w_s(v8f32 _1) { return __lasx_xvftintrp_w_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrp_l_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrp.l.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrp_l_d(v4f64 _1) { return __lasx_xvftintrp_l_d(_1); } // CHECK-LABEL: define dso_local void @xvftintrm_w_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrm.w.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrm_w_s(v8f32 _1) { return __lasx_xvftintrm_w_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrm_l_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrm.l.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrm_l_d(v4f64 _1) { return __lasx_xvftintrm_l_d(_1); } // CHECK-LABEL: define dso_local void @xvftint_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftint.w.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftint_w_d(v4f64 _1, v4f64 _2) { return __lasx_xvftint_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvffint_s_l( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvffint.s.l(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvffint_s_l(v4i64 _1, v4i64 _2) { return __lasx_xvffint_s_l(_1, _2); } // CHECK-LABEL: define dso_local void @xvftintrz_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrz.w.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrz_w_d(v4f64 _1, v4f64 _2) { return __lasx_xvftintrz_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvftintrp_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrp.w.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrp_w_d(v4f64 _1, v4f64 _2) { return __lasx_xvftintrp_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvftintrm_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrm.w.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrm_w_d(v4f64 _1, v4f64 _2) { return __lasx_xvftintrm_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvftintrne_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrne.w.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrne_w_d(v4f64 _1, v4f64 _2) { return __lasx_xvftintrne_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvftinth_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftinth.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftinth_l_s(v8f32 _1) { return __lasx_xvftinth_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintl_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintl.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintl_l_s(v8f32 _1) { return __lasx_xvftintl_l_s(_1); } // CHECK-LABEL: define dso_local void @xvffinth_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvffinth.d.w(<8 x i32> [[_112]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvffinth_d_w(v8i32 _1) { return __lasx_xvffinth_d_w(_1); } // CHECK-LABEL: define dso_local void @xvffintl_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvffintl.d.w(<8 x i32> [[_112]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvffintl_d_w(v8i32 _1) { return __lasx_xvffintl_d_w(_1); } // CHECK-LABEL: define dso_local void @xvftintrzh_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrzh.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrzh_l_s(v8f32 _1) { return __lasx_xvftintrzh_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrzl_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrzl.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrzl_l_s(v8f32 _1) { return __lasx_xvftintrzl_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrph_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrph.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrph_l_s(v8f32 _1) { return __lasx_xvftintrph_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrpl_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrpl.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrpl_l_s(v8f32 _1) { return __lasx_xvftintrpl_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrmh_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrmh.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrmh_l_s(v8f32 _1) { return __lasx_xvftintrmh_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrml_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrml.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrml_l_s(v8f32 _1) { return __lasx_xvftintrml_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrneh_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrneh.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrneh_l_s(v8f32 _1) { return __lasx_xvftintrneh_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrnel_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrnel.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrnel_l_s(v8f32 _1) { return __lasx_xvftintrnel_l_s(_1); } // CHECK-LABEL: define dso_local void @xvfrintrne_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrintrne.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfrintrne_s(v8f32 _1) { return __lasx_xvfrintrne_s(_1); } // CHECK-LABEL: define dso_local void @xvfrintrne_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrintrne.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfrintrne_d(v4f64 _1) { return __lasx_xvfrintrne_d(_1); } // CHECK-LABEL: define dso_local void @xvfrintrz_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrintrz.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfrintrz_s(v8f32 _1) { return __lasx_xvfrintrz_s(_1); } // CHECK-LABEL: define dso_local void @xvfrintrz_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrintrz.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfrintrz_d(v4f64 _1) { return __lasx_xvfrintrz_d(_1); } // CHECK-LABEL: define dso_local void @xvfrintrp_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrintrp.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfrintrp_s(v8f32 _1) { return __lasx_xvfrintrp_s(_1); } // CHECK-LABEL: define dso_local void @xvfrintrp_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrintrp.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfrintrp_d(v4f64 _1) { return __lasx_xvfrintrp_d(_1); } // CHECK-LABEL: define dso_local void @xvfrintrm_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrintrm.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfrintrm_s(v8f32 _1) { return __lasx_xvfrintrm_s(_1); } // CHECK-LABEL: define dso_local void @xvfrintrm_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrintrm.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfrintrm_d(v4f64 _1) { return __lasx_xvfrintrm_d(_1); } @@ -4482,14 +4482,14 @@ v4i64 xvfrintrm_d(v4f64 _1) { return __lasx_xvfrintrm_d(_1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr noundef [[_1:%.*]]) local_unnamed_addr #[[ATTR3:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvld(ptr [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvld(void * _1) { return __lasx_xvld(_1, 1); } // CHECK-LABEL: define dso_local void @xvst( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr noundef [[_2:%.*]]) local_unnamed_addr #[[ATTR5:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: tail call void @llvm.loongarch.lasx.xvst(<32 x i8> [[_1]], ptr [[_2]], i32 1) // CHECK-NEXT: ret void // @@ -4497,7 +4497,7 @@ void xvst(v32i8 _1, void * _2) { return __lasx_xvst(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvstelm_b( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr noundef [[_2:%.*]]) local_unnamed_addr #[[ATTR5]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: tail call void @llvm.loongarch.lasx.xvstelm.b(<32 x i8> [[_1]], ptr [[_2]], i32 1, i32 1) // CHECK-NEXT: ret void // @@ -4505,7 +4505,7 @@ void xvstelm_b(v32i8 _1, void * _2) { return __lasx_xvstelm_b(_1, _2, 1, 1); } // CHECK-LABEL: define dso_local void @xvstelm_h( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr noundef [[_2:%.*]]) local_unnamed_addr #[[ATTR5]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: tail call void @llvm.loongarch.lasx.xvstelm.h(<16 x i16> [[_1]], ptr [[_2]], i32 2, i32 1) // CHECK-NEXT: ret void // @@ -4513,7 +4513,7 @@ void xvstelm_h(v16i16 _1, void * _2) { return __lasx_xvstelm_h(_1, _2, 2, 1); } // CHECK-LABEL: define dso_local void @xvstelm_w( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr noundef [[_2:%.*]]) local_unnamed_addr #[[ATTR5]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: tail call void @llvm.loongarch.lasx.xvstelm.w(<8 x i32> [[_1]], ptr [[_2]], i32 4, i32 1) // CHECK-NEXT: ret void // @@ -4521,7 +4521,7 @@ void xvstelm_w(v8i32 _1, void * _2) { return __lasx_xvstelm_w(_1, _2, 4, 1); } // CHECK-LABEL: define dso_local void @xvstelm_d( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr noundef [[_2:%.*]]) local_unnamed_addr #[[ATTR5]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: tail call void @llvm.loongarch.lasx.xvstelm.d(<4 x i64> [[_1]], ptr [[_2]], i32 8, i32 1) // CHECK-NEXT: ret void // @@ -4529,108 +4529,108 @@ void xvstelm_d(v4i64 _1, void * _2) { return __lasx_xvstelm_d(_1, _2, 8, 1); } // CHECK-LABEL: define dso_local void @xvinsve0_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvinsve0.w(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvinsve0_w(v8i32 _1, v8i32 _2) { return __lasx_xvinsve0_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvinsve0_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvinsve0.d(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvinsve0_d(v4i64 _1, v4i64 _2) { return __lasx_xvinsve0_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvpickve_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvpickve.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvpickve_w(v8i32 _1) { return __lasx_xvpickve_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvpickve_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvpickve.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvpickve_d(v4i64 _1) { return __lasx_xvpickve_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvssrlrn_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrlrn.b.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssrlrn_b_h(v16i16 _1, v16i16 _2) { return __lasx_xvssrlrn_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrlrn_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrlrn.h.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssrlrn_h_w(v8i32 _1, v8i32 _2) { return __lasx_xvssrlrn_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrlrn_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrlrn.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssrlrn_w_d(v4i64 _1, v4i64 _2) { return __lasx_xvssrlrn_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrln_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrln.b.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssrln_b_h(v16i16 _1, v16i16 _2) { return __lasx_xvssrln_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrln_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrln.h.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssrln_h_w(v8i32 _1, v8i32 _2) { return __lasx_xvssrln_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrln_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrln.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssrln_w_d(v4i64 _1, v4i64 _2) { return __lasx_xvssrln_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvorn_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvorn.v(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvorn_v(v32i8 _1, v32i8 _2) { return __lasx_xvorn_v(_1, _2); } @@ -4638,22 +4638,22 @@ v32i8 xvorn_v(v32i8 _1, v32i8 _2) { return __lasx_xvorn_v(_1, _2); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvldi(i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvldi() { return __lasx_xvldi(1); } // CHECK-LABEL: define dso_local void @xvldx( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr noundef [[_1:%.*]]) local_unnamed_addr #[[ATTR3]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvldx(ptr [[_1]], i64 1), !noalias [[META5:![0-9]+]] -// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvldx(ptr [[_1]], i64 1), !noalias [[META7:![0-9]+]] +// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvldx(void * _1) { return __lasx_xvldx(_1, 1); } // CHECK-LABEL: define dso_local void @xvstx( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr noundef [[_2:%.*]]) local_unnamed_addr #[[ATTR5]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: tail call void @llvm.loongarch.lasx.xvstx(<32 x i8> [[_112]], ptr [[_2]], i64 1) // CHECK-NEXT: ret void // @@ -4661,209 +4661,209 @@ void xvstx(v32i8 _1, void * _2) { return __lasx_xvstx(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvextl_qu_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvextl.qu.du(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvextl_qu_du(v4u64 _1) { return __lasx_xvextl_qu_du(_1); } // CHECK-LABEL: define dso_local void @xvinsgr2vr_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvinsgr2vr.w(<8 x i32> [[_1]], i32 1, i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvinsgr2vr_w(v8i32 _1) { return __lasx_xvinsgr2vr_w(_1, 1, 1); } // CHECK-LABEL: define dso_local void @xvinsgr2vr_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvinsgr2vr.d(<4 x i64> [[_1]], i64 1, i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvinsgr2vr_d(v4i64 _1) { return __lasx_xvinsgr2vr_d(_1, 1, 1); } // CHECK-LABEL: define dso_local void @xvreplve0_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvreplve0.b(<32 x i8> [[_112]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvreplve0_b(v32i8 _1) { return __lasx_xvreplve0_b(_1); } // CHECK-LABEL: define dso_local void @xvreplve0_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvreplve0.h(<16 x i16> [[_112]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvreplve0_h(v16i16 _1) { return __lasx_xvreplve0_h(_1); } // CHECK-LABEL: define dso_local void @xvreplve0_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvreplve0.w(<8 x i32> [[_112]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvreplve0_w(v8i32 _1) { return __lasx_xvreplve0_w(_1); } // CHECK-LABEL: define dso_local void @xvreplve0_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvreplve0.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvreplve0_d(v4i64 _1) { return __lasx_xvreplve0_d(_1); } // CHECK-LABEL: define dso_local void @xvreplve0_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvreplve0.q(<32 x i8> [[_112]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvreplve0_q(v32i8 _1) { return __lasx_xvreplve0_q(_1); } // CHECK-LABEL: define dso_local void @vext2xv_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.vext2xv.h.b(<32 x i8> [[_112]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 vext2xv_h_b(v32i8 _1) { return __lasx_vext2xv_h_b(_1); } // CHECK-LABEL: define dso_local void @vext2xv_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.vext2xv.w.h(<16 x i16> [[_112]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 vext2xv_w_h(v16i16 _1) { return __lasx_vext2xv_w_h(_1); } // CHECK-LABEL: define dso_local void @vext2xv_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.vext2xv.d.w(<8 x i32> [[_112]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 vext2xv_d_w(v8i32 _1) { return __lasx_vext2xv_d_w(_1); } // CHECK-LABEL: define dso_local void @vext2xv_w_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.vext2xv.w.b(<32 x i8> [[_112]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 vext2xv_w_b(v32i8 _1) { return __lasx_vext2xv_w_b(_1); } // CHECK-LABEL: define dso_local void @vext2xv_d_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.vext2xv.d.h(<16 x i16> [[_112]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 vext2xv_d_h(v16i16 _1) { return __lasx_vext2xv_d_h(_1); } // CHECK-LABEL: define dso_local void @vext2xv_d_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.vext2xv.d.b(<32 x i8> [[_112]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 vext2xv_d_b(v32i8 _1) { return __lasx_vext2xv_d_b(_1); } // CHECK-LABEL: define dso_local void @vext2xv_hu_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.vext2xv.hu.bu(<32 x i8> [[_112]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 vext2xv_hu_bu(v32i8 _1) { return __lasx_vext2xv_hu_bu(_1); } // CHECK-LABEL: define dso_local void @vext2xv_wu_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.vext2xv.wu.hu(<16 x i16> [[_112]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 vext2xv_wu_hu(v16i16 _1) { return __lasx_vext2xv_wu_hu(_1); } // CHECK-LABEL: define dso_local void @vext2xv_du_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.vext2xv.du.wu(<8 x i32> [[_112]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 vext2xv_du_wu(v8i32 _1) { return __lasx_vext2xv_du_wu(_1); } // CHECK-LABEL: define dso_local void @vext2xv_wu_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.vext2xv.wu.bu(<32 x i8> [[_112]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 vext2xv_wu_bu(v32i8 _1) { return __lasx_vext2xv_wu_bu(_1); } // CHECK-LABEL: define dso_local void @vext2xv_du_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.vext2xv.du.hu(<16 x i16> [[_112]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 vext2xv_du_hu(v16i16 _1) { return __lasx_vext2xv_du_hu(_1); } // CHECK-LABEL: define dso_local void @vext2xv_du_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.vext2xv.du.bu(<32 x i8> [[_112]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 vext2xv_du_bu(v32i8 _1) { return __lasx_vext2xv_du_bu(_1); } // CHECK-LABEL: define dso_local void @xvpermi_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvpermi.q(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvpermi_q(v32i8 _1, v32i8 _2) { return __lasx_xvpermi_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvpermi_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvpermi.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvpermi_d(v4i64 _1) { return __lasx_xvpermi_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvperm_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvperm.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvperm_w(v8i32 _1, v8i32 _2) { return __lasx_xvperm_w(_1, _2); } @@ -4871,7 +4871,7 @@ v8i32 xvperm_w(v8i32 _1, v8i32 _2) { return __lasx_xvperm_w(_1, _2); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr noundef [[_1:%.*]]) local_unnamed_addr #[[ATTR3]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvldrepl.b(ptr [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvldrepl_b(void * _1) { return __lasx_xvldrepl_b(_1, 1); } @@ -4879,7 +4879,7 @@ v32i8 xvldrepl_b(void * _1) { return __lasx_xvldrepl_b(_1, 1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr noundef [[_1:%.*]]) local_unnamed_addr #[[ATTR3]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvldrepl.h(ptr [[_1]], i32 2) -// CHECK-NEXT: store <16 x i16> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvldrepl_h(void * _1) { return __lasx_xvldrepl_h(_1, 2); } @@ -4887,7 +4887,7 @@ v16i16 xvldrepl_h(void * _1) { return __lasx_xvldrepl_h(_1, 2); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr noundef [[_1:%.*]]) local_unnamed_addr #[[ATTR3]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvldrepl.w(ptr [[_1]], i32 4) -// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvldrepl_w(void * _1) { return __lasx_xvldrepl_w(_1, 4); } @@ -4895,14 +4895,14 @@ v8i32 xvldrepl_w(void * _1) { return __lasx_xvldrepl_w(_1, 4); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr noundef [[_1:%.*]]) local_unnamed_addr #[[ATTR3]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvldrepl.d(ptr [[_1]], i32 8) -// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvldrepl_d(void * _1) { return __lasx_xvldrepl_d(_1, 8); } // CHECK-LABEL: define dso_local signext i32 @xvpickve2gr_w( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xvpickve2gr.w(<8 x i32> [[_1]], i32 1) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -4910,7 +4910,7 @@ int xvpickve2gr_w(v8i32 _1) { return __lasx_xvpickve2gr_w(_1, 1); } // CHECK-LABEL: define dso_local signext i32 @xvpickve2gr_wu( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xvpickve2gr.wu(<8 x i32> [[_1]], i32 1) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -4918,7 +4918,7 @@ unsigned int xvpickve2gr_wu(v8i32 _1) { return __lasx_xvpickve2gr_wu(_1, 1); } // CHECK-LABEL: define dso_local i64 @xvpickve2gr_d( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.loongarch.lasx.xvpickve2gr.d(<4 x i64> [[_1]], i32 1) // CHECK-NEXT: ret i64 [[TMP1]] // @@ -4926,7 +4926,7 @@ long xvpickve2gr_d(v4i64 _1) { return __lasx_xvpickve2gr_d(_1, 1); } // CHECK-LABEL: define dso_local i64 @xvpickve2gr_du( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.loongarch.lasx.xvpickve2gr.du(<4 x i64> [[_1]], i32 1) // CHECK-NEXT: ret i64 [[TMP1]] // @@ -4934,1626 +4934,1626 @@ unsigned long int xvpickve2gr_du(v4i64 _1) { return __lasx_xvpickve2gr_du(_1, 1) // CHECK-LABEL: define dso_local void @xvaddwev_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwev.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwev_q_d(v4i64 _1, v4i64 _2) { return __lasx_xvaddwev_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwev.d.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwev_d_w(v8i32 _1, v8i32 _2) { return __lasx_xvaddwev_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvaddwev.w.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvaddwev_w_h(v16i16 _1, v16i16 _2) { return __lasx_xvaddwev_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvaddwev.h.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvaddwev_h_b(v32i8 _1, v32i8 _2) { return __lasx_xvaddwev_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwev.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwev_q_du(v4u64 _1, v4u64 _2) { return __lasx_xvaddwev_q_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwev.d.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwev_d_wu(v8u32 _1, v8u32 _2) { return __lasx_xvaddwev_d_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvaddwev.w.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvaddwev_w_hu(v16u16 _1, v16u16 _2) { return __lasx_xvaddwev_w_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvaddwev.h.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvaddwev_h_bu(v32u8 _1, v32u8 _2) { return __lasx_xvaddwev_h_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwev.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwev_q_d(v4i64 _1, v4i64 _2) { return __lasx_xvsubwev_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwev.d.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwev_d_w(v8i32 _1, v8i32 _2) { return __lasx_xvsubwev_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsubwev.w.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsubwev_w_h(v16i16 _1, v16i16 _2) { return __lasx_xvsubwev_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsubwev.h.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsubwev_h_b(v32i8 _1, v32i8 _2) { return __lasx_xvsubwev_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwev.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwev_q_du(v4u64 _1, v4u64 _2) { return __lasx_xvsubwev_q_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwev.d.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwev_d_wu(v8u32 _1, v8u32 _2) { return __lasx_xvsubwev_d_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsubwev.w.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsubwev_w_hu(v16u16 _1, v16u16 _2) { return __lasx_xvsubwev_w_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsubwev.h.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsubwev_h_bu(v32u8 _1, v32u8 _2) { return __lasx_xvsubwev_h_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwev.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwev_q_d(v4i64 _1, v4i64 _2) { return __lasx_xvmulwev_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwev.d.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwev_d_w(v8i32 _1, v8i32 _2) { return __lasx_xvmulwev_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmulwev.w.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmulwev_w_h(v16i16 _1, v16i16 _2) { return __lasx_xvmulwev_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmulwev.h.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmulwev_h_b(v32i8 _1, v32i8 _2) { return __lasx_xvmulwev_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwev.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwev_q_du(v4u64 _1, v4u64 _2) { return __lasx_xvmulwev_q_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwev.d.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwev_d_wu(v8u32 _1, v8u32 _2) { return __lasx_xvmulwev_d_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmulwev.w.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmulwev_w_hu(v16u16 _1, v16u16 _2) { return __lasx_xvmulwev_w_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmulwev.h.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmulwev_h_bu(v32u8 _1, v32u8 _2) { return __lasx_xvmulwev_h_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwod.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwod_q_d(v4i64 _1, v4i64 _2) { return __lasx_xvaddwod_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwod.d.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwod_d_w(v8i32 _1, v8i32 _2) { return __lasx_xvaddwod_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvaddwod.w.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvaddwod_w_h(v16i16 _1, v16i16 _2) { return __lasx_xvaddwod_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvaddwod.h.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvaddwod_h_b(v32i8 _1, v32i8 _2) { return __lasx_xvaddwod_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwod.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwod_q_du(v4u64 _1, v4u64 _2) { return __lasx_xvaddwod_q_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwod.d.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwod_d_wu(v8u32 _1, v8u32 _2) { return __lasx_xvaddwod_d_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvaddwod.w.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvaddwod_w_hu(v16u16 _1, v16u16 _2) { return __lasx_xvaddwod_w_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvaddwod.h.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvaddwod_h_bu(v32u8 _1, v32u8 _2) { return __lasx_xvaddwod_h_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwod.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwod_q_d(v4i64 _1, v4i64 _2) { return __lasx_xvsubwod_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwod.d.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwod_d_w(v8i32 _1, v8i32 _2) { return __lasx_xvsubwod_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsubwod.w.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsubwod_w_h(v16i16 _1, v16i16 _2) { return __lasx_xvsubwod_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsubwod.h.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsubwod_h_b(v32i8 _1, v32i8 _2) { return __lasx_xvsubwod_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwod.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwod_q_du(v4u64 _1, v4u64 _2) { return __lasx_xvsubwod_q_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwod.d.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwod_d_wu(v8u32 _1, v8u32 _2) { return __lasx_xvsubwod_d_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsubwod.w.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsubwod_w_hu(v16u16 _1, v16u16 _2) { return __lasx_xvsubwod_w_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsubwod.h.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsubwod_h_bu(v32u8 _1, v32u8 _2) { return __lasx_xvsubwod_h_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwod.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwod_q_d(v4i64 _1, v4i64 _2) { return __lasx_xvmulwod_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwod.d.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwod_d_w(v8i32 _1, v8i32 _2) { return __lasx_xvmulwod_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmulwod.w.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmulwod_w_h(v16i16 _1, v16i16 _2) { return __lasx_xvmulwod_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmulwod.h.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmulwod_h_b(v32i8 _1, v32i8 _2) { return __lasx_xvmulwod_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwod.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwod_q_du(v4u64 _1, v4u64 _2) { return __lasx_xvmulwod_q_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwod.d.wu(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwod_d_wu(v8u32 _1, v8u32 _2) { return __lasx_xvmulwod_d_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmulwod.w.hu(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmulwod_w_hu(v16u16 _1, v16u16 _2) { return __lasx_xvmulwod_w_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmulwod.h.bu(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmulwod_h_bu(v32u8 _1, v32u8 _2) { return __lasx_xvmulwod_h_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_d_wu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwev.d.wu.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwev_d_wu_w(v8u32 _1, v8i32 _2) { return __lasx_xvaddwev_d_wu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_w_hu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvaddwev.w.hu.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvaddwev_w_hu_h(v16u16 _1, v16i16 _2) { return __lasx_xvaddwev_w_hu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_h_bu_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvaddwev.h.bu.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvaddwev_h_bu_b(v32u8 _1, v32i8 _2) { return __lasx_xvaddwev_h_bu_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_d_wu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwev.d.wu.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwev_d_wu_w(v8u32 _1, v8i32 _2) { return __lasx_xvmulwev_d_wu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_w_hu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmulwev.w.hu.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmulwev_w_hu_h(v16u16 _1, v16i16 _2) { return __lasx_xvmulwev_w_hu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_h_bu_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmulwev.h.bu.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmulwev_h_bu_b(v32u8 _1, v32i8 _2) { return __lasx_xvmulwev_h_bu_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_d_wu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwod.d.wu.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwod_d_wu_w(v8u32 _1, v8i32 _2) { return __lasx_xvaddwod_d_wu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_w_hu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvaddwod.w.hu.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvaddwod_w_hu_h(v16u16 _1, v16i16 _2) { return __lasx_xvaddwod_w_hu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_h_bu_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvaddwod.h.bu.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvaddwod_h_bu_b(v32u8 _1, v32i8 _2) { return __lasx_xvaddwod_h_bu_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_d_wu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwod.d.wu.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwod_d_wu_w(v8u32 _1, v8i32 _2) { return __lasx_xvmulwod_d_wu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_w_hu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmulwod.w.hu.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmulwod_w_hu_h(v16u16 _1, v16i16 _2) { return __lasx_xvmulwod_w_hu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_h_bu_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmulwod.h.bu.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmulwod_h_bu_b(v32u8 _1, v32i8 _2) { return __lasx_xvmulwod_h_bu_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhaddw.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvhaddw_q_d(v4i64 _1, v4i64 _2) { return __lasx_xvhaddw_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_qu_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhaddw.qu.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvhaddw_qu_du(v4u64 _1, v4u64 _2) { return __lasx_xvhaddw_qu_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhsubw.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvhsubw_q_d(v4i64 _1, v4i64 _2) { return __lasx_xvhsubw_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_qu_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhsubw.qu.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvhsubw_qu_du(v4u64 _1, v4u64 _2) { return __lasx_xvhsubw_qu_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvmaddwev_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwev.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwev_q_d(v4i64 _1, v4i64 _2, v4i64 _3) { return __lasx_xvmaddwev_q_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_346:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_346:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwev.d.w(<4 x i64> [[_1]], <8 x i32> [[_235]], <8 x i32> [[_346]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwev_d_w(v4i64 _1, v8i32 _2, v8i32 _3) { return __lasx_xvmaddwev_d_w(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaddwev.w.h(<8 x i32> [[_136]], <16 x i16> [[_247]], <16 x i16> [[_358]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmaddwev_w_h(v8i32 _1, v16i16 _2, v16i16 _3) { return __lasx_xvmaddwev_w_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaddwev.h.b(<16 x i16> [[_136]], <32 x i8> [[_247]], <32 x i8> [[_358]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmaddwev_h_b(v16i16 _1, v32i8 _2, v32i8 _3) { return __lasx_xvmaddwev_h_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwev.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmaddwev_q_du(v4u64 _1, v4u64 _2, v4u64 _3) { return __lasx_xvmaddwev_q_du(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_346:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_346:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwev.d.wu(<4 x i64> [[_1]], <8 x i32> [[_235]], <8 x i32> [[_346]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmaddwev_d_wu(v4u64 _1, v8u32 _2, v8u32 _3) { return __lasx_xvmaddwev_d_wu(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaddwev.w.hu(<8 x i32> [[_136]], <16 x i16> [[_247]], <16 x i16> [[_358]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmaddwev_w_hu(v8u32 _1, v16u16 _2, v16u16 _3) { return __lasx_xvmaddwev_w_hu(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaddwev.h.bu(<16 x i16> [[_136]], <32 x i8> [[_247]], <32 x i8> [[_358]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmaddwev_h_bu(v16u16 _1, v32u8 _2, v32u8 _3) { return __lasx_xvmaddwev_h_bu(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwod.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwod_q_d(v4i64 _1, v4i64 _2, v4i64 _3) { return __lasx_xvmaddwod_q_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_346:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_346:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwod.d.w(<4 x i64> [[_1]], <8 x i32> [[_235]], <8 x i32> [[_346]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwod_d_w(v4i64 _1, v8i32 _2, v8i32 _3) { return __lasx_xvmaddwod_d_w(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaddwod.w.h(<8 x i32> [[_136]], <16 x i16> [[_247]], <16 x i16> [[_358]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmaddwod_w_h(v8i32 _1, v16i16 _2, v16i16 _3) { return __lasx_xvmaddwod_w_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaddwod.h.b(<16 x i16> [[_136]], <32 x i8> [[_247]], <32 x i8> [[_358]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmaddwod_h_b(v16i16 _1, v32i8 _2, v32i8 _3) { return __lasx_xvmaddwod_h_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwod.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmaddwod_q_du(v4u64 _1, v4u64 _2, v4u64 _3) { return __lasx_xvmaddwod_q_du(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_346:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_346:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwod.d.wu(<4 x i64> [[_1]], <8 x i32> [[_235]], <8 x i32> [[_346]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmaddwod_d_wu(v4u64 _1, v8u32 _2, v8u32 _3) { return __lasx_xvmaddwod_d_wu(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaddwod.w.hu(<8 x i32> [[_136]], <16 x i16> [[_247]], <16 x i16> [[_358]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmaddwod_w_hu(v8u32 _1, v16u16 _2, v16u16 _3) { return __lasx_xvmaddwod_w_hu(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaddwod.h.bu(<16 x i16> [[_136]], <32 x i8> [[_247]], <32 x i8> [[_358]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmaddwod_h_bu(v16u16 _1, v32u8 _2, v32u8 _3) { return __lasx_xvmaddwod_h_bu(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_q_du_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwev.q.du.d(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwev_q_du_d(v4i64 _1, v4u64 _2, v4i64 _3) { return __lasx_xvmaddwev_q_du_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_d_wu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_346:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_346:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwev.d.wu.w(<4 x i64> [[_1]], <8 x i32> [[_235]], <8 x i32> [[_346]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwev_d_wu_w(v4i64 _1, v8u32 _2, v8i32 _3) { return __lasx_xvmaddwev_d_wu_w(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_w_hu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaddwev.w.hu.h(<8 x i32> [[_136]], <16 x i16> [[_247]], <16 x i16> [[_358]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmaddwev_w_hu_h(v8i32 _1, v16u16 _2, v16i16 _3) { return __lasx_xvmaddwev_w_hu_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_h_bu_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaddwev.h.bu.b(<16 x i16> [[_136]], <32 x i8> [[_247]], <32 x i8> [[_358]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmaddwev_h_bu_b(v16i16 _1, v32u8 _2, v32i8 _3) { return __lasx_xvmaddwev_h_bu_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_q_du_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwod.q.du.d(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwod_q_du_d(v4i64 _1, v4u64 _2, v4i64 _3) { return __lasx_xvmaddwod_q_du_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_d_wu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_346:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_346:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwod.d.wu.w(<4 x i64> [[_1]], <8 x i32> [[_235]], <8 x i32> [[_346]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwod_d_wu_w(v4i64 _1, v8u32 _2, v8i32 _3) { return __lasx_xvmaddwod_d_wu_w(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_w_hu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaddwod.w.hu.h(<8 x i32> [[_136]], <16 x i16> [[_247]], <16 x i16> [[_358]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmaddwod_w_hu_h(v8i32 _1, v16u16 _2, v16i16 _3) { return __lasx_xvmaddwod_w_hu_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_h_bu_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_136:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_247:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_358:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaddwod.h.bu.b(<16 x i16> [[_136]], <32 x i8> [[_247]], <32 x i8> [[_358]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmaddwod_h_bu_b(v16i16 _1, v32u8 _2, v32i8 _3) { return __lasx_xvmaddwod_h_bu_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvrotr_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvrotr.b(<32 x i8> [[_124]], <32 x i8> [[_235]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvrotr_b(v32i8 _1, v32i8 _2) { return __lasx_xvrotr_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvrotr_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvrotr.h(<16 x i16> [[_124]], <16 x i16> [[_235]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvrotr_h(v16i16 _1, v16i16 _2) { return __lasx_xvrotr_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvrotr_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_124:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_235:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvrotr.w(<8 x i32> [[_124]], <8 x i32> [[_235]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvrotr_w(v8i32 _1, v8i32 _2) { return __lasx_xvrotr_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvrotr_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvrotr.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvrotr_d(v4i64 _1, v4i64 _2) { return __lasx_xvrotr_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvadd_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvadd.q(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvadd_q(v4i64 _1, v4i64 _2) { return __lasx_xvadd_q(_1, _2); } // CHECK-LABEL: define dso_local void @xvsub_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsub.q(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsub_q(v4i64 _1, v4i64 _2) { return __lasx_xvsub_q(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_q_du_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwev.q.du.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwev_q_du_d(v4u64 _1, v4i64 _2) { return __lasx_xvaddwev_q_du_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_q_du_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwod.q.du.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwod_q_du_d(v4u64 _1, v4i64 _2) { return __lasx_xvaddwod_q_du_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_q_du_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwev.q.du.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwev_q_du_d(v4u64 _1, v4i64 _2) { return __lasx_xvmulwev_q_du_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_q_du_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwod.q.du.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwod_q_du_d(v4u64 _1, v4i64 _2) { return __lasx_xvmulwod_q_du_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmskgez_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmskgez.b(<32 x i8> [[_112]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmskgez_b(v32i8 _1) { return __lasx_xvmskgez_b(_1); } // CHECK-LABEL: define dso_local void @xvmsknz_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmsknz.b(<32 x i8> [[_112]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmsknz_b(v32i8 _1) { return __lasx_xvmsknz_b(_1); } // CHECK-LABEL: define dso_local void @xvexth_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvexth.h.b(<32 x i8> [[_112]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvexth_h_b(v32i8 _1) { return __lasx_xvexth_h_b(_1); } // CHECK-LABEL: define dso_local void @xvexth_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvexth.w.h(<16 x i16> [[_112]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvexth_w_h(v16i16 _1) { return __lasx_xvexth_w_h(_1); } // CHECK-LABEL: define dso_local void @xvexth_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvexth.d.w(<8 x i32> [[_112]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvexth_d_w(v8i32 _1) { return __lasx_xvexth_d_w(_1); } // CHECK-LABEL: define dso_local void @xvexth_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvexth.q.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvexth_q_d(v4i64 _1) { return __lasx_xvexth_q_d(_1); } // CHECK-LABEL: define dso_local void @xvexth_hu_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvexth.hu.bu(<32 x i8> [[_112]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvexth_hu_bu(v32u8 _1) { return __lasx_xvexth_hu_bu(_1); } // CHECK-LABEL: define dso_local void @xvexth_wu_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvexth.wu.hu(<16 x i16> [[_112]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvexth_wu_hu(v16u16 _1) { return __lasx_xvexth_wu_hu(_1); } // CHECK-LABEL: define dso_local void @xvexth_du_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_112:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvexth.du.wu(<8 x i32> [[_112]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvexth_du_wu(v8u32 _1) { return __lasx_xvexth_du_wu(_1); } // CHECK-LABEL: define dso_local void @xvexth_qu_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvexth.qu.du(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvexth_qu_du(v4u64 _1) { return __lasx_xvexth_qu_du(_1); } // CHECK-LABEL: define dso_local void @xvrotri_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvrotri.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvrotri_b(v32i8 _1) { return __lasx_xvrotri_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvrotri_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvrotri.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvrotri_h(v16i16 _1) { return __lasx_xvrotri_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvrotri_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvrotri.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvrotri_w(v8i32 _1) { return __lasx_xvrotri_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvrotri_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvrotri.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvrotri_d(v4i64 _1) { return __lasx_xvrotri_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvextl_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvextl.q.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvextl_q_d(v4i64 _1) { return __lasx_xvextl_q_d(_1); } // CHECK-LABEL: define dso_local void @xvsrlni_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrlni.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrlni_b_h(v32i8 _1, v32i8 _2) { return __lasx_xvsrlni_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrlni_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrlni.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrlni_h_w(v16i16 _1, v16i16 _2) { return __lasx_xvsrlni_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrlni_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrlni.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrlni_w_d(v8i32 _1, v8i32 _2) { return __lasx_xvsrlni_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrlni_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrlni.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrlni_d_q(v4i64 _1, v4i64 _2) { return __lasx_xvsrlni_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrlrni_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrlrni.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrlrni_b_h(v32i8 _1, v32i8 _2) { return __lasx_xvsrlrni_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrlrni_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrlrni.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrlrni_h_w(v16i16 _1, v16i16 _2) { return __lasx_xvsrlrni_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrlrni_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrlrni.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrlrni_w_d(v8i32 _1, v8i32 _2) { return __lasx_xvsrlrni_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrlrni_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrlrni.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrlrni_d_q(v4i64 _1, v4i64 _2) { return __lasx_xvsrlrni_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrlni.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssrlni_b_h(v32i8 _1, v32i8 _2) { return __lasx_xvssrlni_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrlni.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssrlni_h_w(v16i16 _1, v16i16 _2) { return __lasx_xvssrlni_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrlni.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssrlni_w_d(v8i32 _1, v8i32 _2) { return __lasx_xvssrlni_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrlni.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvssrlni_d_q(v4i64 _1, v4i64 _2) { return __lasx_xvssrlni_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrlni.bu.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssrlni_bu_h(v32u8 _1, v32i8 _2) { return __lasx_xvssrlni_bu_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrlni.hu.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssrlni_hu_w(v16u16 _1, v16i16 _2) { return __lasx_xvssrlni_hu_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrlni.wu.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssrlni_wu_d(v8u32 _1, v8i32 _2) { return __lasx_xvssrlni_wu_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_du_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrlni.du.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvssrlni_du_q(v4u64 _1, v4i64 _2) { return __lasx_xvssrlni_du_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrlrni.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssrlrni_b_h(v32i8 _1, v32i8 _2) { return __lasx_xvssrlrni_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrlrni.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssrlrni_h_w(v16i16 _1, v16i16 _2) { return __lasx_xvssrlrni_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrlrni.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssrlrni_w_d(v8i32 _1, v8i32 _2) { return __lasx_xvssrlrni_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrlrni.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvssrlrni_d_q(v4i64 _1, v4i64 _2) { return __lasx_xvssrlrni_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrlrni.bu.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssrlrni_bu_h(v32u8 _1, v32i8 _2) { return __lasx_xvssrlrni_bu_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrlrni.hu.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssrlrni_hu_w(v16u16 _1, v16i16 _2) { return __lasx_xvssrlrni_hu_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrlrni.wu.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssrlrni_wu_d(v8u32 _1, v8i32 _2) { return __lasx_xvssrlrni_wu_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_du_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrlrni.du.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvssrlrni_du_q(v4u64 _1, v4i64 _2) { return __lasx_xvssrlrni_du_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrani_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrani.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrani_b_h(v32i8 _1, v32i8 _2) { return __lasx_xvsrani_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrani_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrani.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrani_h_w(v16i16 _1, v16i16 _2) { return __lasx_xvsrani_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrani_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrani.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrani_w_d(v8i32 _1, v8i32 _2) { return __lasx_xvsrani_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrani_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrani.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrani_d_q(v4i64 _1, v4i64 _2) { return __lasx_xvsrani_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrarni_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrarni.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrarni_b_h(v32i8 _1, v32i8 _2) { return __lasx_xvsrarni_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrarni_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrarni.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrarni_h_w(v16i16 _1, v16i16 _2) { return __lasx_xvsrarni_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrarni_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrarni.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrarni_w_d(v8i32 _1, v8i32 _2) { return __lasx_xvsrarni_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrarni_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrarni.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrarni_d_q(v4i64 _1, v4i64 _2) { return __lasx_xvsrarni_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrani.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssrani_b_h(v32i8 _1, v32i8 _2) { return __lasx_xvssrani_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrani.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssrani_h_w(v16i16 _1, v16i16 _2) { return __lasx_xvssrani_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrani.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssrani_w_d(v8i32 _1, v8i32 _2) { return __lasx_xvssrani_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrani.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvssrani_d_q(v4i64 _1, v4i64 _2) { return __lasx_xvssrani_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrani.bu.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssrani_bu_h(v32u8 _1, v32i8 _2) { return __lasx_xvssrani_bu_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrani.hu.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssrani_hu_w(v16u16 _1, v16i16 _2) { return __lasx_xvssrani_hu_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrani.wu.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssrani_wu_d(v8u32 _1, v8i32 _2) { return __lasx_xvssrani_wu_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_du_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrani.du.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvssrani_du_q(v4u64 _1, v4i64 _2) { return __lasx_xvssrani_du_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrarni.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssrarni_b_h(v32i8 _1, v32i8 _2) { return __lasx_xvssrarni_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrarni.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssrarni_h_w(v16i16 _1, v16i16 _2) { return __lasx_xvssrarni_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrarni.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssrarni_w_d(v8i32 _1, v8i32 _2) { return __lasx_xvssrarni_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrarni.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvssrarni_d_q(v4i64 _1, v4i64 _2) { return __lasx_xvssrarni_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrarni.bu.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssrarni_bu_h(v32u8 _1, v32i8 _2) { return __lasx_xvssrarni_bu_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrarni.hu.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssrarni_hu_w(v16u16 _1, v16i16 _2) { return __lasx_xvssrarni_hu_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrarni.wu.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssrarni_wu_d(v8u32 _1, v8i32 _2) { return __lasx_xvssrarni_wu_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_du_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrarni.du.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvssrarni_du_q(v4u64 _1, v4i64 _2) { return __lasx_xvssrarni_du_q(_1, _2, 1); } // CHECK-LABEL: define dso_local signext i32 @xbnz_b( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbnz.b(<32 x i8> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6561,7 +6561,7 @@ int xbnz_b(v32u8 _1) { return __lasx_xbnz_b(_1); } // CHECK-LABEL: define dso_local signext i32 @xbnz_d( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbnz.d(<4 x i64> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6569,7 +6569,7 @@ int xbnz_d(v4u64 _1) { return __lasx_xbnz_d(_1); } // CHECK-LABEL: define dso_local signext i32 @xbnz_h( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbnz.h(<16 x i16> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6577,7 +6577,7 @@ int xbnz_h(v16u16 _1) { return __lasx_xbnz_h(_1); } // CHECK-LABEL: define dso_local signext i32 @xbnz_v( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbnz.v(<32 x i8> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6585,7 +6585,7 @@ int xbnz_v(v32u8 _1) { return __lasx_xbnz_v(_1); } // CHECK-LABEL: define dso_local signext i32 @xbnz_w( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbnz.w(<8 x i32> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6593,7 +6593,7 @@ int xbnz_w(v8u32 _1) { return __lasx_xbnz_w(_1); } // CHECK-LABEL: define dso_local signext i32 @xbz_b( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbz.b(<32 x i8> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6601,7 +6601,7 @@ int xbz_b(v32u8 _1) { return __lasx_xbz_b(_1); } // CHECK-LABEL: define dso_local signext i32 @xbz_d( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbz.d(<4 x i64> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6609,7 +6609,7 @@ int xbz_d(v4u64 _1) { return __lasx_xbz_d(_1); } // CHECK-LABEL: define dso_local signext i32 @xbz_h( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbz.h(<16 x i16> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6617,7 +6617,7 @@ int xbz_h(v16u16 _1) { return __lasx_xbz_h(_1); } // CHECK-LABEL: define dso_local signext i32 @xbz_v( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbz.v(<32 x i8> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6625,7 +6625,7 @@ int xbz_v(v32u8 _1) { return __lasx_xbz_v(_1); } // CHECK-LABEL: define dso_local signext i32 @xbz_w( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbz.w(<8 x i32> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6633,458 +6633,458 @@ int xbz_w(v8u32 _1) { return __lasx_xbz_w(_1); } // CHECK-LABEL: define dso_local void @xvfcmp_caf_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.caf.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_caf_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_caf_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_caf_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.caf.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_caf_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_caf_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_ceq_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.ceq.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_ceq_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_ceq_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_ceq_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.ceq.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_ceq_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_ceq_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cle_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cle.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cle_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_cle_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cle_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cle.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cle_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_cle_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_clt_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.clt.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_clt_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_clt_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_clt_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.clt.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_clt_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_clt_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cne_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cne.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cne_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_cne_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cne_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cne.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cne_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_cne_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cor_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cor.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cor_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_cor_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cor_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cor.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cor_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_cor_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cueq_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cueq.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cueq_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_cueq_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cueq_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cueq.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cueq_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_cueq_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cule_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cule.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cule_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_cule_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cule_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cule.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cule_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_cule_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cult_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cult.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cult_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_cult_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cult_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cult.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cult_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_cult_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cun_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cun.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cun_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_cun_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cune_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cune.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cune_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_cune_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cune_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cune.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cune_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_cune_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cun_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cun.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cun_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_cun_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_saf_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.saf.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_saf_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_saf_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_saf_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.saf.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_saf_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_saf_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_seq_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.seq.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_seq_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_seq_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_seq_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.seq.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_seq_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_seq_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sle_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sle.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sle_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_sle_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sle_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sle.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sle_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_sle_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_slt_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.slt.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_slt_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_slt_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_slt_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.slt.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_slt_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_slt_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sne_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sne.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sne_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_sne_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sne_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sne.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sne_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_sne_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sor_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sor.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sor_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_sor_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sor_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sor.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sor_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_sor_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sueq_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sueq.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sueq_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_sueq_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sueq_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sueq.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sueq_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_sueq_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sule_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sule.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sule_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_sule_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sule_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sule.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sule_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_sule_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sult_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sult.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sult_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_sult_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sult_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sult.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sult_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_sult_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sun_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sun.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sun_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_sun_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sune_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sune.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sune_d(v4f64 _1, v4f64 _2) { return __lasx_xvfcmp_sune_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sune_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sune.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sune_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_sune_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sun_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sun.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sun_s(v8f32 _1, v8f32 _2) { return __lasx_xvfcmp_sun_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickve_d_f( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvpickve.d.f(<4 x double> [[_1]], i32 1) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvpickve_d_f(v4f64 _1) { return __lasx_xvpickve_d_f(_1, 1); } // CHECK-LABEL: define dso_local void @xvpickve_w_f( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvpickve.w.f(<8 x float> [[_1]], i32 1) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvpickve_w_f(v8f32 _1) { return __lasx_xvpickve_w_f(_1, 1); } @@ -7092,7 +7092,7 @@ v8f32 xvpickve_w_f(v8f32 _1) { return __lasx_xvpickve_w_f(_1, 1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvrepli.b(i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvrepli_b() { return __lasx_xvrepli_b(1); } @@ -7100,7 +7100,7 @@ v32i8 xvrepli_b() { return __lasx_xvrepli_b(1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvrepli.d(i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvrepli_d() { return __lasx_xvrepli_d(1); } @@ -7108,7 +7108,7 @@ v4i64 xvrepli_d() { return __lasx_xvrepli_d(1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvrepli.h(i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvrepli_h() { return __lasx_xvrepli_h(1); } @@ -7116,15 +7116,15 @@ v16i16 xvrepli_h() { return __lasx_xvrepli_h(1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvrepli.w(i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvrepli_w() { return __lasx_xvrepli_w(1); } //. -// CHECK: [[CHAR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"Simple C/C++ TBAA"} -// CHECK: [[META5]] = !{[[META6:![0-9]+]]} -// CHECK: [[META6]] = distinct !{[[META6]], [[META7:![0-9]+]], !"__lasx_xvldx: %agg.result"} -// CHECK: [[META7]] = distinct !{[[META7]], !"__lasx_xvldx"} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[CHAR_TBAA6]] = !{[[META4]], [[META4]], i64 0} +// CHECK: [[META7]] = !{[[META8:![0-9]+]]} +// CHECK: [[META8]] = distinct !{[[META8]], [[META9:![0-9]+]], !"__lasx_xvldx: %agg.result"} +// CHECK: [[META9]] = distinct !{[[META9]], !"__lasx_xvldx"} //. diff --git a/clang/test/CodeGen/LoongArch/lasx/builtin-approximate-alias.c b/clang/test/CodeGen/LoongArch/lasx/builtin-approximate-alias.c index b194ea8..1cbcdcf 100644 --- a/clang/test/CodeGen/LoongArch/lasx/builtin-approximate-alias.c +++ b/clang/test/CodeGen/LoongArch/lasx/builtin-approximate-alias.c @@ -6,41 +6,41 @@ // CHECK-LABEL: define dso_local void @xvfrecipe_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2:![0-9]+]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6:![0-9]+]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrecipe.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfrecipe_s(v8f32 _1) { return __lasx_xvfrecipe_s(_1); } // CHECK-LABEL: define dso_local void @xvfrecipe_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrecipe.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfrecipe_d(v4f64 _1) { return __lasx_xvfrecipe_d(_1); } // CHECK-LABEL: define dso_local void @xvfrsqrte_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrsqrte.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfrsqrte_s(v8f32 _1) { return __lasx_xvfrsqrte_s(_1); } // CHECK-LABEL: define dso_local void @xvfrsqrte_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrsqrte.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfrsqrte_d(v4f64 _1) { return __lasx_xvfrsqrte_d(_1); } //. -// CHECK: [[CHAR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[CHAR_TBAA6]] = !{[[META4]], [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/LoongArch/lasx/builtin-approximate.c b/clang/test/CodeGen/LoongArch/lasx/builtin-approximate.c index 9d543df..5276a48 100644 --- a/clang/test/CodeGen/LoongArch/lasx/builtin-approximate.c +++ b/clang/test/CodeGen/LoongArch/lasx/builtin-approximate.c @@ -7,41 +7,41 @@ typedef double v4f64 __attribute__((vector_size(32), aligned(32))); // CHECK-LABEL: define dso_local void @xvfrecipe_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2:![0-9]+]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6:![0-9]+]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrecipe.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfrecipe_s(v8f32 _1) { return __builtin_lasx_xvfrecipe_s(_1); } // CHECK-LABEL: define dso_local void @xvfrecipe_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrecipe.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfrecipe_d(v4f64 _1) { return __builtin_lasx_xvfrecipe_d(_1); } // CHECK-LABEL: define dso_local void @xvfrsqrte_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrsqrte.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfrsqrte_s(v8f32 _1) { return __builtin_lasx_xvfrsqrte_s(_1); } // CHECK-LABEL: define dso_local void @xvfrsqrte_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrsqrte.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfrsqrte_d(v4f64 _1) { return __builtin_lasx_xvfrsqrte_d(_1); } //. -// CHECK: [[CHAR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[CHAR_TBAA6]] = !{[[META4]], [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/LoongArch/lasx/builtin.c b/clang/test/CodeGen/LoongArch/lasx/builtin.c index 9b21c7e..700e845 100644 --- a/clang/test/CodeGen/LoongArch/lasx/builtin.c +++ b/clang/test/CodeGen/LoongArch/lasx/builtin.c @@ -28,2808 +28,2808 @@ typedef double v4f64_d __attribute__((vector_size(32), aligned(8))); // CHECK-LABEL: define dso_local void @xvsll_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2:![0-9]+]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6:![0-9]+]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsll.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsll_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsll_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsll_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsll.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsll_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsll_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsll_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsll.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsll_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsll_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsll_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsll.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsll_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsll_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvslli_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvslli.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvslli_b(v32i8 _1) { return __builtin_lasx_xvslli_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvslli_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvslli.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvslli_h(v16i16 _1) { return __builtin_lasx_xvslli_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvslli_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvslli.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvslli_w(v8i32 _1) { return __builtin_lasx_xvslli_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvslli_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvslli.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvslli_d(v4i64 _1) { return __builtin_lasx_xvslli_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvsra_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsra.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsra_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsra_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsra_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsra.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsra_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsra_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsra_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsra.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsra_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsra_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsra_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsra.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsra_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsra_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrai_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrai.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrai_b(v32i8 _1) { return __builtin_lasx_xvsrai_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrai_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrai.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrai_h(v16i16 _1) { return __builtin_lasx_xvsrai_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrai_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrai.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrai_w(v8i32 _1) { return __builtin_lasx_xvsrai_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrai_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrai.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrai_d(v4i64 _1) { return __builtin_lasx_xvsrai_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrar_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrar.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrar_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsrar_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrar_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrar.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrar_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsrar_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrar_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrar.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrar_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsrar_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrar_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrar.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrar_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsrar_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrari_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrari.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrari_b(v32i8 _1) { return __builtin_lasx_xvsrari_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrari_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrari.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrari_h(v16i16 _1) { return __builtin_lasx_xvsrari_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrari_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrari.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrari_w(v8i32 _1) { return __builtin_lasx_xvsrari_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrari_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrari.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrari_d(v4i64 _1) { return __builtin_lasx_xvsrari_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrl_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrl.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrl_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsrl_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrl_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrl.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrl_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsrl_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrl_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrl.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrl_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsrl_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrl_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrl.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrl_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsrl_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrli_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrli.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrli_b(v32i8 _1) { return __builtin_lasx_xvsrli_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrli_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrli.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrli_h(v16i16 _1) { return __builtin_lasx_xvsrli_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrli_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrli.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrli_w(v8i32 _1) { return __builtin_lasx_xvsrli_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrli_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrli.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrli_d(v4i64 _1) { return __builtin_lasx_xvsrli_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrlr_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrlr.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrlr_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsrlr_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrlr_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrlr.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrlr_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsrlr_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrlr_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrlr.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrlr_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsrlr_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrlr_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrlr.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrlr_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsrlr_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrlri_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrlri.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrlri_b(v32i8 _1) { return __builtin_lasx_xvsrlri_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrlri_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrlri.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrlri_h(v16i16 _1) { return __builtin_lasx_xvsrlri_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrlri_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrlri.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrlri_w(v8i32 _1) { return __builtin_lasx_xvsrlri_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvsrlri_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrlri.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrlri_d(v4i64 _1) { return __builtin_lasx_xvsrlri_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitclr_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitclr.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitclr_b(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvbitclr_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitclr_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvbitclr.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvbitclr_h(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvbitclr_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitclr_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvbitclr.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvbitclr_w(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvbitclr_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitclr_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvbitclr.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvbitclr_d(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvbitclr_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitclri_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitclri.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitclri_b(v32u8 _1) { return __builtin_lasx_xvbitclri_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitclri_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvbitclri.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvbitclri_h(v16u16 _1) { return __builtin_lasx_xvbitclri_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitclri_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvbitclri.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvbitclri_w(v8u32 _1) { return __builtin_lasx_xvbitclri_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitclri_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvbitclri.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvbitclri_d(v4u64 _1) { return __builtin_lasx_xvbitclri_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitset_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitset.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitset_b(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvbitset_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitset_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvbitset.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvbitset_h(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvbitset_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitset_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvbitset.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvbitset_w(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvbitset_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitset_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvbitset.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvbitset_d(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvbitset_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitseti_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitseti.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitseti_b(v32u8 _1) { return __builtin_lasx_xvbitseti_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitseti_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvbitseti.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvbitseti_h(v16u16 _1) { return __builtin_lasx_xvbitseti_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitseti_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvbitseti.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvbitseti_w(v8u32 _1) { return __builtin_lasx_xvbitseti_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitseti_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvbitseti.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvbitseti_d(v4u64 _1) { return __builtin_lasx_xvbitseti_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitrev_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitrev.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitrev_b(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvbitrev_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitrev_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvbitrev.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvbitrev_h(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvbitrev_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitrev_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvbitrev.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvbitrev_w(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvbitrev_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitrev_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvbitrev.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvbitrev_d(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvbitrev_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvbitrevi_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitrevi.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitrevi_b(v32u8 _1) { return __builtin_lasx_xvbitrevi_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitrevi_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvbitrevi.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvbitrevi_h(v16u16 _1) { return __builtin_lasx_xvbitrevi_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitrevi_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvbitrevi.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvbitrevi_w(v8u32 _1) { return __builtin_lasx_xvbitrevi_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitrevi_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvbitrevi.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvbitrevi_d(v4u64 _1) { return __builtin_lasx_xvbitrevi_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvadd_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvadd.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvadd_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvadd_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvadd_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvadd.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvadd_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvadd_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvadd_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvadd.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvadd_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvadd_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvadd_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvadd.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvadd_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvadd_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddi_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvaddi.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvaddi_bu(v32i8 _1) { return __builtin_lasx_xvaddi_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvaddi_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvaddi.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvaddi_hu(v16i16 _1) { return __builtin_lasx_xvaddi_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvaddi_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvaddi.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvaddi_wu(v8i32 _1) { return __builtin_lasx_xvaddi_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvaddi_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddi.du(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddi_du(v4i64 _1) { return __builtin_lasx_xvaddi_du(_1, 1); } // CHECK-LABEL: define dso_local void @xvsub_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsub.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsub_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsub_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsub_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsub.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsub_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsub_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsub_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsub.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsub_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsub_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsub_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsub.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsub_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsub_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubi_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsubi.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsubi_bu(v32i8 _1) { return __builtin_lasx_xvsubi_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsubi_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsubi.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsubi_hu(v16i16 _1) { return __builtin_lasx_xvsubi_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsubi_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsubi.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsubi_wu(v8i32 _1) { return __builtin_lasx_xvsubi_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsubi_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubi.du(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubi_du(v4i64 _1) { return __builtin_lasx_xvsubi_du(_1, 1); } // CHECK-LABEL: define dso_local void @xvmax_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmax.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmax_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvmax_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmax_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmax.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmax_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvmax_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmax_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmax.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmax_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvmax_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmax_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmax.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmax_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvmax_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmaxi_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmaxi.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmaxi_b(v32i8 _1) { return __builtin_lasx_xvmaxi_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvmaxi_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaxi.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmaxi_h(v16i16 _1) { return __builtin_lasx_xvmaxi_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvmaxi_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaxi.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmaxi_w(v8i32 _1) { return __builtin_lasx_xvmaxi_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvmaxi_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaxi.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaxi_d(v4i64 _1) { return __builtin_lasx_xvmaxi_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvmax_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmax.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvmax_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvmax_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmax_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmax.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmax_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvmax_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmax_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmax.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmax_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvmax_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmax_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmax.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmax_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvmax_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvmaxi_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmaxi.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvmaxi_bu(v32u8 _1) { return __builtin_lasx_xvmaxi_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvmaxi_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaxi.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmaxi_hu(v16u16 _1) { return __builtin_lasx_xvmaxi_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvmaxi_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaxi.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmaxi_wu(v8u32 _1) { return __builtin_lasx_xvmaxi_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvmaxi_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaxi.du(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmaxi_du(v4u64 _1) { return __builtin_lasx_xvmaxi_du(_1, 1); } // CHECK-LABEL: define dso_local void @xvmin_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmin.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmin_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvmin_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmin_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmin.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmin_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvmin_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmin_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmin.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmin_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvmin_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmin_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmin.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmin_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvmin_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmini_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmini.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmini_b(v32i8 _1) { return __builtin_lasx_xvmini_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvmini_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmini.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmini_h(v16i16 _1) { return __builtin_lasx_xvmini_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvmini_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmini.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmini_w(v8i32 _1) { return __builtin_lasx_xvmini_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvmini_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmini.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmini_d(v4i64 _1) { return __builtin_lasx_xvmini_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvmin_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmin.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvmin_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvmin_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmin_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmin.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmin_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvmin_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmin_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmin.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmin_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvmin_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmin_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmin.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmin_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvmin_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvmini_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmini.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvmini_bu(v32u8 _1) { return __builtin_lasx_xvmini_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvmini_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmini.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmini_hu(v16u16 _1) { return __builtin_lasx_xvmini_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvmini_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmini.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmini_wu(v8u32 _1) { return __builtin_lasx_xvmini_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvmini_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmini.du(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmini_du(v4u64 _1) { return __builtin_lasx_xvmini_du(_1, 1); } // CHECK-LABEL: define dso_local void @xvseq_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvseq.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvseq_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvseq_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvseq_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvseq.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvseq_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvseq_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvseq_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvseq.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvseq_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvseq_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvseq_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvseq.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvseq_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvseq_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvseqi_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvseqi.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvseqi_b(v32i8 _1) { return __builtin_lasx_xvseqi_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvseqi_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvseqi.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvseqi_h(v16i16 _1) { return __builtin_lasx_xvseqi_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvseqi_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvseqi.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvseqi_w(v8i32 _1) { return __builtin_lasx_xvseqi_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvseqi_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvseqi.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvseqi_d(v4i64 _1) { return __builtin_lasx_xvseqi_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvslt_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvslt.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvslt_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvslt_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvslt_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvslt.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvslt_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvslt_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvslt_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvslt.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvslt_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvslt_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvslt_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvslt.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvslt_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvslt_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvslti_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvslti.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvslti_b(v32i8 _1) { return __builtin_lasx_xvslti_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvslti_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvslti.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvslti_h(v16i16 _1) { return __builtin_lasx_xvslti_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvslti_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvslti.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvslti_w(v8i32 _1) { return __builtin_lasx_xvslti_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvslti_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvslti.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvslti_d(v4i64 _1) { return __builtin_lasx_xvslti_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvslt_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvslt.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvslt_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvslt_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvslt_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvslt.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvslt_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvslt_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvslt_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvslt.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvslt_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvslt_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvslt_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvslt.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvslt_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvslt_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvslti_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvslti.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvslti_bu(v32u8 _1) { return __builtin_lasx_xvslti_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvslti_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvslti.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvslti_hu(v16u16 _1) { return __builtin_lasx_xvslti_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvslti_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvslti.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvslti_wu(v8u32 _1) { return __builtin_lasx_xvslti_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvslti_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvslti.du(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvslti_du(v4u64 _1) { return __builtin_lasx_xvslti_du(_1, 1); } // CHECK-LABEL: define dso_local void @xvsle_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsle.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsle_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsle_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsle_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsle.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsle_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsle_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsle_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsle.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsle_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsle_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsle_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsle.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsle_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsle_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvslei_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvslei.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvslei_b(v32i8 _1) { return __builtin_lasx_xvslei_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvslei_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvslei.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvslei_h(v16i16 _1) { return __builtin_lasx_xvslei_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvslei_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvslei.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvslei_w(v8i32 _1) { return __builtin_lasx_xvslei_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvslei_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvslei.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvslei_d(v4i64 _1) { return __builtin_lasx_xvslei_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvsle_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsle.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsle_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvsle_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsle_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsle.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsle_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvsle_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsle_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsle.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsle_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvsle_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsle_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsle.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsle_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvsle_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvslei_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvslei.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvslei_bu(v32u8 _1) { return __builtin_lasx_xvslei_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvslei_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvslei.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvslei_hu(v16u16 _1) { return __builtin_lasx_xvslei_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvslei_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvslei.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvslei_wu(v8u32 _1) { return __builtin_lasx_xvslei_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvslei_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvslei.du(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvslei_du(v4u64 _1) { return __builtin_lasx_xvslei_du(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsat.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsat_b(v32i8 _1) { return __builtin_lasx_xvsat_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsat.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsat_h(v16i16 _1) { return __builtin_lasx_xvsat_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsat.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsat_w(v8i32 _1) { return __builtin_lasx_xvsat_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsat.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsat_d(v4i64 _1) { return __builtin_lasx_xvsat_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsat.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvsat_bu(v32u8 _1) { return __builtin_lasx_xvsat_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsat.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvsat_hu(v16u16 _1) { return __builtin_lasx_xvsat_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsat.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvsat_wu(v8u32 _1) { return __builtin_lasx_xvsat_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsat_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsat.du(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvsat_du(v4u64 _1) { return __builtin_lasx_xvsat_du(_1, 1); } // CHECK-LABEL: define dso_local void @xvadda_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvadda.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvadda_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvadda_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvadda_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvadda.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvadda_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvadda_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvadda_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvadda.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvadda_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvadda_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvadda_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvadda.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvadda_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvadda_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsadd.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsadd_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsadd_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsadd.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsadd_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsadd_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsadd.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsadd_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsadd_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsadd.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsadd_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsadd_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsadd.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvsadd_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvsadd_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsadd.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvsadd_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvsadd_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsadd.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvsadd_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvsadd_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsadd_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsadd.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvsadd_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvsadd_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvavg.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvavg_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvavg_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvavg.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvavg_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvavg_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvavg.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvavg_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvavg_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvavg.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvavg_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvavg_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvavg.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvavg_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvavg_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvavg.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvavg_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvavg_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvavg.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvavg_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvavg_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvavg_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvavg.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvavg_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvavg_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvavgr.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvavgr_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvavgr_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvavgr.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvavgr_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvavgr_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvavgr.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvavgr_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvavgr_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvavgr.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvavgr_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvavgr_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvavgr.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvavgr_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvavgr_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvavgr.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvavgr_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvavgr_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvavgr.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvavgr_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvavgr_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvavgr_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvavgr.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvavgr_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvavgr_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssub.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssub_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvssub_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssub.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssub_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvssub_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssub.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssub_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvssub_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssub.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvssub_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvssub_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssub.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssub_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvssub_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssub.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssub_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvssub_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssub.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssub_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvssub_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvssub_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssub.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvssub_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvssub_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvabsd.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvabsd_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvabsd_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvabsd.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvabsd_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvabsd_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvabsd.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvabsd_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvabsd_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvabsd.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvabsd_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvabsd_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvabsd.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvabsd_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvabsd_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvabsd.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvabsd_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvabsd_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvabsd.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvabsd_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvabsd_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvabsd_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvabsd.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvabsd_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvabsd_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvmul_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmul.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmul_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvmul_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmul_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmul.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmul_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvmul_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmul_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmul.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmul_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvmul_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmul_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmul.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmul_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvmul_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmadd_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmadd.b(<32 x i8> [[_1]], <32 x i8> [[_2]], <32 x i8> [[_3]]) -// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmadd_b(v32i8 _1, v32i8 _2, v32i8 _3) { return __builtin_lasx_xvmadd_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmadd_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmadd.h(<16 x i16> [[_1]], <16 x i16> [[_2]], <16 x i16> [[_3]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmadd_h(v16i16 _1, v16i16 _2, v16i16 _3) { return __builtin_lasx_xvmadd_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmadd_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmadd.w(<8 x i32> [[_1]], <8 x i32> [[_2]], <8 x i32> [[_3]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmadd_w(v8i32 _1, v8i32 _2, v8i32 _3) { return __builtin_lasx_xvmadd_w(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmadd_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmadd.d(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmadd_d(v4i64 _1, v4i64 _2, v4i64 _3) { return __builtin_lasx_xvmadd_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmsub_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmsub.b(<32 x i8> [[_1]], <32 x i8> [[_2]], <32 x i8> [[_3]]) -// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmsub_b(v32i8 _1, v32i8 _2, v32i8 _3) { return __builtin_lasx_xvmsub_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmsub_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmsub.h(<16 x i16> [[_1]], <16 x i16> [[_2]], <16 x i16> [[_3]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmsub_h(v16i16 _1, v16i16 _2, v16i16 _3) { return __builtin_lasx_xvmsub_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmsub_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmsub.w(<8 x i32> [[_1]], <8 x i32> [[_2]], <8 x i32> [[_3]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmsub_w(v8i32 _1, v8i32 _2, v8i32 _3) { return __builtin_lasx_xvmsub_w(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmsub_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmsub.d(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmsub_d(v4i64 _1, v4i64 _2, v4i64 _3) { return __builtin_lasx_xvmsub_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvdiv_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvdiv.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvdiv_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvdiv_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvdiv_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvdiv.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvdiv_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvdiv_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvdiv_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvdiv.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvdiv_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvdiv_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvdiv_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvdiv.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvdiv_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvdiv_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvdiv_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvdiv.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvdiv_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvdiv_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvdiv_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvdiv.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvdiv_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvdiv_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvdiv_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvdiv.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvdiv_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvdiv_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvdiv_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvdiv.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvdiv_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvdiv_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvhaddw.h.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvhaddw_h_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvhaddw_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvhaddw.w.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvhaddw_w_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvhaddw_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhaddw.d.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvhaddw_d_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvhaddw_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_hu_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvhaddw.hu.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvhaddw_hu_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvhaddw_hu_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_wu_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvhaddw.wu.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvhaddw_wu_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvhaddw_wu_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_du_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhaddw.du.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvhaddw_du_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvhaddw_du_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvhsubw.h.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvhsubw_h_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvhsubw_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvhsubw.w.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvhsubw_w_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvhsubw_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhsubw.d.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvhsubw_d_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvhsubw_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_hu_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvhsubw.hu.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvhsubw_hu_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvhsubw_hu_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_wu_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvhsubw.wu.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvhsubw_wu_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvhsubw_wu_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_du_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhsubw.du.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvhsubw_du_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvhsubw_du_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmod.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmod_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvmod_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmod.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmod_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvmod_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmod.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmod_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvmod_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmod.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmod_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvmod_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmod.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvmod_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvmod_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmod.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmod_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvmod_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmod.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmod_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvmod_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmod_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmod.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmod_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvmod_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvrepl128vei_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvrepl128vei.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvrepl128vei_b(v32i8 _1) { return __builtin_lasx_xvrepl128vei_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvrepl128vei_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvrepl128vei.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvrepl128vei_h(v16i16 _1) { return __builtin_lasx_xvrepl128vei_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvrepl128vei_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvrepl128vei.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvrepl128vei_w(v8i32 _1) { return __builtin_lasx_xvrepl128vei_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvrepl128vei_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvrepl128vei.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvrepl128vei_d(v4i64 _1) { return __builtin_lasx_xvrepl128vei_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvpickev_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvpickev.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvpickev_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvpickev_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickev_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvpickev.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvpickev_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvpickev_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickev_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvpickev.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvpickev_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvpickev_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickev_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvpickev.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvpickev_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvpickev_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickod_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvpickod.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvpickod_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvpickod_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickod_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvpickod.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvpickod_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvpickod_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickod_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvpickod.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvpickod_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvpickod_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickod_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvpickod.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvpickod_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvpickod_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvh_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvilvh.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvilvh_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvilvh_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvh_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvilvh.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvilvh_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvilvh_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvh_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvilvh.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvilvh_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvilvh_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvh_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvilvh.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvilvh_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvilvh_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvl_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvilvl.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvilvl_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvilvl_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvl_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvilvl.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvilvl_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvilvl_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvl_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvilvl.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvilvl_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvilvl_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvilvl_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvilvl.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvilvl_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvilvl_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackev_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvpackev.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvpackev_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvpackev_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackev_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvpackev.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvpackev_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvpackev_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackev_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvpackev.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvpackev_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvpackev_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackev_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvpackev.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvpackev_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvpackev_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackod_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvpackod.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvpackod_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvpackod_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackod_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvpackod.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvpackod_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvpackod_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackod_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvpackod.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvpackod_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvpackod_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvpackod_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvpackod.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvpackod_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvpackod_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvshuf_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvshuf.b(<32 x i8> [[_1]], <32 x i8> [[_2]], <32 x i8> [[_3]]) -// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvshuf_b(v32i8 _1, v32i8 _2, v32i8 _3) { return __builtin_lasx_xvshuf_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvshuf_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvshuf.h(<16 x i16> [[_1]], <16 x i16> [[_2]], <16 x i16> [[_3]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvshuf_h(v16i16 _1, v16i16 _2, v16i16 _3) { return __builtin_lasx_xvshuf_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvshuf_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvshuf.w(<8 x i32> [[_1]], <8 x i32> [[_2]], <8 x i32> [[_3]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvshuf_w(v8i32 _1, v8i32 _2, v8i32 _3) { return __builtin_lasx_xvshuf_w(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvshuf_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvshuf.d(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvshuf_d(v4i64 _1, v4i64 _2, v4i64 _3) { return __builtin_lasx_xvshuf_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvand_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvand.v(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvand_v(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvand_v(_1, _2); } // CHECK-LABEL: define dso_local void @xvandi_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvandi.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvandi_b(v32u8 _1) { return __builtin_lasx_xvandi_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvor_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvor.v(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvor_v(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvor_v(_1, _2); } // CHECK-LABEL: define dso_local void @xvori_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvori.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvori_b(v32u8 _1) { return __builtin_lasx_xvori_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvnor_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvnor.v(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvnor_v(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvnor_v(_1, _2); } // CHECK-LABEL: define dso_local void @xvnori_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvnori.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvnori_b(v32u8 _1) { return __builtin_lasx_xvnori_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvxor_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvxor.v(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvxor_v(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvxor_v(_1, _2); } // CHECK-LABEL: define dso_local void @xvxori_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvxori.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvxori_b(v32u8 _1) { return __builtin_lasx_xvxori_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvbitsel_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitsel.v(<32 x i8> [[_1]], <32 x i8> [[_2]], <32 x i8> [[_3]]) -// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitsel_v(v32u8 _1, v32u8 _2, v32u8 _3) { return __builtin_lasx_xvbitsel_v(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvbitseli_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbitseli.b(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvbitseli_b(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvbitseli_b(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvshuf4i_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvshuf4i.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvshuf4i_b(v32i8 _1) { return __builtin_lasx_xvshuf4i_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvshuf4i_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvshuf4i.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvshuf4i_h(v16i16 _1) { return __builtin_lasx_xvshuf4i_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvshuf4i_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvshuf4i.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvshuf4i_w(v8i32 _1) { return __builtin_lasx_xvshuf4i_w(_1, 1); } @@ -2837,7 +2837,7 @@ v8i32 xvshuf4i_w(v8i32 _1) { return __builtin_lasx_xvshuf4i_w(_1, 1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], i32 noundef signext [[_1:%.*]]) local_unnamed_addr #[[ATTR2:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvreplgr2vr.b(i32 [[_1]]) -// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvreplgr2vr_b(int _1) { return __builtin_lasx_xvreplgr2vr_b(_1); } @@ -2845,7 +2845,7 @@ v32i8 xvreplgr2vr_b(int _1) { return __builtin_lasx_xvreplgr2vr_b(_1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], i32 noundef signext [[_1:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvreplgr2vr.h(i32 [[_1]]) -// CHECK-NEXT: store <16 x i16> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvreplgr2vr_h(int _1) { return __builtin_lasx_xvreplgr2vr_h(_1); } @@ -2853,7 +2853,7 @@ v16i16 xvreplgr2vr_h(int _1) { return __builtin_lasx_xvreplgr2vr_h(_1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], i32 noundef signext [[_1:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvreplgr2vr.w(i32 [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvreplgr2vr_w(int _1) { return __builtin_lasx_xvreplgr2vr_w(_1); } @@ -2862,1641 +2862,1641 @@ v8i32 xvreplgr2vr_w(int _1) { return __builtin_lasx_xvreplgr2vr_w(_1); } // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[CONV:%.*]] = sext i32 [[_1]] to i64 // CHECK-NEXT: [[TMP0:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvreplgr2vr.d(i64 [[CONV]]) -// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvreplgr2vr_d(int _1) { return __builtin_lasx_xvreplgr2vr_d(_1); } // CHECK-LABEL: define dso_local void @xvpcnt_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvpcnt.b(<32 x i8> [[_1]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvpcnt_b(v32i8 _1) { return __builtin_lasx_xvpcnt_b(_1); } // CHECK-LABEL: define dso_local void @xvpcnt_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvpcnt.h(<16 x i16> [[_1]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvpcnt_h(v16i16 _1) { return __builtin_lasx_xvpcnt_h(_1); } // CHECK-LABEL: define dso_local void @xvpcnt_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvpcnt.w(<8 x i32> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvpcnt_w(v8i32 _1) { return __builtin_lasx_xvpcnt_w(_1); } // CHECK-LABEL: define dso_local void @xvpcnt_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvpcnt.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvpcnt_d(v4i64 _1) { return __builtin_lasx_xvpcnt_d(_1); } // CHECK-LABEL: define dso_local void @xvclo_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvclo.b(<32 x i8> [[_1]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvclo_b(v32i8 _1) { return __builtin_lasx_xvclo_b(_1); } // CHECK-LABEL: define dso_local void @xvclo_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvclo.h(<16 x i16> [[_1]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvclo_h(v16i16 _1) { return __builtin_lasx_xvclo_h(_1); } // CHECK-LABEL: define dso_local void @xvclo_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvclo.w(<8 x i32> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvclo_w(v8i32 _1) { return __builtin_lasx_xvclo_w(_1); } // CHECK-LABEL: define dso_local void @xvclo_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvclo.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvclo_d(v4i64 _1) { return __builtin_lasx_xvclo_d(_1); } // CHECK-LABEL: define dso_local void @xvclz_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvclz.b(<32 x i8> [[_1]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvclz_b(v32i8 _1) { return __builtin_lasx_xvclz_b(_1); } // CHECK-LABEL: define dso_local void @xvclz_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvclz.h(<16 x i16> [[_1]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvclz_h(v16i16 _1) { return __builtin_lasx_xvclz_h(_1); } // CHECK-LABEL: define dso_local void @xvclz_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvclz.w(<8 x i32> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvclz_w(v8i32 _1) { return __builtin_lasx_xvclz_w(_1); } // CHECK-LABEL: define dso_local void @xvclz_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvclz.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvclz_d(v4i64 _1) { return __builtin_lasx_xvclz_d(_1); } // CHECK-LABEL: define dso_local void @xvfadd_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfadd.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfadd_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfadd_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfadd_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfadd.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfadd_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfadd_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfsub_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfsub.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfsub_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfsub_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfsub_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfsub.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfsub_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfsub_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmul_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfmul.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfmul_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfmul_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmul_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfmul.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfmul_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfmul_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfdiv_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfdiv.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfdiv_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfdiv_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfdiv_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfdiv.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfdiv_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfdiv_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcvt_h_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvfcvt.h.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvfcvt_h_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcvt_h_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcvt_s_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfcvt.s.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfcvt_s_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcvt_s_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmin_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfmin.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfmin_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfmin_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmin_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfmin.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfmin_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfmin_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmina_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfmina.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfmina_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfmina_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmina_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfmina.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfmina_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfmina_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmax_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfmax.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfmax_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfmax_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmax_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfmax.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfmax_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfmax_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmaxa_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfmaxa.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfmaxa_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfmaxa_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmaxa_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfmaxa.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfmaxa_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfmaxa_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfclass_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfclass.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfclass_s(v8f32 _1) { return __builtin_lasx_xvfclass_s(_1); } // CHECK-LABEL: define dso_local void @xvfclass_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfclass.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfclass_d(v4f64 _1) { return __builtin_lasx_xvfclass_d(_1); } // CHECK-LABEL: define dso_local void @xvfsqrt_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfsqrt.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfsqrt_s(v8f32 _1) { return __builtin_lasx_xvfsqrt_s(_1); } // CHECK-LABEL: define dso_local void @xvfsqrt_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfsqrt.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfsqrt_d(v4f64 _1) { return __builtin_lasx_xvfsqrt_d(_1); } // CHECK-LABEL: define dso_local void @xvfrecip_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrecip.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfrecip_s(v8f32 _1) { return __builtin_lasx_xvfrecip_s(_1); } // CHECK-LABEL: define dso_local void @xvfrecip_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrecip.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfrecip_d(v4f64 _1) { return __builtin_lasx_xvfrecip_d(_1); } // CHECK-LABEL: define dso_local void @xvfrint_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrint.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfrint_s(v8f32 _1) { return __builtin_lasx_xvfrint_s(_1); } // CHECK-LABEL: define dso_local void @xvfrint_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrint.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfrint_d(v4f64 _1) { return __builtin_lasx_xvfrint_d(_1); } // CHECK-LABEL: define dso_local void @xvfrsqrt_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrsqrt.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfrsqrt_s(v8f32 _1) { return __builtin_lasx_xvfrsqrt_s(_1); } // CHECK-LABEL: define dso_local void @xvfrsqrt_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrsqrt.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfrsqrt_d(v4f64 _1) { return __builtin_lasx_xvfrsqrt_d(_1); } // CHECK-LABEL: define dso_local void @xvflogb_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvflogb.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvflogb_s(v8f32 _1) { return __builtin_lasx_xvflogb_s(_1); } // CHECK-LABEL: define dso_local void @xvflogb_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvflogb.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvflogb_d(v4f64 _1) { return __builtin_lasx_xvflogb_d(_1); } // CHECK-LABEL: define dso_local void @xvfcvth_s_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfcvth.s.h(<16 x i16> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfcvth_s_h(v16i16 _1) { return __builtin_lasx_xvfcvth_s_h(_1); } // CHECK-LABEL: define dso_local void @xvfcvth_d_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfcvth.d.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfcvth_d_s(v8f32 _1) { return __builtin_lasx_xvfcvth_d_s(_1); } // CHECK-LABEL: define dso_local void @xvfcvtl_s_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfcvtl.s.h(<16 x i16> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfcvtl_s_h(v16i16 _1) { return __builtin_lasx_xvfcvtl_s_h(_1); } // CHECK-LABEL: define dso_local void @xvfcvtl_d_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfcvtl.d.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfcvtl_d_s(v8f32 _1) { return __builtin_lasx_xvfcvtl_d_s(_1); } // CHECK-LABEL: define dso_local void @xvftint_w_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftint.w.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftint_w_s(v8f32 _1) { return __builtin_lasx_xvftint_w_s(_1); } // CHECK-LABEL: define dso_local void @xvftint_l_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftint.l.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftint_l_d(v4f64 _1) { return __builtin_lasx_xvftint_l_d(_1); } // CHECK-LABEL: define dso_local void @xvftint_wu_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftint.wu.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvftint_wu_s(v8f32 _1) { return __builtin_lasx_xvftint_wu_s(_1); } // CHECK-LABEL: define dso_local void @xvftint_lu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftint.lu.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvftint_lu_d(v4f64 _1) { return __builtin_lasx_xvftint_lu_d(_1); } // CHECK-LABEL: define dso_local void @xvftintrz_w_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrz.w.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrz_w_s(v8f32 _1) { return __builtin_lasx_xvftintrz_w_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrz_l_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrz.l.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrz_l_d(v4f64 _1) { return __builtin_lasx_xvftintrz_l_d(_1); } // CHECK-LABEL: define dso_local void @xvftintrz_wu_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrz.wu.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvftintrz_wu_s(v8f32 _1) { return __builtin_lasx_xvftintrz_wu_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrz_lu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrz.lu.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvftintrz_lu_d(v4f64 _1) { return __builtin_lasx_xvftintrz_lu_d(_1); } // CHECK-LABEL: define dso_local void @xvffint_s_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvffint.s.w(<8 x i32> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvffint_s_w(v8i32 _1) { return __builtin_lasx_xvffint_s_w(_1); } // CHECK-LABEL: define dso_local void @xvffint_d_l( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvffint.d.l(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvffint_d_l(v4i64 _1) { return __builtin_lasx_xvffint_d_l(_1); } // CHECK-LABEL: define dso_local void @xvffint_s_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvffint.s.wu(<8 x i32> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvffint_s_wu(v8u32 _1) { return __builtin_lasx_xvffint_s_wu(_1); } // CHECK-LABEL: define dso_local void @xvffint_d_lu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvffint.d.lu(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvffint_d_lu(v4u64 _1) { return __builtin_lasx_xvffint_d_lu(_1); } // CHECK-LABEL: define dso_local void @xvreplve_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], i32 noundef signext [[_2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvreplve.b(<32 x i8> [[_1]], i32 [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvreplve_b(v32i8 _1, int _2) { return __builtin_lasx_xvreplve_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvreplve_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], i32 noundef signext [[_2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvreplve.h(<16 x i16> [[_1]], i32 [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvreplve_h(v16i16 _1, int _2) { return __builtin_lasx_xvreplve_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvreplve_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], i32 noundef signext [[_2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvreplve.w(<8 x i32> [[_1]], i32 [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvreplve_w(v8i32 _1, int _2) { return __builtin_lasx_xvreplve_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvreplve_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], i32 noundef signext [[_2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvreplve.d(<4 x i64> [[_1]], i32 [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvreplve_d(v4i64 _1, int _2) { return __builtin_lasx_xvreplve_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvpermi_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvpermi.w(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvpermi_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvpermi_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvandn_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvandn.v(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvandn_v(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvandn_v(_1, _2); } // CHECK-LABEL: define dso_local void @xvneg_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvneg.b(<32 x i8> [[_1]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvneg_b(v32i8 _1) { return __builtin_lasx_xvneg_b(_1); } // CHECK-LABEL: define dso_local void @xvneg_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvneg.h(<16 x i16> [[_1]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvneg_h(v16i16 _1) { return __builtin_lasx_xvneg_h(_1); } // CHECK-LABEL: define dso_local void @xvneg_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvneg.w(<8 x i32> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvneg_w(v8i32 _1) { return __builtin_lasx_xvneg_w(_1); } // CHECK-LABEL: define dso_local void @xvneg_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvneg.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvneg_d(v4i64 _1) { return __builtin_lasx_xvneg_d(_1); } // CHECK-LABEL: define dso_local void @xvmuh_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmuh.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmuh_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvmuh_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmuh_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmuh.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmuh_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvmuh_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmuh_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmuh.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmuh_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvmuh_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmuh_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmuh.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmuh_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvmuh_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmuh_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmuh.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvmuh_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvmuh_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmuh_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmuh.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmuh_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvmuh_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmuh_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmuh.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmuh_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvmuh_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmuh_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmuh.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmuh_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvmuh_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvsllwil_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsllwil.h.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsllwil_h_b(v32i8 _1) { return __builtin_lasx_xvsllwil_h_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvsllwil_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsllwil.w.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsllwil_w_h(v16i16 _1) { return __builtin_lasx_xvsllwil_w_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvsllwil_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsllwil.d.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsllwil_d_w(v8i32 _1) { return __builtin_lasx_xvsllwil_d_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvsllwil_hu_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsllwil.hu.bu(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvsllwil_hu_bu(v32u8 _1) { return __builtin_lasx_xvsllwil_hu_bu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsllwil_wu_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsllwil.wu.hu(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvsllwil_wu_hu(v16u16 _1) { return __builtin_lasx_xvsllwil_wu_hu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsllwil_du_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsllwil.du.wu(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvsllwil_du_wu(v8u32 _1) { return __builtin_lasx_xvsllwil_du_wu(_1, 1); } // CHECK-LABEL: define dso_local void @xvsran_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsran.b.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsran_b_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsran_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsran_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsran.h.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsran_h_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsran_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsran_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsran.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsran_w_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsran_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssran_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssran.b.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssran_b_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvssran_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssran_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssran.h.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssran_h_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvssran_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssran_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssran.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssran_w_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvssran_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssran_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssran.bu.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssran_bu_h(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvssran_bu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssran_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssran.hu.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssran_hu_w(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvssran_hu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssran_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssran.wu.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssran_wu_d(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvssran_wu_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrarn_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrarn.b.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrarn_b_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsrarn_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrarn_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrarn.h.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrarn_h_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsrarn_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrarn_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrarn.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrarn_w_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsrarn_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrarn_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrarn.b.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssrarn_b_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvssrarn_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrarn_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrarn.h.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssrarn_h_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvssrarn_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrarn_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrarn.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssrarn_w_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvssrarn_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrarn_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrarn.bu.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssrarn_bu_h(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvssrarn_bu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrarn_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrarn.hu.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssrarn_hu_w(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvssrarn_hu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrarn_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrarn.wu.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssrarn_wu_d(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvssrarn_wu_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrln_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrln.b.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrln_b_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsrln_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrln_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrln.h.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrln_h_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsrln_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrln_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrln.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrln_w_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsrln_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrln_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrln.bu.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssrln_bu_h(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvssrln_bu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrln_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrln.hu.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssrln_hu_w(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvssrln_hu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrln_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrln.wu.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssrln_wu_d(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvssrln_wu_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrlrn_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrlrn.b.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrlrn_b_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsrlrn_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrlrn_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrlrn.h.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrlrn_h_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsrlrn_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsrlrn_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrlrn.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrlrn_w_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsrlrn_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrlrn_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrlrn.bu.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssrlrn_bu_h(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvssrlrn_bu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrlrn_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrlrn.hu.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssrlrn_hu_w(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvssrlrn_hu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrlrn_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrlrn.wu.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssrlrn_wu_d(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvssrlrn_wu_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfrstpi_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvfrstpi.b(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvfrstpi_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvfrstpi_b(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvfrstpi_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvfrstpi.h(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvfrstpi_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvfrstpi_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvfrstp_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvfrstp.b(<32 x i8> [[_1]], <32 x i8> [[_2]], <32 x i8> [[_3]]) -// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvfrstp_b(v32i8 _1, v32i8 _2, v32i8 _3) { return __builtin_lasx_xvfrstp_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfrstp_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvfrstp.h(<16 x i16> [[_1]], <16 x i16> [[_2]], <16 x i16> [[_3]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvfrstp_h(v16i16 _1, v16i16 _2, v16i16 _3) { return __builtin_lasx_xvfrstp_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvshuf4i_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvshuf4i.d(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvshuf4i_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvshuf4i_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvbsrl_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbsrl.v(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvbsrl_v(v32i8 _1) { return __builtin_lasx_xvbsrl_v(_1, 1); } // CHECK-LABEL: define dso_local void @xvbsll_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvbsll.v(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvbsll_v(v32i8 _1) { return __builtin_lasx_xvbsll_v(_1, 1); } // CHECK-LABEL: define dso_local void @xvextrins_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvextrins.b(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvextrins_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvextrins_b(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvextrins_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvextrins.h(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvextrins_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvextrins_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvextrins_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvextrins.w(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvextrins_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvextrins_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvextrins_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvextrins.d(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvextrins_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvextrins_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvmskltz_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmskltz.b(<32 x i8> [[_1]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmskltz_b(v32i8 _1) { return __builtin_lasx_xvmskltz_b(_1); } // CHECK-LABEL: define dso_local void @xvmskltz_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmskltz.h(<16 x i16> [[_1]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmskltz_h(v16i16 _1) { return __builtin_lasx_xvmskltz_h(_1); } // CHECK-LABEL: define dso_local void @xvmskltz_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmskltz.w(<8 x i32> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmskltz_w(v8i32 _1) { return __builtin_lasx_xvmskltz_w(_1); } // CHECK-LABEL: define dso_local void @xvmskltz_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmskltz.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmskltz_d(v4i64 _1) { return __builtin_lasx_xvmskltz_d(_1); } // CHECK-LABEL: define dso_local void @xvsigncov_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsigncov.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsigncov_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsigncov_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsigncov_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsigncov.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsigncov_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsigncov_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsigncov_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsigncov.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsigncov_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsigncov_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsigncov_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsigncov.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsigncov_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsigncov_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfmadd_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfmadd.s(<8 x float> [[_1]], <8 x float> [[_2]], <8 x float> [[_3]]) -// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfmadd_s(v8f32 _1, v8f32 _2, v8f32 _3) { return __builtin_lasx_xvfmadd_s(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfmadd_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfmadd.d(<4 x double> [[_1]], <4 x double> [[_2]], <4 x double> [[_3]]) -// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfmadd_d(v4f64 _1, v4f64 _2, v4f64 _3) { return __builtin_lasx_xvfmadd_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfmsub_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfmsub.s(<8 x float> [[_1]], <8 x float> [[_2]], <8 x float> [[_3]]) -// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfmsub_s(v8f32 _1, v8f32 _2, v8f32 _3) { return __builtin_lasx_xvfmsub_s(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfmsub_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfmsub.d(<4 x double> [[_1]], <4 x double> [[_2]], <4 x double> [[_3]]) -// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfmsub_d(v4f64 _1, v4f64 _2, v4f64 _3) { return __builtin_lasx_xvfmsub_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfnmadd_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfnmadd.s(<8 x float> [[_1]], <8 x float> [[_2]], <8 x float> [[_3]]) -// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfnmadd_s(v8f32 _1, v8f32 _2, v8f32 _3) { return __builtin_lasx_xvfnmadd_s(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfnmadd_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfnmadd.d(<4 x double> [[_1]], <4 x double> [[_2]], <4 x double> [[_3]]) -// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfnmadd_d(v4f64 _1, v4f64 _2, v4f64 _3) { return __builtin_lasx_xvfnmadd_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfnmsub_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x float>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfnmsub.s(<8 x float> [[_1]], <8 x float> [[_2]], <8 x float> [[_3]]) -// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvfnmsub_s(v8f32 _1, v8f32 _2, v8f32 _3) { return __builtin_lasx_xvfnmsub_s(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvfnmsub_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x double>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfnmsub.d(<4 x double> [[_1]], <4 x double> [[_2]], <4 x double> [[_3]]) -// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvfnmsub_d(v4f64 _1, v4f64 _2, v4f64 _3) { return __builtin_lasx_xvfnmsub_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvftintrne_w_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrne.w.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrne_w_s(v8f32 _1) { return __builtin_lasx_xvftintrne_w_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrne_l_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrne.l.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrne_l_d(v4f64 _1) { return __builtin_lasx_xvftintrne_l_d(_1); } // CHECK-LABEL: define dso_local void @xvftintrp_w_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrp.w.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrp_w_s(v8f32 _1) { return __builtin_lasx_xvftintrp_w_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrp_l_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrp.l.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrp_l_d(v4f64 _1) { return __builtin_lasx_xvftintrp_l_d(_1); } // CHECK-LABEL: define dso_local void @xvftintrm_w_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrm.w.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrm_w_s(v8f32 _1) { return __builtin_lasx_xvftintrm_w_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrm_l_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrm.l.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrm_l_d(v4f64 _1) { return __builtin_lasx_xvftintrm_l_d(_1); } // CHECK-LABEL: define dso_local void @xvftint_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftint.w.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftint_w_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvftint_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvffint_s_l( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvffint.s.l(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvffint_s_l(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvffint_s_l(_1, _2); } // CHECK-LABEL: define dso_local void @xvftintrz_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrz.w.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrz_w_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvftintrz_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvftintrp_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrp.w.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrp_w_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvftintrp_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvftintrm_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrm.w.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrm_w_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvftintrm_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvftintrne_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvftintrne.w.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvftintrne_w_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvftintrne_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvftinth_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftinth.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftinth_l_s(v8f32 _1) { return __builtin_lasx_xvftinth_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintl_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintl.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintl_l_s(v8f32 _1) { return __builtin_lasx_xvftintl_l_s(_1); } // CHECK-LABEL: define dso_local void @xvffinth_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvffinth.d.w(<8 x i32> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvffinth_d_w(v8i32 _1) { return __builtin_lasx_xvffinth_d_w(_1); } // CHECK-LABEL: define dso_local void @xvffintl_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvffintl.d.w(<8 x i32> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvffintl_d_w(v8i32 _1) { return __builtin_lasx_xvffintl_d_w(_1); } // CHECK-LABEL: define dso_local void @xvftintrzh_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrzh.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrzh_l_s(v8f32 _1) { return __builtin_lasx_xvftintrzh_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrzl_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrzl.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrzl_l_s(v8f32 _1) { return __builtin_lasx_xvftintrzl_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrph_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrph.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrph_l_s(v8f32 _1) { return __builtin_lasx_xvftintrph_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrpl_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrpl.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrpl_l_s(v8f32 _1) { return __builtin_lasx_xvftintrpl_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrmh_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrmh.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrmh_l_s(v8f32 _1) { return __builtin_lasx_xvftintrmh_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrml_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrml.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrml_l_s(v8f32 _1) { return __builtin_lasx_xvftintrml_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrneh_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrneh.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrneh_l_s(v8f32 _1) { return __builtin_lasx_xvftintrneh_l_s(_1); } // CHECK-LABEL: define dso_local void @xvftintrnel_l_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvftintrnel.l.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvftintrnel_l_s(v8f32 _1) { return __builtin_lasx_xvftintrnel_l_s(_1); } // CHECK-LABEL: define dso_local void @xvfrintrne_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrintrne.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfrintrne_s(v8f32 _1) { return __builtin_lasx_xvfrintrne_s(_1); } // CHECK-LABEL: define dso_local void @xvfrintrne_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrintrne.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfrintrne_d(v4f64 _1) { return __builtin_lasx_xvfrintrne_d(_1); } // CHECK-LABEL: define dso_local void @xvfrintrz_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrintrz.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfrintrz_s(v8f32 _1) { return __builtin_lasx_xvfrintrz_s(_1); } // CHECK-LABEL: define dso_local void @xvfrintrz_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrintrz.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfrintrz_d(v4f64 _1) { return __builtin_lasx_xvfrintrz_d(_1); } // CHECK-LABEL: define dso_local void @xvfrintrp_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrintrp.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfrintrp_s(v8f32 _1) { return __builtin_lasx_xvfrintrp_s(_1); } // CHECK-LABEL: define dso_local void @xvfrintrp_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrintrp.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfrintrp_d(v4f64 _1) { return __builtin_lasx_xvfrintrp_d(_1); } // CHECK-LABEL: define dso_local void @xvfrintrm_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvfrintrm.s(<8 x float> [[_1]]) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfrintrm_s(v8f32 _1) { return __builtin_lasx_xvfrintrm_s(_1); } // CHECK-LABEL: define dso_local void @xvfrintrm_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvfrintrm.d(<4 x double> [[_1]]) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfrintrm_d(v4f64 _1) { return __builtin_lasx_xvfrintrm_d(_1); } @@ -4504,14 +4504,14 @@ v4i64 xvfrintrm_d(v4f64 _1) { return __builtin_lasx_xvfrintrm_d(_1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr noundef [[_1:%.*]]) local_unnamed_addr #[[ATTR3:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvld(ptr [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvld(void *_1) { return __builtin_lasx_xvld(_1, 1); } // CHECK-LABEL: define dso_local void @xvst( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr noundef [[_2:%.*]]) local_unnamed_addr #[[ATTR5:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: tail call void @llvm.loongarch.lasx.xvst(<32 x i8> [[_1]], ptr [[_2]], i32 1) // CHECK-NEXT: ret void // @@ -4519,7 +4519,7 @@ void xvst(v32i8 _1, void *_2) { return __builtin_lasx_xvst(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvstelm_b( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr noundef [[_2:%.*]]) local_unnamed_addr #[[ATTR5]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: tail call void @llvm.loongarch.lasx.xvstelm.b(<32 x i8> [[_1]], ptr [[_2]], i32 1, i32 1) // CHECK-NEXT: ret void // @@ -4527,7 +4527,7 @@ void xvstelm_b(v32i8 _1, void * _2) { return __builtin_lasx_xvstelm_b(_1, _2, 1, // CHECK-LABEL: define dso_local void @xvstelm_h( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr noundef [[_2:%.*]]) local_unnamed_addr #[[ATTR5]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: tail call void @llvm.loongarch.lasx.xvstelm.h(<16 x i16> [[_1]], ptr [[_2]], i32 2, i32 1) // CHECK-NEXT: ret void // @@ -4535,7 +4535,7 @@ void xvstelm_h(v16i16 _1, void * _2) { return __builtin_lasx_xvstelm_h(_1, _2, 2 // CHECK-LABEL: define dso_local void @xvstelm_w( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr noundef [[_2:%.*]]) local_unnamed_addr #[[ATTR5]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: tail call void @llvm.loongarch.lasx.xvstelm.w(<8 x i32> [[_1]], ptr [[_2]], i32 4, i32 1) // CHECK-NEXT: ret void // @@ -4543,7 +4543,7 @@ void xvstelm_w(v8i32 _1, void * _2) { return __builtin_lasx_xvstelm_w(_1, _2, 4, // CHECK-LABEL: define dso_local void @xvstelm_d( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr noundef [[_2:%.*]]) local_unnamed_addr #[[ATTR5]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: tail call void @llvm.loongarch.lasx.xvstelm.d(<4 x i64> [[_1]], ptr [[_2]], i32 8, i32 1) // CHECK-NEXT: ret void // @@ -4551,108 +4551,108 @@ void xvstelm_d(v4i64 _1, void * _2) { return __builtin_lasx_xvstelm_d(_1, _2, 8, // CHECK-LABEL: define dso_local void @xvinsve0_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvinsve0.w(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvinsve0_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvinsve0_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvinsve0_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvinsve0.d(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvinsve0_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvinsve0_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvpickve_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvpickve.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvpickve_w(v8i32 _1) { return __builtin_lasx_xvpickve_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvpickve_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvpickve.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvpickve_d(v4i64 _1) { return __builtin_lasx_xvpickve_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvssrlrn_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrlrn.b.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssrlrn_b_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvssrlrn_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrlrn_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrlrn.h.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssrlrn_h_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvssrlrn_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrlrn_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrlrn.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssrlrn_w_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvssrlrn_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrln_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrln.b.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssrln_b_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvssrln_b_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrln_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrln.h.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssrln_h_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvssrln_h_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvssrln_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrln.w.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssrln_w_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvssrln_w_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvorn_v( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvorn.v(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvorn_v(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvorn_v(_1, _2); } @@ -4660,7 +4660,7 @@ v32i8 xvorn_v(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvorn_v(_1, _2); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvldi(i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvldi() { return __builtin_lasx_xvldi(1); } @@ -4668,14 +4668,14 @@ v4i64 xvldi() { return __builtin_lasx_xvldi(1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr noundef [[_1:%.*]]) local_unnamed_addr #[[ATTR3]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvldx(ptr [[_1]], i64 1) -// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvldx(void *_1) { return __builtin_lasx_xvldx(_1, 1); } // CHECK-LABEL: define dso_local void @xvstx( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr noundef [[_2:%.*]]) local_unnamed_addr #[[ATTR5]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: tail call void @llvm.loongarch.lasx.xvstx(<32 x i8> [[_1]], ptr [[_2]], i64 1) // CHECK-NEXT: ret void // @@ -4683,209 +4683,209 @@ void xvstx(v32i8 _1, void *_2) { return __builtin_lasx_xvstx(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvextl_qu_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvextl.qu.du(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvextl_qu_du(v4u64 _1) { return __builtin_lasx_xvextl_qu_du(_1); } // CHECK-LABEL: define dso_local void @xvinsgr2vr_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvinsgr2vr.w(<8 x i32> [[_1]], i32 1, i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvinsgr2vr_w(v8i32 _1) { return __builtin_lasx_xvinsgr2vr_w(_1, 1, 1); } // CHECK-LABEL: define dso_local void @xvinsgr2vr_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvinsgr2vr.d(<4 x i64> [[_1]], i64 1, i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvinsgr2vr_d(v4i64 _1) { return __builtin_lasx_xvinsgr2vr_d(_1, 1, 1); } // CHECK-LABEL: define dso_local void @xvreplve0_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvreplve0.b(<32 x i8> [[_1]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvreplve0_b(v32i8 _1) { return __builtin_lasx_xvreplve0_b(_1); } // CHECK-LABEL: define dso_local void @xvreplve0_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvreplve0.h(<16 x i16> [[_1]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvreplve0_h(v16i16 _1) { return __builtin_lasx_xvreplve0_h(_1); } // CHECK-LABEL: define dso_local void @xvreplve0_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvreplve0.w(<8 x i32> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvreplve0_w(v8i32 _1) { return __builtin_lasx_xvreplve0_w(_1); } // CHECK-LABEL: define dso_local void @xvreplve0_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvreplve0.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvreplve0_d(v4i64 _1) { return __builtin_lasx_xvreplve0_d(_1); } // CHECK-LABEL: define dso_local void @xvreplve0_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvreplve0.q(<32 x i8> [[_1]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvreplve0_q(v32i8 _1) { return __builtin_lasx_xvreplve0_q(_1); } // CHECK-LABEL: define dso_local void @vext2xv_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.vext2xv.h.b(<32 x i8> [[_1]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 vext2xv_h_b(v32i8 _1) { return __builtin_lasx_vext2xv_h_b(_1); } // CHECK-LABEL: define dso_local void @vext2xv_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.vext2xv.w.h(<16 x i16> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 vext2xv_w_h(v16i16 _1) { return __builtin_lasx_vext2xv_w_h(_1); } // CHECK-LABEL: define dso_local void @vext2xv_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.vext2xv.d.w(<8 x i32> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 vext2xv_d_w(v8i32 _1) { return __builtin_lasx_vext2xv_d_w(_1); } // CHECK-LABEL: define dso_local void @vext2xv_w_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.vext2xv.w.b(<32 x i8> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 vext2xv_w_b(v32i8 _1) { return __builtin_lasx_vext2xv_w_b(_1); } // CHECK-LABEL: define dso_local void @vext2xv_d_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.vext2xv.d.h(<16 x i16> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 vext2xv_d_h(v16i16 _1) { return __builtin_lasx_vext2xv_d_h(_1); } // CHECK-LABEL: define dso_local void @vext2xv_d_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.vext2xv.d.b(<32 x i8> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 vext2xv_d_b(v32i8 _1) { return __builtin_lasx_vext2xv_d_b(_1); } // CHECK-LABEL: define dso_local void @vext2xv_hu_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.vext2xv.hu.bu(<32 x i8> [[_1]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 vext2xv_hu_bu(v32i8 _1) { return __builtin_lasx_vext2xv_hu_bu(_1); } // CHECK-LABEL: define dso_local void @vext2xv_wu_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.vext2xv.wu.hu(<16 x i16> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 vext2xv_wu_hu(v16i16 _1) { return __builtin_lasx_vext2xv_wu_hu(_1); } // CHECK-LABEL: define dso_local void @vext2xv_du_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.vext2xv.du.wu(<8 x i32> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 vext2xv_du_wu(v8i32 _1) { return __builtin_lasx_vext2xv_du_wu(_1); } // CHECK-LABEL: define dso_local void @vext2xv_wu_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.vext2xv.wu.bu(<32 x i8> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 vext2xv_wu_bu(v32i8 _1) { return __builtin_lasx_vext2xv_wu_bu(_1); } // CHECK-LABEL: define dso_local void @vext2xv_du_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.vext2xv.du.hu(<16 x i16> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 vext2xv_du_hu(v16i16 _1) { return __builtin_lasx_vext2xv_du_hu(_1); } // CHECK-LABEL: define dso_local void @vext2xv_du_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.vext2xv.du.bu(<32 x i8> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 vext2xv_du_bu(v32i8 _1) { return __builtin_lasx_vext2xv_du_bu(_1); } // CHECK-LABEL: define dso_local void @xvpermi_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvpermi.q(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvpermi_q(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvpermi_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvpermi_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvpermi.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvpermi_d(v4i64 _1) { return __builtin_lasx_xvpermi_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvperm_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvperm.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvperm_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvperm_w(_1, _2); } @@ -4893,7 +4893,7 @@ v8i32 xvperm_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvperm_w(_1, _2); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr noundef [[_1:%.*]]) local_unnamed_addr #[[ATTR3]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvldrepl.b(ptr [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvldrepl_b(void *_1) { return __builtin_lasx_xvldrepl_b(_1, 1); } @@ -4901,7 +4901,7 @@ v32i8 xvldrepl_b(void *_1) { return __builtin_lasx_xvldrepl_b(_1, 1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr noundef [[_1:%.*]]) local_unnamed_addr #[[ATTR3]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvldrepl.h(ptr [[_1]], i32 2) -// CHECK-NEXT: store <16 x i16> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvldrepl_h(void *_1) { return __builtin_lasx_xvldrepl_h(_1, 2); } @@ -4909,7 +4909,7 @@ v16i16 xvldrepl_h(void *_1) { return __builtin_lasx_xvldrepl_h(_1, 2); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr noundef [[_1:%.*]]) local_unnamed_addr #[[ATTR3]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvldrepl.w(ptr [[_1]], i32 4) -// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvldrepl_w(void *_1) { return __builtin_lasx_xvldrepl_w(_1, 4); } @@ -4917,14 +4917,14 @@ v8i32 xvldrepl_w(void *_1) { return __builtin_lasx_xvldrepl_w(_1, 4); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr noundef [[_1:%.*]]) local_unnamed_addr #[[ATTR3]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvldrepl.d(ptr [[_1]], i32 8) -// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvldrepl_d(void *_1) { return __builtin_lasx_xvldrepl_d(_1, 8); } // CHECK-LABEL: define dso_local signext i32 @xvpickve2gr_w( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xvpickve2gr.w(<8 x i32> [[_1]], i32 1) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -4932,7 +4932,7 @@ int xvpickve2gr_w(v8i32 _1) { return __builtin_lasx_xvpickve2gr_w(_1, 1); } // CHECK-LABEL: define dso_local signext i32 @xvpickve2gr_wu( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xvpickve2gr.wu(<8 x i32> [[_1]], i32 1) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -4940,7 +4940,7 @@ unsigned int xvpickve2gr_wu(v8i32 _1) { return __builtin_lasx_xvpickve2gr_wu(_1, // CHECK-LABEL: define dso_local i64 @xvpickve2gr_d( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.loongarch.lasx.xvpickve2gr.d(<4 x i64> [[_1]], i32 1) // CHECK-NEXT: ret i64 [[TMP1]] // @@ -4948,7 +4948,7 @@ long xvpickve2gr_d(v4i64 _1) { return __builtin_lasx_xvpickve2gr_d(_1, 1); } // CHECK-LABEL: define dso_local i64 @xvpickve2gr_du( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i64 @llvm.loongarch.lasx.xvpickve2gr.du(<4 x i64> [[_1]], i32 1) // CHECK-NEXT: ret i64 [[TMP1]] // @@ -4956,1626 +4956,1626 @@ unsigned long int xvpickve2gr_du(v4i64 _1) { return __builtin_lasx_xvpickve2gr_d // CHECK-LABEL: define dso_local void @xvaddwev_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwev.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwev_q_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvaddwev_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwev.d.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwev_d_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvaddwev_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvaddwev.w.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvaddwev_w_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvaddwev_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvaddwev.h.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvaddwev_h_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvaddwev_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwev.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwev_q_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvaddwev_q_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwev.d.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwev_d_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvaddwev_d_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvaddwev.w.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvaddwev_w_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvaddwev_w_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvaddwev.h.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvaddwev_h_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvaddwev_h_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwev.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwev_q_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsubwev_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwev.d.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwev_d_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsubwev_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsubwev.w.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsubwev_w_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsubwev_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsubwev.h.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsubwev_h_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsubwev_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwev.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwev_q_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvsubwev_q_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwev.d.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwev_d_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvsubwev_d_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsubwev.w.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsubwev_w_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvsubwev_w_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwev_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsubwev.h.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsubwev_h_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvsubwev_h_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwev.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwev_q_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvmulwev_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwev.d.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwev_d_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvmulwev_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmulwev.w.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmulwev_w_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvmulwev_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmulwev.h.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmulwev_h_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvmulwev_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwev.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwev_q_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvmulwev_q_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwev.d.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwev_d_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvmulwev_d_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmulwev.w.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmulwev_w_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvmulwev_w_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmulwev.h.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmulwev_h_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvmulwev_h_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwod.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwod_q_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvaddwod_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwod.d.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwod_d_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvaddwod_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvaddwod.w.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvaddwod_w_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvaddwod_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvaddwod.h.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvaddwod_h_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvaddwod_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwod.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwod_q_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvaddwod_q_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwod.d.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwod_d_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvaddwod_d_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvaddwod.w.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvaddwod_w_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvaddwod_w_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvaddwod.h.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvaddwod_h_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvaddwod_h_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwod.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwod_q_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsubwod_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwod.d.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwod_d_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsubwod_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsubwod.w.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsubwod_w_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsubwod_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsubwod.h.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsubwod_h_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsubwod_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwod.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwod_q_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvsubwod_q_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsubwod.d.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsubwod_d_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvsubwod_d_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsubwod.w.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsubwod_w_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvsubwod_w_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvsubwod_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsubwod.h.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsubwod_h_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvsubwod_h_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwod.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwod_q_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvmulwod_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwod.d.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwod_d_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvmulwod_d_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmulwod.w.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmulwod_w_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvmulwod_w_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmulwod.h.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmulwod_h_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvmulwod_h_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwod.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwod_q_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvmulwod_q_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwod.d.wu(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwod_d_wu(v8u32 _1, v8u32 _2) { return __builtin_lasx_xvmulwod_d_wu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmulwod.w.hu(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmulwod_w_hu(v16u16 _1, v16u16 _2) { return __builtin_lasx_xvmulwod_w_hu(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmulwod.h.bu(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmulwod_h_bu(v32u8 _1, v32u8 _2) { return __builtin_lasx_xvmulwod_h_bu(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_d_wu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwev.d.wu.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwev_d_wu_w(v8u32 _1, v8i32 _2) { return __builtin_lasx_xvaddwev_d_wu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_w_hu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvaddwev.w.hu.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvaddwev_w_hu_h(v16u16 _1, v16i16 _2) { return __builtin_lasx_xvaddwev_w_hu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_h_bu_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvaddwev.h.bu.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvaddwev_h_bu_b(v32u8 _1, v32i8 _2) { return __builtin_lasx_xvaddwev_h_bu_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_d_wu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwev.d.wu.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwev_d_wu_w(v8u32 _1, v8i32 _2) { return __builtin_lasx_xvmulwev_d_wu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_w_hu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmulwev.w.hu.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmulwev_w_hu_h(v16u16 _1, v16i16 _2) { return __builtin_lasx_xvmulwev_w_hu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_h_bu_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmulwev.h.bu.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmulwev_h_bu_b(v32u8 _1, v32i8 _2) { return __builtin_lasx_xvmulwev_h_bu_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_d_wu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwod.d.wu.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwod_d_wu_w(v8u32 _1, v8i32 _2) { return __builtin_lasx_xvaddwod_d_wu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_w_hu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvaddwod.w.hu.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvaddwod_w_hu_h(v16u16 _1, v16i16 _2) { return __builtin_lasx_xvaddwod_w_hu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_h_bu_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvaddwod.h.bu.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvaddwod_h_bu_b(v32u8 _1, v32i8 _2) { return __builtin_lasx_xvaddwod_h_bu_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_d_wu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwod.d.wu.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwod_d_wu_w(v8u32 _1, v8i32 _2) { return __builtin_lasx_xvmulwod_d_wu_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_w_hu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmulwod.w.hu.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmulwod_w_hu_h(v16u16 _1, v16i16 _2) { return __builtin_lasx_xvmulwod_w_hu_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_h_bu_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmulwod.h.bu.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmulwod_h_bu_b(v32u8 _1, v32i8 _2) { return __builtin_lasx_xvmulwod_h_bu_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhaddw.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvhaddw_q_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvhaddw_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvhaddw_qu_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhaddw.qu.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvhaddw_qu_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvhaddw_qu_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhsubw.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvhsubw_q_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvhsubw_q_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvhsubw_qu_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvhsubw.qu.du(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvhsubw_qu_du(v4u64 _1, v4u64 _2) { return __builtin_lasx_xvhsubw_qu_du(_1, _2); } // CHECK-LABEL: define dso_local void @xvmaddwev_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwev.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwev_q_d(v4i64 _1, v4i64 _2, v4i64 _3) { return __builtin_lasx_xvmaddwev_q_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwev.d.w(<4 x i64> [[_1]], <8 x i32> [[_2]], <8 x i32> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwev_d_w(v4i64 _1, v8i32 _2, v8i32 _3) { return __builtin_lasx_xvmaddwev_d_w(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaddwev.w.h(<8 x i32> [[_1]], <16 x i16> [[_2]], <16 x i16> [[_3]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmaddwev_w_h(v8i32 _1, v16i16 _2, v16i16 _3) { return __builtin_lasx_xvmaddwev_w_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaddwev.h.b(<16 x i16> [[_1]], <32 x i8> [[_2]], <32 x i8> [[_3]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmaddwev_h_b(v16i16 _1, v32i8 _2, v32i8 _3) { return __builtin_lasx_xvmaddwev_h_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwev.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmaddwev_q_du(v4u64 _1, v4u64 _2, v4u64 _3) { return __builtin_lasx_xvmaddwev_q_du(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwev.d.wu(<4 x i64> [[_1]], <8 x i32> [[_2]], <8 x i32> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmaddwev_d_wu(v4u64 _1, v8u32 _2, v8u32 _3) { return __builtin_lasx_xvmaddwev_d_wu(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaddwev.w.hu(<8 x i32> [[_1]], <16 x i16> [[_2]], <16 x i16> [[_3]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmaddwev_w_hu(v8u32 _1, v16u16 _2, v16u16 _3) { return __builtin_lasx_xvmaddwev_w_hu(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaddwev.h.bu(<16 x i16> [[_1]], <32 x i8> [[_2]], <32 x i8> [[_3]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmaddwev_h_bu(v16u16 _1, v32u8 _2, v32u8 _3) { return __builtin_lasx_xvmaddwev_h_bu(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwod.q.d(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwod_q_d(v4i64 _1, v4i64 _2, v4i64 _3) { return __builtin_lasx_xvmaddwod_q_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwod.d.w(<4 x i64> [[_1]], <8 x i32> [[_2]], <8 x i32> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwod_d_w(v4i64 _1, v8i32 _2, v8i32 _3) { return __builtin_lasx_xvmaddwod_d_w(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaddwod.w.h(<8 x i32> [[_1]], <16 x i16> [[_2]], <16 x i16> [[_3]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmaddwod_w_h(v8i32 _1, v16i16 _2, v16i16 _3) { return __builtin_lasx_xvmaddwod_w_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaddwod.h.b(<16 x i16> [[_1]], <32 x i8> [[_2]], <32 x i8> [[_3]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmaddwod_h_b(v16i16 _1, v32i8 _2, v32i8 _3) { return __builtin_lasx_xvmaddwod_h_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_q_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwod.q.du(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmaddwod_q_du(v4u64 _1, v4u64 _2, v4u64 _3) { return __builtin_lasx_xvmaddwod_q_du(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_d_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwod.d.wu(<4 x i64> [[_1]], <8 x i32> [[_2]], <8 x i32> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvmaddwod_d_wu(v4u64 _1, v8u32 _2, v8u32 _3) { return __builtin_lasx_xvmaddwod_d_wu(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_w_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaddwod.w.hu(<8 x i32> [[_1]], <16 x i16> [[_2]], <16 x i16> [[_3]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvmaddwod_w_hu(v8u32 _1, v16u16 _2, v16u16 _3) { return __builtin_lasx_xvmaddwod_w_hu(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_h_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaddwod.h.bu(<16 x i16> [[_1]], <32 x i8> [[_2]], <32 x i8> [[_3]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvmaddwod_h_bu(v16u16 _1, v32u8 _2, v32u8 _3) { return __builtin_lasx_xvmaddwod_h_bu(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_q_du_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwev.q.du.d(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwev_q_du_d(v4i64 _1, v4u64 _2, v4i64 _3) { return __builtin_lasx_xvmaddwev_q_du_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_d_wu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwev.d.wu.w(<4 x i64> [[_1]], <8 x i32> [[_2]], <8 x i32> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwev_d_wu_w(v4i64 _1, v8u32 _2, v8i32 _3) { return __builtin_lasx_xvmaddwev_d_wu_w(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_w_hu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaddwev.w.hu.h(<8 x i32> [[_1]], <16 x i16> [[_2]], <16 x i16> [[_3]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmaddwev_w_hu_h(v8i32 _1, v16u16 _2, v16i16 _3) { return __builtin_lasx_xvmaddwev_w_hu_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwev_h_bu_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaddwev.h.bu.b(<16 x i16> [[_1]], <32 x i8> [[_2]], <32 x i8> [[_3]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmaddwev_h_bu_b(v16i16 _1, v32u8 _2, v32i8 _3) { return __builtin_lasx_xvmaddwev_h_bu_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_q_du_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <4 x i64>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwod.q.du.d(<4 x i64> [[_1]], <4 x i64> [[_2]], <4 x i64> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwod_q_du_d(v4i64 _1, v4u64 _2, v4i64 _3) { return __builtin_lasx_xvmaddwod_q_du_d(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_d_wu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmaddwod.d.wu.w(<4 x i64> [[_1]], <8 x i32> [[_2]], <8 x i32> [[_3]]) -// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmaddwod_d_wu_w(v4i64 _1, v8u32 _2, v8i32 _3) { return __builtin_lasx_xvmaddwod_d_wu_w(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_w_hu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <16 x i16>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvmaddwod.w.hu.h(<8 x i32> [[_1]], <16 x i16> [[_2]], <16 x i16> [[_3]]) -// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvmaddwod_w_hu_h(v8i32 _1, v16u16 _2, v16i16 _3) { return __builtin_lasx_xvmaddwod_w_hu_h(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvmaddwod_h_bu_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP2:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_3:%.*]] = load <32 x i8>, ptr [[TMP2]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP3:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvmaddwod.h.bu.b(<16 x i16> [[_1]], <32 x i8> [[_2]], <32 x i8> [[_3]]) -// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP3]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvmaddwod_h_bu_b(v16i16 _1, v32u8 _2, v32i8 _3) { return __builtin_lasx_xvmaddwod_h_bu_b(_1, _2, _3); } // CHECK-LABEL: define dso_local void @xvrotr_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvrotr.b(<32 x i8> [[_1]], <32 x i8> [[_2]]) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvrotr_b(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvrotr_b(_1, _2); } // CHECK-LABEL: define dso_local void @xvrotr_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvrotr.h(<16 x i16> [[_1]], <16 x i16> [[_2]]) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvrotr_h(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvrotr_h(_1, _2); } // CHECK-LABEL: define dso_local void @xvrotr_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvrotr.w(<8 x i32> [[_1]], <8 x i32> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvrotr_w(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvrotr_w(_1, _2); } // CHECK-LABEL: define dso_local void @xvrotr_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvrotr.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvrotr_d(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvrotr_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvadd_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvadd.q(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvadd_q(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvadd_q(_1, _2); } // CHECK-LABEL: define dso_local void @xvsub_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsub.q(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsub_q(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsub_q(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwev_q_du_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwev.q.du.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwev_q_du_d(v4u64 _1, v4i64 _2) { return __builtin_lasx_xvaddwev_q_du_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvaddwod_q_du_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvaddwod.q.du.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvaddwod_q_du_d(v4u64 _1, v4i64 _2) { return __builtin_lasx_xvaddwod_q_du_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwev_q_du_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwev.q.du.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwev_q_du_d(v4u64 _1, v4i64 _2) { return __builtin_lasx_xvmulwev_q_du_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmulwod_q_du_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvmulwod.q.du.d(<4 x i64> [[_1]], <4 x i64> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvmulwod_q_du_d(v4u64 _1, v4i64 _2) { return __builtin_lasx_xvmulwod_q_du_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvmskgez_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmskgez.b(<32 x i8> [[_1]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmskgez_b(v32i8 _1) { return __builtin_lasx_xvmskgez_b(_1); } // CHECK-LABEL: define dso_local void @xvmsknz_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvmsknz.b(<32 x i8> [[_1]]) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvmsknz_b(v32i8 _1) { return __builtin_lasx_xvmsknz_b(_1); } // CHECK-LABEL: define dso_local void @xvexth_h_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvexth.h.b(<32 x i8> [[_1]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvexth_h_b(v32i8 _1) { return __builtin_lasx_xvexth_h_b(_1); } // CHECK-LABEL: define dso_local void @xvexth_w_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvexth.w.h(<16 x i16> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvexth_w_h(v16i16 _1) { return __builtin_lasx_xvexth_w_h(_1); } // CHECK-LABEL: define dso_local void @xvexth_d_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvexth.d.w(<8 x i32> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvexth_d_w(v8i32 _1) { return __builtin_lasx_xvexth_d_w(_1); } // CHECK-LABEL: define dso_local void @xvexth_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvexth.q.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvexth_q_d(v4i64 _1) { return __builtin_lasx_xvexth_q_d(_1); } // CHECK-LABEL: define dso_local void @xvexth_hu_bu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvexth.hu.bu(<32 x i8> [[_1]]) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvexth_hu_bu(v32u8 _1) { return __builtin_lasx_xvexth_hu_bu(_1); } // CHECK-LABEL: define dso_local void @xvexth_wu_hu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvexth.wu.hu(<16 x i16> [[_1]]) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvexth_wu_hu(v16u16 _1) { return __builtin_lasx_xvexth_wu_hu(_1); } // CHECK-LABEL: define dso_local void @xvexth_du_wu( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvexth.du.wu(<8 x i32> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvexth_du_wu(v8u32 _1) { return __builtin_lasx_xvexth_du_wu(_1); } // CHECK-LABEL: define dso_local void @xvexth_qu_du( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvexth.qu.du(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvexth_qu_du(v4u64 _1) { return __builtin_lasx_xvexth_qu_du(_1); } // CHECK-LABEL: define dso_local void @xvrotri_b( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvrotri.b(<32 x i8> [[_1]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvrotri_b(v32i8 _1) { return __builtin_lasx_xvrotri_b(_1, 1); } // CHECK-LABEL: define dso_local void @xvrotri_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvrotri.h(<16 x i16> [[_1]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvrotri_h(v16i16 _1) { return __builtin_lasx_xvrotri_h(_1, 1); } // CHECK-LABEL: define dso_local void @xvrotri_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvrotri.w(<8 x i32> [[_1]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvrotri_w(v8i32 _1) { return __builtin_lasx_xvrotri_w(_1, 1); } // CHECK-LABEL: define dso_local void @xvrotri_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvrotri.d(<4 x i64> [[_1]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvrotri_d(v4i64 _1) { return __builtin_lasx_xvrotri_d(_1, 1); } // CHECK-LABEL: define dso_local void @xvextl_q_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvextl.q.d(<4 x i64> [[_1]]) -// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvextl_q_d(v4i64 _1) { return __builtin_lasx_xvextl_q_d(_1); } // CHECK-LABEL: define dso_local void @xvsrlni_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrlni.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrlni_b_h(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsrlni_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrlni_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrlni.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrlni_h_w(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsrlni_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrlni_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrlni.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrlni_w_d(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsrlni_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrlni_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrlni.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrlni_d_q(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsrlni_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrlrni_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrlrni.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrlrni_b_h(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsrlrni_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrlrni_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrlrni.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrlrni_h_w(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsrlrni_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrlrni_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrlrni.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrlrni_w_d(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsrlrni_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrlrni_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrlrni.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrlrni_d_q(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsrlrni_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrlni.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssrlni_b_h(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvssrlni_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrlni.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssrlni_h_w(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvssrlni_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrlni.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssrlni_w_d(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvssrlni_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrlni.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvssrlni_d_q(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvssrlni_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrlni.bu.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssrlni_bu_h(v32u8 _1, v32i8 _2) { return __builtin_lasx_xvssrlni_bu_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrlni.hu.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssrlni_hu_w(v16u16 _1, v16i16 _2) { return __builtin_lasx_xvssrlni_hu_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrlni.wu.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssrlni_wu_d(v8u32 _1, v8i32 _2) { return __builtin_lasx_xvssrlni_wu_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlni_du_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrlni.du.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvssrlni_du_q(v4u64 _1, v4i64 _2) { return __builtin_lasx_xvssrlni_du_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrlrni.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssrlrni_b_h(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvssrlrni_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrlrni.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssrlrni_h_w(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvssrlrni_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrlrni.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssrlrni_w_d(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvssrlrni_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrlrni.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvssrlrni_d_q(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvssrlrni_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrlrni.bu.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssrlrni_bu_h(v32u8 _1, v32i8 _2) { return __builtin_lasx_xvssrlrni_bu_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrlrni.hu.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssrlrni_hu_w(v16u16 _1, v16i16 _2) { return __builtin_lasx_xvssrlrni_hu_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrlrni.wu.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssrlrni_wu_d(v8u32 _1, v8i32 _2) { return __builtin_lasx_xvssrlrni_wu_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrlrni_du_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrlrni.du.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvssrlrni_du_q(v4u64 _1, v4i64 _2) { return __builtin_lasx_xvssrlrni_du_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrani_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrani.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrani_b_h(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsrani_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrani_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrani.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrani_h_w(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsrani_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrani_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrani.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrani_w_d(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsrani_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrani_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrani.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrani_d_q(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsrani_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrarni_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvsrarni.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvsrarni_b_h(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvsrarni_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrarni_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvsrarni.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvsrarni_h_w(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvsrarni_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrarni_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvsrarni.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvsrarni_w_d(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvsrarni_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvsrarni_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvsrarni.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvsrarni_d_q(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvsrarni_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrani.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssrani_b_h(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvssrani_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrani.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssrani_h_w(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvssrani_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrani.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssrani_w_d(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvssrani_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrani.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvssrani_d_q(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvssrani_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrani.bu.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssrani_bu_h(v32u8 _1, v32i8 _2) { return __builtin_lasx_xvssrani_bu_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrani.hu.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssrani_hu_w(v16u16 _1, v16i16 _2) { return __builtin_lasx_xvssrani_hu_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrani.wu.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssrani_wu_d(v8u32 _1, v8i32 _2) { return __builtin_lasx_xvssrani_wu_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrani_du_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrani.du.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvssrani_du_q(v4u64 _1, v4i64 _2) { return __builtin_lasx_xvssrani_du_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_b_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrarni.b.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvssrarni_b_h(v32i8 _1, v32i8 _2) { return __builtin_lasx_xvssrarni_b_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_h_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrarni.h.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvssrarni_h_w(v16i16 _1, v16i16 _2) { return __builtin_lasx_xvssrarni_h_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_w_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrarni.w.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvssrarni_w_d(v8i32 _1, v8i32 _2) { return __builtin_lasx_xvssrarni_w_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_d_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrarni.d.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvssrarni_d_q(v4i64 _1, v4i64 _2) { return __builtin_lasx_xvssrarni_d_q(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_bu_h( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <32 x i8>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvssrarni.bu.h(<32 x i8> [[_1]], <32 x i8> [[_2]], i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32u8 xvssrarni_bu_h(v32u8 _1, v32i8 _2) { return __builtin_lasx_xvssrarni_bu_h(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_hu_w( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <16 x i16>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvssrarni.hu.w(<16 x i16> [[_1]], <16 x i16> [[_2]], i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16u16 xvssrarni_hu_w(v16u16 _1, v16i16 _2) { return __builtin_lasx_xvssrarni_hu_w(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_wu_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvssrarni.wu.d(<8 x i32> [[_1]], <8 x i32> [[_2]], i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8u32 xvssrarni_wu_d(v8u32 _1, v8i32 _2) { return __builtin_lasx_xvssrarni_wu_d(_1, _2, 1); } // CHECK-LABEL: define dso_local void @xvssrarni_du_q( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvssrarni.du.q(<4 x i64> [[_1]], <4 x i64> [[_2]], i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4u64 xvssrarni_du_q(v4u64 _1, v4i64 _2) { return __builtin_lasx_xvssrarni_du_q(_1, _2, 1); } // CHECK-LABEL: define dso_local signext i32 @xbnz_b( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbnz.b(<32 x i8> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6583,7 +6583,7 @@ int xbnz_b(v32u8 _1) { return __builtin_lasx_xbnz_b(_1); } // CHECK-LABEL: define dso_local signext i32 @xbnz_d( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbnz.d(<4 x i64> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6591,7 +6591,7 @@ int xbnz_d(v4u64 _1) { return __builtin_lasx_xbnz_d(_1); } // CHECK-LABEL: define dso_local signext i32 @xbnz_h( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbnz.h(<16 x i16> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6599,7 +6599,7 @@ int xbnz_h(v16u16 _1) { return __builtin_lasx_xbnz_h(_1); } // CHECK-LABEL: define dso_local signext i32 @xbnz_v( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbnz.v(<32 x i8> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6607,7 +6607,7 @@ int xbnz_v(v32u8 _1) { return __builtin_lasx_xbnz_v(_1); } // CHECK-LABEL: define dso_local signext i32 @xbnz_w( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbnz.w(<8 x i32> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6615,7 +6615,7 @@ int xbnz_w(v8u32 _1) { return __builtin_lasx_xbnz_w(_1); } // CHECK-LABEL: define dso_local signext i32 @xbz_b( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbz.b(<32 x i8> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6623,7 +6623,7 @@ int xbz_b(v32u8 _1) { return __builtin_lasx_xbz_b(_1); } // CHECK-LABEL: define dso_local signext i32 @xbz_d( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x i64>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbz.d(<4 x i64> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6631,7 +6631,7 @@ int xbz_d(v4u64 _1) { return __builtin_lasx_xbz_d(_1); } // CHECK-LABEL: define dso_local signext i32 @xbz_h( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <16 x i16>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbz.h(<16 x i16> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6639,7 +6639,7 @@ int xbz_h(v16u16 _1) { return __builtin_lasx_xbz_h(_1); } // CHECK-LABEL: define dso_local signext i32 @xbz_v( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <32 x i8>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbz.v(<32 x i8> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6647,7 +6647,7 @@ int xbz_v(v32u8 _1) { return __builtin_lasx_xbz_v(_1); } // CHECK-LABEL: define dso_local signext i32 @xbz_w( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR7]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call i32 @llvm.loongarch.lasx.xbz.w(<8 x i32> [[_1]]) // CHECK-NEXT: ret i32 [[TMP1]] // @@ -6655,458 +6655,458 @@ int xbz_w(v8u32 _1) { return __builtin_lasx_xbz_w(_1); } // CHECK-LABEL: define dso_local void @xvfcmp_caf_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.caf.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_caf_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_caf_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_caf_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.caf.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_caf_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_caf_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_ceq_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.ceq.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_ceq_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_ceq_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_ceq_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.ceq.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_ceq_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_ceq_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cle_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cle.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cle_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_cle_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cle_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cle.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cle_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_cle_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_clt_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.clt.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_clt_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_clt_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_clt_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.clt.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_clt_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_clt_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cne_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cne.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cne_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_cne_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cne_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cne.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cne_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_cne_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cor_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cor.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cor_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_cor_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cor_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cor.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cor_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_cor_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cueq_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cueq.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cueq_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_cueq_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cueq_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cueq.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cueq_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_cueq_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cule_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cule.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cule_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_cule_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cule_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cule.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cule_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_cule_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cult_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cult.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cult_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_cult_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cult_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cult.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cult_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_cult_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cun_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cun.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cun_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_cun_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cune_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.cune.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_cune_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_cune_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cune_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cune.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cune_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_cune_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_cun_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.cun.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_cun_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_cun_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_saf_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.saf.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_saf_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_saf_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_saf_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.saf.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_saf_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_saf_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_seq_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.seq.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_seq_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_seq_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_seq_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.seq.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_seq_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_seq_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sle_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sle.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sle_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_sle_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sle_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sle.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sle_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_sle_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_slt_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.slt.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_slt_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_slt_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_slt_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.slt.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_slt_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_slt_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sne_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sne.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sne_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_sne_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sne_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sne.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sne_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_sne_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sor_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sor.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sor_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_sor_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sor_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sor.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sor_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_sor_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sueq_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sueq.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sueq_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_sueq_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sueq_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sueq.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sueq_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_sueq_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sule_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sule.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sule_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_sule_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sule_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sule.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sule_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_sule_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sult_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sult.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sult_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_sult_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sult_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sult.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sult_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_sult_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sun_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sun.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sun_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_sun_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sune_d( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <4 x double>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvfcmp.sune.d(<4 x double> [[_1]], <4 x double> [[_2]]) -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvfcmp_sune_d(v4f64 _1, v4f64 _2) { return __builtin_lasx_xvfcmp_sune_d(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sune_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sune.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sune_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_sune_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvfcmp_sun_s( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[_2:%.*]] = load <8 x float>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvfcmp.sun.s(<8 x float> [[_1]], <8 x float> [[_2]]) -// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP2]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvfcmp_sun_s(v8f32 _1, v8f32 _2) { return __builtin_lasx_xvfcmp_sun_s(_1, _2); } // CHECK-LABEL: define dso_local void @xvpickve_d_f( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x double>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <4 x double>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x double> @llvm.loongarch.lasx.xvpickve.d.f(<4 x double> [[_1]], i32 1) -// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x double> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4f64 xvpickve_d_f(v4f64 _1) { return __builtin_lasx_xvpickve_d_f(_1, 1); } // CHECK-LABEL: define dso_local void @xvpickve_w_f( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x float>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[_1:%.*]] = load <8 x float>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.loongarch.lasx.xvpickve.w.f(<8 x float> [[_1]], i32 1) -// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x float> [[TMP1]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8f32 xvpickve_w_f(v8f32 _1) { return __builtin_lasx_xvpickve_w_f(_1, 1); } @@ -7114,7 +7114,7 @@ v8f32 xvpickve_w_f(v8f32 _1) { return __builtin_lasx_xvpickve_w_f(_1, 1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<32 x i8>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <32 x i8> @llvm.loongarch.lasx.xvrepli.b(i32 1) -// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <32 x i8> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v32i8 xvrepli_b() { return __builtin_lasx_xvrepli_b(1); } @@ -7122,7 +7122,7 @@ v32i8 xvrepli_b() { return __builtin_lasx_xvrepli_b(1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <4 x i64> @llvm.loongarch.lasx.xvrepli.d(i32 1) -// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v4i64 xvrepli_d() { return __builtin_lasx_xvrepli_d(1); } @@ -7130,7 +7130,7 @@ v4i64 xvrepli_d() { return __builtin_lasx_xvrepli_d(1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i16>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <16 x i16> @llvm.loongarch.lasx.xvrepli.h(i32 1) -// CHECK-NEXT: store <16 x i16> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i16> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v16i16 xvrepli_h() { return __builtin_lasx_xvrepli_h(1); } @@ -7138,12 +7138,12 @@ v16i16 xvrepli_h() { return __builtin_lasx_xvrepli_h(1); } // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.loongarch.lasx.xvrepli.w(i32 1) -// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x i32> [[TMP0]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // v8i32 xvrepli_w() { return __builtin_lasx_xvrepli_w(1); } //. -// CHECK: [[CHAR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[CHAR_TBAA6]] = !{[[META4]], [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/LoongArch/lasx/inline-asm-gcc-regs.c b/clang/test/CodeGen/LoongArch/lasx/inline-asm-gcc-regs.c index ed1a966..0dc74ff 100644 --- a/clang/test/CodeGen/LoongArch/lasx/inline-asm-gcc-regs.c +++ b/clang/test/CodeGen/LoongArch/lasx/inline-asm-gcc-regs.c @@ -4,7 +4,7 @@ typedef signed char v32i8 __attribute__((vector_size(32), aligned(32))); // CHECK-LABEL: @test_xr0( -// CHECK: tail call void asm sideeffect "", "{$xr0}"(<32 x i8> undef) #[[ATTR1:[0-9]+]], !srcloc !2 +// CHECK: tail call void asm sideeffect "", "{$xr0}"(<32 x i8> undef) #[[ATTR1:[0-9]+]], !srcloc [[META6:![0-9]+]] // void test_xr0() { register v32i8 a asm ("$xr0"); @@ -12,7 +12,7 @@ void test_xr0() { } // CHECK-LABEL: @test_xr7( -// CHECK: tail call void asm sideeffect "", "{$xr7}"(<32 x i8> undef) #[[ATTR1]], !srcloc !3 +// CHECK: tail call void asm sideeffect "", "{$xr7}"(<32 x i8> undef) #[[ATTR1]], !srcloc [[META7:![0-9]+]] // void test_xr7() { register v32i8 a asm ("$xr7"); @@ -20,7 +20,7 @@ void test_xr7() { } // CHECK-LABEL: @test_xr15( -// CHECK: tail call void asm sideeffect "", "{$xr15}"(<32 x i8> undef) #[[ATTR1]], !srcloc !4 +// CHECK: tail call void asm sideeffect "", "{$xr15}"(<32 x i8> undef) #[[ATTR1]], !srcloc [[META8:![0-9]+]] // void test_xr15() { register v32i8 a asm ("$xr15"); @@ -28,7 +28,7 @@ void test_xr15() { } // CHECK-LABEL: @test_xr31( -// CHECK: tail call void asm sideeffect "", "{$xr31}"(<32 x i8> undef) #[[ATTR1]], !srcloc !5 +// CHECK: tail call void asm sideeffect "", "{$xr31}"(<32 x i8> undef) #[[ATTR1]], !srcloc [[META9:![0-9]+]] // void test_xr31() { register v32i8 a asm ("$xr31"); diff --git a/clang/test/CodeGen/LoongArch/lasx/inline-asm-operand-modifier.c b/clang/test/CodeGen/LoongArch/lasx/inline-asm-operand-modifier.c index a5cc879..cb5e689 100644 --- a/clang/test/CodeGen/LoongArch/lasx/inline-asm-operand-modifier.c +++ b/clang/test/CodeGen/LoongArch/lasx/inline-asm-operand-modifier.c @@ -6,7 +6,7 @@ typedef long long v4i64 __attribute__ ((vector_size(32), aligned(32))); // CHECK-LABEL: define dso_local void @test_u // CHECK-SAME: () local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: entry: -// CHECK-NEXT: [[TMP0:%.*]] = tail call <4 x i64> asm sideeffect "xvldi ${0:u}, 1", "=f"() #[[ATTR1:[0-9]+]], !srcloc !2 +// CHECK-NEXT: [[TMP0:%.*]] = tail call <4 x i64> asm sideeffect "xvldi ${0:u}, 1", "=f"() #[[ATTR1:[0-9]+]], !srcloc [[META6:![0-9]+]] // CHECK-NEXT: ret void // void test_u() { diff --git a/clang/test/CodeGen/LoongArch/lsx/inline-asm-gcc-regs.c b/clang/test/CodeGen/LoongArch/lsx/inline-asm-gcc-regs.c index b05b1c8..588a3a1 100644 --- a/clang/test/CodeGen/LoongArch/lsx/inline-asm-gcc-regs.c +++ b/clang/test/CodeGen/LoongArch/lsx/inline-asm-gcc-regs.c @@ -4,7 +4,7 @@ typedef signed char v16i8 __attribute__((vector_size(16), aligned(16))); // CHECK-LABEL: @test_vr0( -// CHECK: tail call void asm sideeffect "", "{$vr0}"(<16 x i8> undef) #[[ATTR1:[0-9]+]], !srcloc !2 +// CHECK: tail call void asm sideeffect "", "{$vr0}"(<16 x i8> undef) #[[ATTR1:[0-9]+]], !srcloc [[META6:![0-9]+]] // void test_vr0() { register v16i8 a asm ("$vr0"); @@ -12,7 +12,7 @@ void test_vr0() { } // CHECK-LABEL: @test_vr7( -// CHECK: tail call void asm sideeffect "", "{$vr7}"(<16 x i8> undef) #[[ATTR1]], !srcloc !3 +// CHECK: tail call void asm sideeffect "", "{$vr7}"(<16 x i8> undef) #[[ATTR1]], !srcloc [[META7:![0-9]+]] // void test_vr7() { register v16i8 a asm ("$vr7"); @@ -20,7 +20,7 @@ void test_vr7() { } // CHECK-LABEL: @test_vr15( -// CHECK: tail call void asm sideeffect "", "{$vr15}"(<16 x i8> undef) #[[ATTR1]], !srcloc !4 +// CHECK: tail call void asm sideeffect "", "{$vr15}"(<16 x i8> undef) #[[ATTR1]], !srcloc [[META8:![0-9]+]] // void test_vr15() { register v16i8 a asm ("$vr15"); @@ -28,7 +28,7 @@ void test_vr15() { } // CHECK-LABEL: @test_vr31( -// CHECK: tail call void asm sideeffect "", "{$vr31}"(<16 x i8> undef) #[[ATTR1]], !srcloc !5 +// CHECK: tail call void asm sideeffect "", "{$vr31}"(<16 x i8> undef) #[[ATTR1]], !srcloc [[META9:![0-9]+]] // void test_vr31() { register v16i8 a asm ("$vr31"); diff --git a/clang/test/CodeGen/LoongArch/lsx/inline-asm-operand-modifier.c b/clang/test/CodeGen/LoongArch/lsx/inline-asm-operand-modifier.c index 5e0fae9..f0fb6e3 100644 --- a/clang/test/CodeGen/LoongArch/lsx/inline-asm-operand-modifier.c +++ b/clang/test/CodeGen/LoongArch/lsx/inline-asm-operand-modifier.c @@ -6,7 +6,7 @@ typedef long long v2i64 __attribute__ ((vector_size(16), aligned(16))); // CHECK-LABEL: define dso_local void @test_w // CHECK-SAME: () local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: entry: -// CHECK-NEXT: [[TMP0:%.*]] = tail call <2 x i64> asm sideeffect "vldi ${0:w}, 1", "=f"() #[[ATTR1:[0-9]+]], !srcloc !2 +// CHECK-NEXT: [[TMP0:%.*]] = tail call <2 x i64> asm sideeffect "vldi ${0:w}, 1", "=f"() #[[ATTR1:[0-9]+]], !srcloc [[META6:![0-9]+]] // CHECK-NEXT: ret void // void test_w() { diff --git a/clang/test/CodeGen/PowerPC/builtins-dmf-vsx-vector-float.c b/clang/test/CodeGen/PowerPC/builtins-dmf-vsx-vector-float.c index 8fc9a68..f59a964 100644 --- a/clang/test/CodeGen/PowerPC/builtins-dmf-vsx-vector-float.c +++ b/clang/test/CodeGen/PowerPC/builtins-dmf-vsx-vector-float.c @@ -1,17 +1,26 @@ -// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 6 // Update then manual applied to commonize the checks for AIX and LoP. // RUN: %clang_cc1 -O3 -triple powerpc64le-unknown-unknown -target-cpu future \ // RUN: -emit-llvm %s -o - | FileCheck %s // RUN: %clang_cc1 -O3 -triple powerpc64-ibm-aix -target-cpu future \ -// RUN: -emit-llvm %s -o - | FileCheck %s +// RUN: -emit-llvm %s -o - | FileCheck %s --check-prefix=AIX -// CHECK-LABEL: void @test_dmxvbf16gerx2( +// CHECK-LABEL: define dso_local void @test_dmxvbf16gerx2( +// CHECK-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2:![0-9]+]] -// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6:![0-9]+]] +// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8:![0-9]+]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvbf16gerx2( +// AIX-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6:![0-9]+]] +// AIX-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8:![0-9]+]] +// AIX-NEXT: ret void +// void test_dmxvbf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -19,14 +28,24 @@ void test_dmxvbf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigne *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvbf16gerx2nn( +// CHECK-LABEL: define dso_local void @test_dmxvbf16gerx2nn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvbf16gerx2nn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvbf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -34,14 +53,24 @@ void test_dmxvbf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsig *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvbf16gerx2np( +// CHECK-LABEL: define dso_local void @test_dmxvbf16gerx2np( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvbf16gerx2np( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvbf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -49,14 +78,24 @@ void test_dmxvbf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsig *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvbf16gerx2pn( +// CHECK-LABEL: define dso_local void @test_dmxvbf16gerx2pn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvbf16gerx2pn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvbf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -64,14 +103,24 @@ void test_dmxvbf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsig *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvbf16gerx2pp( +// CHECK-LABEL: define dso_local void @test_dmxvbf16gerx2pp( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvbf16gerx2pp( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvbf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -79,13 +128,22 @@ void test_dmxvbf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsig *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvbf16gerx2( +// CHECK-LABEL: define dso_local void @test_pmdmxvbf16gerx2( +// CHECK-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvbf16gerx2( +// AIX-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvbf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -93,14 +151,24 @@ void test_pmdmxvbf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsig *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvbf16gerx2nn( +// CHECK-LABEL: define dso_local void @test_pmdmxvbf16gerx2nn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvbf16gerx2nn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvbf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -108,14 +176,24 @@ void test_pmdmxvbf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector uns *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvbf16gerx2np( +// CHECK-LABEL: define dso_local void @test_pmdmxvbf16gerx2np( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvbf16gerx2np( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvbf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -123,14 +201,24 @@ void test_pmdmxvbf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector uns *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvbf16gerx2pn( +// CHECK-LABEL: define dso_local void @test_pmdmxvbf16gerx2pn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvbf16gerx2pn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvbf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -138,14 +226,24 @@ void test_pmdmxvbf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector uns *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvbf16gerx2pp( +// CHECK-LABEL: define dso_local void @test_pmdmxvbf16gerx2pp( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvbf16gerx2pp( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvbf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -153,13 +251,22 @@ void test_pmdmxvbf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector uns *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvf16gerx2( +// CHECK-LABEL: define dso_local void @test_dmxvf16gerx2( +// CHECK-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2:![0-9]+]] -// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvf16gerx2( +// AIX-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -167,14 +274,24 @@ void test_dmxvf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigned *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvf16gerx2nn( +// CHECK-LABEL: define dso_local void @test_dmxvf16gerx2nn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvf16gerx2nn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -182,14 +299,24 @@ void test_dmxvf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsign *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvf16gerx2np( +// CHECK-LABEL: define dso_local void @test_dmxvf16gerx2np( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvf16gerx2np( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -197,14 +324,24 @@ void test_dmxvf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsign *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvf16gerx2pn( +// CHECK-LABEL: define dso_local void @test_dmxvf16gerx2pn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvf16gerx2pn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -212,14 +349,24 @@ void test_dmxvf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsign *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_dmxvf16gerx2pp( +// CHECK-LABEL: define dso_local void @test_dmxvf16gerx2pp( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_dmxvf16gerx2pp( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_dmxvf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -227,13 +374,22 @@ void test_dmxvf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsign *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvf16gerx2( +// CHECK-LABEL: define dso_local void @test_pmdmxvf16gerx2( +// CHECK-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvf16gerx2( +// AIX-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -241,14 +397,24 @@ void test_pmdmxvf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsign *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvf16gerx2nn( +// CHECK-LABEL: define dso_local void @test_pmdmxvf16gerx2nn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvf16gerx2nn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -256,14 +422,24 @@ void test_pmdmxvf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsi *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvf16gerx2np( +// CHECK-LABEL: define dso_local void @test_pmdmxvf16gerx2np( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvf16gerx2np( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -271,14 +447,24 @@ void test_pmdmxvf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsi *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvf16gerx2pn( +// CHECK-LABEL: define dso_local void @test_pmdmxvf16gerx2pn( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvf16gerx2pn( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -286,14 +472,24 @@ void test_pmdmxvf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsi *((__dmr1024 *)resp) = vdmr; } -// CHECK-LABEL: void @test_pmdmxvf16gerx2pp( +// CHECK-LABEL: define dso_local void @test_pmdmxvf16gerx2pp( +// CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // +// AIX-LABEL: define void @test_pmdmxvf16gerx2pp( +// AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { +// AIX-NEXT: [[ENTRY:.*:]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: ret void +// void test_pmdmxvf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { __dmr1024 vdmr = *((__dmr1024 *)vdmrp); __vector_pair vp = *((__vector_pair *)vpp); @@ -301,9 +497,18 @@ void test_pmdmxvf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsi *((__dmr1024 *)resp) = vdmr; } -// CHECK: [[TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"__vector_pair", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +//. +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} -// CHECK: [[TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK: [[META7]] = !{!"__dmr1024", [[META4]], i64 0} +// CHECK: [[__VECTOR_PAIR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK: [[META7]] = !{!"__vector_pair", [[META4]], i64 0} +// CHECK: [[__DMR1024_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} +// CHECK: [[META9]] = !{!"__dmr1024", [[META4]], i64 0} +//. +// AIX: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// AIX: [[META5]] = !{!"Simple C/C++ TBAA"} +// AIX: [[__VECTOR_PAIR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// AIX: [[META7]] = !{!"__vector_pair", [[META4]], i64 0} +// AIX: [[__DMR1024_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} +// AIX: [[META9]] = !{!"__dmr1024", [[META4]], i64 0} +//. diff --git a/clang/test/CodeGen/PowerPC/builtins-ppc-build-pair-mma.c b/clang/test/CodeGen/PowerPC/builtins-ppc-build-pair-mma.c index 59b71cd..e602be7c 100644 --- a/clang/test/CodeGen/PowerPC/builtins-ppc-build-pair-mma.c +++ b/clang/test/CodeGen/PowerPC/builtins-ppc-build-pair-mma.c @@ -10,14 +10,14 @@ // CHECK-LE-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC1:%.*]], <16 x i8> noundef [[VC2:%.*]], <16 x i8> noundef [[VC3:%.*]], <16 x i8> noundef [[VC4:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-LE-NEXT: [[ENTRY:.*:]] // CHECK-LE-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.assemble.acc(<16 x i8> [[VC4]], <16 x i8> [[VC3]], <16 x i8> [[VC2]], <16 x i8> [[VC1]]) -// CHECK-LE-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2:![0-9]+]] +// CHECK-LE-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6:![0-9]+]] // CHECK-LE-NEXT: ret void // // CHECK-BE-LABEL: define dso_local void @test1( // CHECK-BE-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC1:%.*]], <16 x i8> noundef [[VC2:%.*]], <16 x i8> noundef [[VC3:%.*]], <16 x i8> noundef [[VC4:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-BE-NEXT: [[ENTRY:.*:]] // CHECK-BE-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.assemble.acc(<16 x i8> [[VC1]], <16 x i8> [[VC2]], <16 x i8> [[VC3]], <16 x i8> [[VC4]]) -// CHECK-BE-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2:![0-9]+]] +// CHECK-BE-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6:![0-9]+]] // CHECK-BE-NEXT: ret void // // CHECK-LE-NOOPT-LABEL: define dso_local void @test1( @@ -70,14 +70,14 @@ void test1(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc1, vec // CHECK-LE-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC1:%.*]], <16 x i8> noundef [[VC2:%.*]], ptr noundef writeonly captures(none) initializes((0, 32)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-LE-NEXT: [[ENTRY:.*:]] // CHECK-LE-NEXT: [[TMP0:%.*]] = tail call <256 x i1> @llvm.ppc.vsx.assemble.pair(<16 x i8> [[VC2]], <16 x i8> [[VC1]]) -// CHECK-LE-NEXT: store <256 x i1> [[TMP0]], ptr [[RESP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6:![0-9]+]] +// CHECK-LE-NEXT: store <256 x i1> [[TMP0]], ptr [[RESP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA8:![0-9]+]] // CHECK-LE-NEXT: ret void // // CHECK-BE-LABEL: define dso_local void @test2( // CHECK-BE-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC1:%.*]], <16 x i8> noundef [[VC2:%.*]], ptr noundef writeonly captures(none) initializes((0, 32)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-BE-NEXT: [[ENTRY:.*:]] // CHECK-BE-NEXT: [[TMP0:%.*]] = tail call <256 x i1> @llvm.ppc.vsx.assemble.pair(<16 x i8> [[VC1]], <16 x i8> [[VC2]]) -// CHECK-BE-NEXT: store <256 x i1> [[TMP0]], ptr [[RESP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6:![0-9]+]] +// CHECK-BE-NEXT: store <256 x i1> [[TMP0]], ptr [[RESP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA8:![0-9]+]] // CHECK-BE-NEXT: ret void // // CHECK-LE-NOOPT-LABEL: define dso_local void @test2( @@ -120,17 +120,17 @@ void test2(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc1, *((__vector_pair *)resp) = res; } //. -// CHECK-LE: [[__VECTOR_QUAD_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK-LE: [[META3]] = !{!"__vector_quad", [[META4:![0-9]+]], i64 0} -// CHECK-LE: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK-LE: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK-LE: [[META5]] = !{!"Simple C/C++ TBAA"} -// CHECK-LE: [[__VECTOR_PAIR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK-LE: [[META7]] = !{!"__vector_pair", [[META4]], i64 0} +// CHECK-LE: [[__VECTOR_QUAD_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK-LE: [[META7]] = !{!"__vector_quad", [[META4]], i64 0} +// CHECK-LE: [[__VECTOR_PAIR_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} +// CHECK-LE: [[META9]] = !{!"__vector_pair", [[META4]], i64 0} //. -// CHECK-BE: [[__VECTOR_QUAD_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK-BE: [[META3]] = !{!"__vector_quad", [[META4:![0-9]+]], i64 0} -// CHECK-BE: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK-BE: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK-BE: [[META5]] = !{!"Simple C/C++ TBAA"} -// CHECK-BE: [[__VECTOR_PAIR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK-BE: [[META7]] = !{!"__vector_pair", [[META4]], i64 0} +// CHECK-BE: [[__VECTOR_QUAD_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK-BE: [[META7]] = !{!"__vector_quad", [[META4]], i64 0} +// CHECK-BE: [[__VECTOR_PAIR_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} +// CHECK-BE: [[META9]] = !{!"__vector_pair", [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/PowerPC/builtins-ppc-dmf.c b/clang/test/CodeGen/PowerPC/builtins-ppc-dmf.c index d8306a7..585d8bac 100644 --- a/clang/test/CodeGen/PowerPC/builtins-ppc-dmf.c +++ b/clang/test/CodeGen/PowerPC/builtins-ppc-dmf.c @@ -8,17 +8,17 @@ // CHECK-LABEL: define dso_local void @test_dmxvi8gerx4( // CHECK-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA2:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6:![0-9]+]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvi8gerx4(<256 x i1> [[TMP0]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6:![0-9]+]] +// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8:![0-9]+]] // CHECK-NEXT: ret void // // AIX-LABEL: define void @test_dmxvi8gerx4( // AIX-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // AIX-NEXT: [[ENTRY:.*:]] -// AIX-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA2:![0-9]+]] +// AIX-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6:![0-9]+]] // AIX-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvi8gerx4(<256 x i1> [[TMP0]], <16 x i8> [[VC]]) -// AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6:![0-9]+]] +// AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8:![0-9]+]] // AIX-NEXT: ret void // void test_dmxvi8gerx4(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -31,17 +31,17 @@ void test_dmxvi8gerx4(unsigned char *vdmrp, unsigned char *vpp, vector unsigned // CHECK-LABEL: define dso_local void @test_pmdmxvi8gerx4( // CHECK-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvi8gerx4(<256 x i1> [[TMP0]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // // AIX-LABEL: define void @test_pmdmxvi8gerx4( // AIX-SAME: ptr noundef readnone captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // AIX-NEXT: [[ENTRY:.*:]] -// AIX-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA2]] +// AIX-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] // AIX-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvi8gerx4(<256 x i1> [[TMP0]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // AIX-NEXT: ret void // void test_pmdmxvi8gerx4(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -54,19 +54,19 @@ void test_pmdmxvi8gerx4(unsigned char *vdmrp, unsigned char *vpp, vector unsigne // CHECK-LABEL: define dso_local void @test_dmxvi8gerx4pp( // CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvi8gerx4pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // // AIX-LABEL: define void @test_dmxvi8gerx4pp( // AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // AIX-NEXT: [[ENTRY:.*:]] -// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA6]] -// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA2]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] // AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvi8gerx4pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) -// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // AIX-NEXT: ret void // void test_dmxvi8gerx4pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -79,19 +79,19 @@ void test_dmxvi8gerx4pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigne // CHECK-LABEL: define dso_local void @test_pmdmxvi8gerx4pp( // CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvi8gerx4pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // // AIX-LABEL: define void @test_pmdmxvi8gerx4pp( // AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // AIX-NEXT: [[ENTRY:.*:]] -// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA6]] -// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA2]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] // AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvi8gerx4pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // AIX-NEXT: ret void // void test_pmdmxvi8gerx4pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -104,19 +104,19 @@ void test_pmdmxvi8gerx4pp(unsigned char *vdmrp, unsigned char *vpp, vector unsig // CHECK-LABEL: define dso_local void @test_dmxvi8gerx4spp( // CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvi8gerx4spp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // // AIX-LABEL: define void @test_dmxvi8gerx4spp( // AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // AIX-NEXT: [[ENTRY:.*:]] -// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA6]] -// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA2]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] // AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvi8gerx4spp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) -// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // AIX-NEXT: ret void // void test_dmxvi8gerx4spp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -129,19 +129,19 @@ void test_dmxvi8gerx4spp(unsigned char *vdmrp, unsigned char *vpp, vector unsign // CHECK-LABEL: define dso_local void @test_pmdmxvi8gerx4spp( // CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvi8gerx4spp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // // AIX-LABEL: define void @test_pmdmxvi8gerx4spp( // AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // AIX-NEXT: [[ENTRY:.*:]] -// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA6]] -// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA2]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] // AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvi8gerx4spp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // AIX-NEXT: ret void // void test_pmdmxvi8gerx4spp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -185,7 +185,7 @@ void test_dmf_basic(char *p, char *res1, char *res2) { // CHECK-LABEL: define dso_local void @test_dmf_basic2( // CHECK-SAME: ptr noundef readonly captures(none) [[P1:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RES1:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RES2:%.*]], ptr noundef readonly captures(none) [[V:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <16 x i8>, ptr [[V]], align 16, !tbaa [[CHAR_TBAA8:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load <16 x i8>, ptr [[V]], align 16, !tbaa [[CHAR_TBAA10:![0-9]+]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.build.dmr(<16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]]) // CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RES2]], align 128 // CHECK-NEXT: [[TMP2:%.*]] = load <1024 x i1>, ptr [[P1]], align 128 @@ -195,7 +195,7 @@ void test_dmf_basic(char *p, char *res1, char *res2) { // AIX-LABEL: define void @test_dmf_basic2( // AIX-SAME: ptr noundef readonly captures(none) [[P1:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RES1:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RES2:%.*]], ptr noundef readonly captures(none) [[V:%.*]]) local_unnamed_addr #[[ATTR0]] { // AIX-NEXT: [[ENTRY:.*:]] -// AIX-NEXT: [[TMP0:%.*]] = load <16 x i8>, ptr [[V]], align 16, !tbaa [[CHAR_TBAA8:![0-9]+]] +// AIX-NEXT: [[TMP0:%.*]] = load <16 x i8>, ptr [[V]], align 16, !tbaa [[CHAR_TBAA10:![0-9]+]] // AIX-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.build.dmr(<16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]], <16 x i8> [[TMP0]]) // AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RES2]], align 128 // AIX-NEXT: [[TMP2:%.*]] = load <1024 x i1>, ptr [[P1]], align 128 @@ -212,19 +212,19 @@ void test_dmf_basic2(char *p1, char *res1, char *res2, // CHECK-LABEL: define dso_local void @test_dmsha2hash( // CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP1:%.*]], ptr noundef readonly captures(none) [[VDMRP2:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP1]], align 128, !tbaa [[__DMR1024_TBAA6]] -// CHECK-NEXT: [[TMP1:%.*]] = load <1024 x i1>, ptr [[VDMRP2]], align 128, !tbaa [[__DMR1024_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP1]], align 128, !tbaa [[__DMR1024_TBAA8]] +// CHECK-NEXT: [[TMP1:%.*]] = load <1024 x i1>, ptr [[VDMRP2]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmsha2hash(<1024 x i1> [[TMP0]], <1024 x i1> [[TMP1]], i32 1) -// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// CHECK-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // // AIX-LABEL: define void @test_dmsha2hash( // AIX-SAME: ptr noundef readonly captures(none) [[VDMRP1:%.*]], ptr noundef readonly captures(none) [[VDMRP2:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // AIX-NEXT: [[ENTRY:.*:]] -// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP1]], align 128, !tbaa [[__DMR1024_TBAA6]] -// AIX-NEXT: [[TMP1:%.*]] = load <1024 x i1>, ptr [[VDMRP2]], align 128, !tbaa [[__DMR1024_TBAA6]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP1]], align 128, !tbaa [[__DMR1024_TBAA8]] +// AIX-NEXT: [[TMP1:%.*]] = load <1024 x i1>, ptr [[VDMRP2]], align 128, !tbaa [[__DMR1024_TBAA8]] // AIX-NEXT: [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmsha2hash(<1024 x i1> [[TMP0]], <1024 x i1> [[TMP1]], i32 1) -// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// AIX-NEXT: store <1024 x i1> [[TMP2]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // AIX-NEXT: ret void // void test_dmsha2hash(unsigned char *vdmrp1, unsigned char *vdmrp2, unsigned char *resp) { @@ -237,17 +237,17 @@ void test_dmsha2hash(unsigned char *vdmrp1, unsigned char *vdmrp2, unsigned char // CHECK-LABEL: define dso_local void @test_dmsha3hash( // CHECK-SAME: ptr noundef readonly captures(none) [[VDMRPP:%.*]], ptr noundef writeonly captures(none) initializes((0, 256)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <2048 x i1>, ptr [[VDMRPP]], align 256, !tbaa [[__DMR2048_TBAA9:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load <2048 x i1>, ptr [[VDMRPP]], align 256, !tbaa [[__DMR2048_TBAA11:![0-9]+]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <2048 x i1> @llvm.ppc.mma.dmsha3hash(<2048 x i1> [[TMP0]], i32 4) -// CHECK-NEXT: store <2048 x i1> [[TMP1]], ptr [[RESP]], align 256, !tbaa [[__DMR2048_TBAA9]] +// CHECK-NEXT: store <2048 x i1> [[TMP1]], ptr [[RESP]], align 256, !tbaa [[__DMR2048_TBAA11]] // CHECK-NEXT: ret void // // AIX-LABEL: define void @test_dmsha3hash( // AIX-SAME: ptr noundef readonly captures(none) [[VDMRPP:%.*]], ptr noundef writeonly captures(none) initializes((0, 256)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // AIX-NEXT: [[ENTRY:.*:]] -// AIX-NEXT: [[TMP0:%.*]] = load <2048 x i1>, ptr [[VDMRPP]], align 256, !tbaa [[__DMR2048_TBAA9:![0-9]+]] +// AIX-NEXT: [[TMP0:%.*]] = load <2048 x i1>, ptr [[VDMRPP]], align 256, !tbaa [[__DMR2048_TBAA11:![0-9]+]] // AIX-NEXT: [[TMP1:%.*]] = tail call <2048 x i1> @llvm.ppc.mma.dmsha3hash(<2048 x i1> [[TMP0]], i32 4) -// AIX-NEXT: store <2048 x i1> [[TMP1]], ptr [[RESP]], align 256, !tbaa [[__DMR2048_TBAA9]] +// AIX-NEXT: store <2048 x i1> [[TMP1]], ptr [[RESP]], align 256, !tbaa [[__DMR2048_TBAA11]] // AIX-NEXT: ret void // void test_dmsha3hash(unsigned char *vdmrpp, unsigned char *resp) { @@ -259,17 +259,17 @@ void test_dmsha3hash(unsigned char *vdmrpp, unsigned char *resp) { // CHECK-LABEL: define dso_local void @test_dmxxshapad( // CHECK-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxxshapad(<1024 x i1> [[TMP0]], <16 x i8> [[VC]], i32 2, i32 1, i32 5) -// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// CHECK-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // CHECK-NEXT: ret void // // AIX-LABEL: define void @test_dmxxshapad( // AIX-SAME: ptr noundef readonly captures(none) [[VDMRP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 128)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // AIX-NEXT: [[ENTRY:.*:]] -// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// AIX-NEXT: [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP]], align 128, !tbaa [[__DMR1024_TBAA8]] // AIX-NEXT: [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxxshapad(<1024 x i1> [[TMP0]], <16 x i8> [[VC]], i32 2, i32 1, i32 5) -// AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA6]] +// AIX-NEXT: store <1024 x i1> [[TMP1]], ptr [[RESP]], align 128, !tbaa [[__DMR1024_TBAA8]] // AIX-NEXT: ret void // void test_dmxxshapad(unsigned char *vdmrp, vector unsigned char vc, unsigned char *resp) { @@ -278,23 +278,23 @@ void test_dmxxshapad(unsigned char *vdmrp, vector unsigned char vc, unsigned cha *((__dmr1024 *)resp) = vdmr; } //. -// CHECK: [[__VECTOR_PAIR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"__vector_pair", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} -// CHECK: [[__DMR1024_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK: [[META7]] = !{!"__dmr1024", [[META4]], i64 0} -// CHECK: [[CHAR_TBAA8]] = !{[[META4]], [[META4]], i64 0} -// CHECK: [[__DMR2048_TBAA9]] = !{[[META10:![0-9]+]], [[META10]], i64 0} -// CHECK: [[META10]] = !{!"__dmr2048", [[META4]], i64 0} +// CHECK: [[__VECTOR_PAIR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK: [[META7]] = !{!"__vector_pair", [[META4]], i64 0} +// CHECK: [[__DMR1024_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} +// CHECK: [[META9]] = !{!"__dmr1024", [[META4]], i64 0} +// CHECK: [[CHAR_TBAA10]] = !{[[META4]], [[META4]], i64 0} +// CHECK: [[__DMR2048_TBAA11]] = !{[[META12:![0-9]+]], [[META12]], i64 0} +// CHECK: [[META12]] = !{!"__dmr2048", [[META4]], i64 0} //. -// AIX: [[__VECTOR_PAIR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// AIX: [[META3]] = !{!"__vector_pair", [[META4:![0-9]+]], i64 0} -// AIX: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// AIX: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // AIX: [[META5]] = !{!"Simple C/C++ TBAA"} -// AIX: [[__DMR1024_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// AIX: [[META7]] = !{!"__dmr1024", [[META4]], i64 0} -// AIX: [[CHAR_TBAA8]] = !{[[META4]], [[META4]], i64 0} -// AIX: [[__DMR2048_TBAA9]] = !{[[META10:![0-9]+]], [[META10]], i64 0} -// AIX: [[META10]] = !{!"__dmr2048", [[META4]], i64 0} +// AIX: [[__VECTOR_PAIR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// AIX: [[META7]] = !{!"__vector_pair", [[META4]], i64 0} +// AIX: [[__DMR1024_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} +// AIX: [[META9]] = !{!"__dmr1024", [[META4]], i64 0} +// AIX: [[CHAR_TBAA10]] = !{[[META4]], [[META4]], i64 0} +// AIX: [[__DMR2048_TBAA11]] = !{[[META12:![0-9]+]], [[META12]], i64 0} +// AIX: [[META12]] = !{!"__dmr2048", [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c b/clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c index 5c7b222..74f9714 100644 --- a/clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c +++ b/clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c @@ -8,7 +8,7 @@ // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.assemble.acc(<16 x i8> [[VC]], <16 x i8> [[VC]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2:![0-9]+]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6:![0-9]+]] // CHECK-NEXT: ret void // void test1(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -45,7 +45,7 @@ void test2(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsi // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 32)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <256 x i1> @llvm.ppc.vsx.assemble.pair(<16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <256 x i1> [[TMP0]], ptr [[RESP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6:![0-9]+]] +// CHECK-NEXT: store <256 x i1> [[TMP0]], ptr [[RESP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA8:![0-9]+]] // CHECK-NEXT: ret void // void test3(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -75,9 +75,9 @@ void test4(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsi // CHECK-LABEL: define dso_local void @test5( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xxmtacc(<512 x i1> [[TMP0]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test5(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -90,9 +90,9 @@ void test5(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsi // CHECK-LABEL: define dso_local void @test6( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xxmfacc(<512 x i1> [[TMP0]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test6(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -106,7 +106,7 @@ void test6(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsi // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xxsetaccz() -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test7(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -120,7 +120,7 @@ void test7(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsi // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvi4ger8(<16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test8(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -134,7 +134,7 @@ void test8(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsi // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvi8ger4(<16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test9(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -148,7 +148,7 @@ void test9(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsi // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvi16ger2(<16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test10(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -162,7 +162,7 @@ void test10(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvi16ger2s(<16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test11(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -176,7 +176,7 @@ void test11(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf16ger2(<16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test12(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -190,7 +190,7 @@ void test12(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf32ger(<16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test13(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -203,9 +203,9 @@ void test13(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test14( // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA8]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf64ger(<256 x i1> [[TMP0]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test14(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -219,7 +219,7 @@ void test14(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvi4ger8(<16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test15(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -233,7 +233,7 @@ void test15(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvi8ger4(<16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test16(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -247,7 +247,7 @@ void test16(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvi16ger2(<16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test17(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -261,7 +261,7 @@ void test17(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvi16ger2s(<16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test18(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -275,7 +275,7 @@ void test18(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf16ger2(<16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test19(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -289,7 +289,7 @@ void test19(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf32ger(<16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test20(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -302,9 +302,9 @@ void test20(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test21( // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA8]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf64ger(<256 x i1> [[TMP0]], <16 x i8> [[VC]], i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test21(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -317,9 +317,9 @@ void test21(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test22( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvi4ger8pp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test22(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -332,9 +332,9 @@ void test22(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test23( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvi8ger4pp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test23(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -347,9 +347,9 @@ void test23(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test24( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvi8ger4spp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test24(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -362,9 +362,9 @@ void test24(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test25( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvi16ger2pp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test25(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -377,9 +377,9 @@ void test25(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test26( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvi16ger2spp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test26(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -392,9 +392,9 @@ void test26(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test27( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvi4ger8pp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test27(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -407,9 +407,9 @@ void test27(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test28( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvi8ger4pp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test28(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -422,9 +422,9 @@ void test28(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test29( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvi8ger4spp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test29(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -437,9 +437,9 @@ void test29(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test30( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvi16ger2pp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test30(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -452,9 +452,9 @@ void test30(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test31( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvi16ger2spp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test31(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -467,9 +467,9 @@ void test31(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test32( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf16ger2pp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test32(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -482,9 +482,9 @@ void test32(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test33( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf16ger2pn(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test33(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -497,9 +497,9 @@ void test33(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test34( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf16ger2np(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test34(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -512,9 +512,9 @@ void test34(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test35( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf16ger2nn(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test35(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -527,9 +527,9 @@ void test35(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test36( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf16ger2pp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test36(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -542,9 +542,9 @@ void test36(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test37( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf16ger2pn(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test37(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -557,9 +557,9 @@ void test37(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test38( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf16ger2np(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test38(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -572,9 +572,9 @@ void test38(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test39( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf16ger2nn(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test39(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -587,9 +587,9 @@ void test39(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test40( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf32gerpp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test40(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -602,9 +602,9 @@ void test40(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test41( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf32gerpn(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test41(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -617,9 +617,9 @@ void test41(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test42( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf32gernp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test42(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -632,9 +632,9 @@ void test42(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test43( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf32gernn(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test43(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -647,9 +647,9 @@ void test43(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test44( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf32gerpp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test44(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -662,9 +662,9 @@ void test44(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test45( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf32gerpn(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test45(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -677,9 +677,9 @@ void test45(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test46( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf32gernp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test46(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -692,9 +692,9 @@ void test46(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test47( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf32gernn(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test47(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -707,10 +707,10 @@ void test47(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test48( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA8]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf64gerpp(<512 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test48(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -723,10 +723,10 @@ void test48(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test49( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA8]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf64gerpn(<512 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test49(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -739,10 +739,10 @@ void test49(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test50( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA8]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf64gernp(<512 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test50(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -755,10 +755,10 @@ void test50(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test51( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA8]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf64gernn(<512 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test51(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -771,10 +771,10 @@ void test51(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test52( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA8]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf64gerpp(<512 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test52(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -787,10 +787,10 @@ void test52(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test53( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA8]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf64gerpn(<512 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test53(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -803,10 +803,10 @@ void test53(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test54( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA8]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf64gernp(<512 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test54(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -819,10 +819,10 @@ void test54(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test55( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readonly captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA8]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf64gernn(<512 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]], i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test55(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -836,7 +836,7 @@ void test55(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvbf16ger2(<16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test56(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -850,7 +850,7 @@ void test56(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvbf16ger2(<16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP0]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test57(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -863,9 +863,9 @@ void test57(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test58( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvbf16ger2pp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test58(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -878,9 +878,9 @@ void test58(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test59( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvbf16ger2pn(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test59(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -893,9 +893,9 @@ void test59(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test60( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvbf16ger2np(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test60(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -908,9 +908,9 @@ void test60(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test61( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvbf16ger2nn(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test61(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -923,9 +923,9 @@ void test61(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test62( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvbf16ger2pp(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test62(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -938,9 +938,9 @@ void test62(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test63( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvbf16ger2pn(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test63(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -953,9 +953,9 @@ void test63(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test64( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvbf16ger2np(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test64(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -968,9 +968,9 @@ void test64(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns // CHECK-LABEL: define dso_local void @test65( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvbf16ger2nn(<512 x i1> [[TMP0]], <16 x i8> [[VC]], <16 x i8> [[VC]], i32 0, i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP1]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test65(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -1079,11 +1079,11 @@ void test72(const __vector_pair *vpp, __vector_pair *vp2) { // CHECK-LABEL: define dso_local void @test73( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[VPP]], i64 8 // CHECK-NEXT: [[TMP2:%.*]] = tail call <256 x i1> @llvm.ppc.vsx.lxvp(ptr [[TMP1]]) // CHECK-NEXT: [[TMP3:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf64gernn(<512 x i1> [[TMP0]], <256 x i1> [[TMP2]], <16 x i8> [[VC]], i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP3]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP3]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test73(unsigned char *vqp, const __vector_pair *vpp, vector unsigned char vc, unsigned char *resp) { @@ -1096,10 +1096,10 @@ void test73(unsigned char *vqp, const __vector_pair *vpp, vector unsigned char v // CHECK-LABEL: define dso_local void @test74( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <256 x i1> @llvm.ppc.vsx.lxvp(ptr [[VPP]]) // CHECK-NEXT: [[TMP2:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf64gernp(<512 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test74(unsigned char *vqp, const __vector_pair *vpp, vector unsigned char vc, unsigned char *resp) { @@ -1112,11 +1112,11 @@ void test74(unsigned char *vqp, const __vector_pair *vpp, vector unsigned char v // CHECK-LABEL: define dso_local void @test75( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], i64 noundef [[OFFS:%.*]], ptr noundef [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[VPP]], i64 [[OFFS]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <256 x i1> @llvm.ppc.vsx.lxvp(ptr [[TMP1]]) // CHECK-NEXT: [[TMP3:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf64gernp(<512 x i1> [[TMP0]], <256 x i1> [[TMP2]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP3]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP3]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test75(unsigned char *vqp, signed long offs, const __vector_pair *vpp, vector unsigned char vc, unsigned char *resp) { @@ -1130,7 +1130,7 @@ void test75(unsigned char *vqp, signed long offs, const __vector_pair *vpp, vect // CHECK-SAME: ptr noundef readnone captures(none) [[VQP:%.*]], ptr noundef readnone captures(none) [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 32)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = tail call <256 x i1> @llvm.ppc.vsx.assemble.pair(<16 x i8> [[VC]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <256 x i1> [[TMP0]], ptr [[RESP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA6]] +// CHECK-NEXT: store <256 x i1> [[TMP0]], ptr [[RESP]], align 32, !tbaa [[__VECTOR_PAIR_TBAA8]] // CHECK-NEXT: ret void // void test76(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) { @@ -1256,11 +1256,11 @@ void test84(const __vector_pair *vpp, __vector_pair *vp2) { // CHECK-LABEL: define dso_local void @test85( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[VPP]], i64 8 // CHECK-NEXT: [[TMP2:%.*]] = tail call <256 x i1> @llvm.ppc.vsx.lxvp(ptr [[TMP1]]) // CHECK-NEXT: [[TMP3:%.*]] = tail call <512 x i1> @llvm.ppc.mma.pmxvf64gernn(<512 x i1> [[TMP0]], <256 x i1> [[TMP2]], <16 x i8> [[VC]], i32 0, i32 0) -// CHECK-NEXT: store <512 x i1> [[TMP3]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP3]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test85(unsigned char *vqp, const __vector_pair *vpp, vector unsigned char vc, unsigned char *resp) { @@ -1273,10 +1273,10 @@ void test85(unsigned char *vqp, const __vector_pair *vpp, vector unsigned char v // CHECK-LABEL: define dso_local void @test86( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], ptr noundef [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = tail call <256 x i1> @llvm.ppc.vsx.lxvp(ptr [[VPP]]) // CHECK-NEXT: [[TMP2:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf64gernp(<512 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP2]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test86(unsigned char *vqp, const __vector_pair *vpp, vector unsigned char vc, unsigned char *resp) { @@ -1289,11 +1289,11 @@ void test86(unsigned char *vqp, const __vector_pair *vpp, vector unsigned char v // CHECK-LABEL: define dso_local void @test87( // CHECK-SAME: ptr noundef readonly captures(none) [[VQP:%.*]], i64 noundef [[OFFS:%.*]], ptr noundef [[VPP:%.*]], <16 x i8> noundef [[VC:%.*]], ptr noundef writeonly captures(none) initializes((0, 64)) [[RESP:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <512 x i1>, ptr [[VQP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[VPP]], i64 [[OFFS]] // CHECK-NEXT: [[TMP2:%.*]] = tail call <256 x i1> @llvm.ppc.vsx.lxvp(ptr [[TMP1]]) // CHECK-NEXT: [[TMP3:%.*]] = tail call <512 x i1> @llvm.ppc.mma.xvf64gernp(<512 x i1> [[TMP0]], <256 x i1> [[TMP2]], <16 x i8> [[VC]]) -// CHECK-NEXT: store <512 x i1> [[TMP3]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA2]] +// CHECK-NEXT: store <512 x i1> [[TMP3]], ptr [[RESP]], align 64, !tbaa [[__VECTOR_QUAD_TBAA6]] // CHECK-NEXT: ret void // void test87(unsigned char *vqp, signed long offs, const __vector_pair *vpp, vector unsigned char vc, unsigned char *resp) { @@ -1303,10 +1303,10 @@ void test87(unsigned char *vqp, signed long offs, const __vector_pair *vpp, vect *((__vector_quad *)resp) = vq; } //. -// CHECK: [[__VECTOR_QUAD_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"__vector_quad", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} -// CHECK: [[__VECTOR_PAIR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK: [[META7]] = !{!"__vector_pair", [[META4]], i64 0} +// CHECK: [[__VECTOR_QUAD_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK: [[META7]] = !{!"__vector_quad", [[META4]], i64 0} +// CHECK: [[__VECTOR_PAIR_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} +// CHECK: [[META9]] = !{!"__vector_pair", [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-bitcast-less-8.c b/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-bitcast-less-8.c index 1f0b3d4..c943649 100644 --- a/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-bitcast-less-8.c +++ b/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-bitcast-less-8.c @@ -57,7 +57,7 @@ DEFINE_STRUCT(bool64) // CHECK-128-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 1 -// CHECK-128-NEXT: [[TMP0:%.*]] = load <1 x i8>, ptr [[Y]], align 1, !tbaa [[CHAR_TBAA6:![0-9]+]] +// CHECK-128-NEXT: [[TMP0:%.*]] = load <1 x i8>, ptr [[Y]], align 1, !tbaa [[CHAR_TBAA10:![0-9]+]] // CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x i8> @llvm.vector.insert.nxv1i8.v1i8(<vscale x 1 x i8> poison, <1 x i8> [[TMP0]], i64 0) // CHECK-128-NEXT: [[TMP1:%.*]] = bitcast <vscale x 1 x i8> [[CAST_SCALABLE]] to <vscale x 8 x i1> // CHECK-128-NEXT: [[TMP2:%.*]] = tail call <vscale x 2 x i1> @llvm.vector.extract.nxv2i1.nxv8i1(<vscale x 8 x i1> [[TMP1]], i64 0) @@ -74,7 +74,7 @@ vbool32_t read_bool32(struct struct_bool32 *s) { // CHECK-128-NEXT: [[TMP1:%.*]] = bitcast <vscale x 8 x i1> [[TMP0]] to <vscale x 1 x i8> // CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <1 x i8> @llvm.vector.extract.v1i8.nxv1i8(<vscale x 1 x i8> [[TMP1]], i64 0) // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 1 -// CHECK-128-NEXT: store <1 x i8> [[CAST_FIXED]], ptr [[Y]], align 1, !tbaa [[CHAR_TBAA6]] +// CHECK-128-NEXT: store <1 x i8> [[CAST_FIXED]], ptr [[Y]], align 1, !tbaa [[CHAR_TBAA10]] // CHECK-128-NEXT: ret void // void write_bool32(struct struct_bool32 *s, vbool32_t x) { @@ -85,7 +85,7 @@ void write_bool32(struct struct_bool32 *s, vbool32_t x) { // CHECK-128-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 1 -// CHECK-128-NEXT: [[TMP0:%.*]] = load <1 x i8>, ptr [[Y]], align 1, !tbaa [[CHAR_TBAA6]] +// CHECK-128-NEXT: [[TMP0:%.*]] = load <1 x i8>, ptr [[Y]], align 1, !tbaa [[CHAR_TBAA10]] // CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x i8> @llvm.vector.insert.nxv1i8.v1i8(<vscale x 1 x i8> poison, <1 x i8> [[TMP0]], i64 0) // CHECK-128-NEXT: [[TMP1:%.*]] = bitcast <vscale x 1 x i8> [[CAST_SCALABLE]] to <vscale x 8 x i1> // CHECK-128-NEXT: [[TMP2:%.*]] = tail call <vscale x 1 x i1> @llvm.vector.extract.nxv1i1.nxv8i1(<vscale x 8 x i1> [[TMP1]], i64 0) @@ -102,14 +102,14 @@ vbool64_t read_bool64(struct struct_bool64 *s) { // CHECK-128-NEXT: [[TMP1:%.*]] = bitcast <vscale x 8 x i1> [[TMP0]] to <vscale x 1 x i8> // CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <1 x i8> @llvm.vector.extract.v1i8.nxv1i8(<vscale x 1 x i8> [[TMP1]], i64 0) // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 1 -// CHECK-128-NEXT: store <1 x i8> [[CAST_FIXED]], ptr [[Y]], align 1, !tbaa [[CHAR_TBAA6]] +// CHECK-128-NEXT: store <1 x i8> [[CAST_FIXED]], ptr [[Y]], align 1, !tbaa [[CHAR_TBAA10]] // CHECK-128-NEXT: ret void // void write_bool64(struct struct_bool64 *s, vbool64_t x) { s->y[0] = x; } //. -// CHECK-128: [[CHAR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK-128: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0} -// CHECK-128: [[META8]] = !{!"Simple C/C++ TBAA"} +// CHECK-128: [[META8:![0-9]+]] = !{!"omnipotent char", [[META9:![0-9]+]], i64 0} +// CHECK-128: [[META9]] = !{!"Simple C/C++ TBAA"} +// CHECK-128: [[CHAR_TBAA10]] = !{[[META8]], [[META8]], i64 0} //. diff --git a/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-bitcast.c b/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-bitcast.c index b92e6df..71d6f46 100644 --- a/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-bitcast.c +++ b/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-bitcast.c @@ -71,7 +71,7 @@ DEFINE_STRUCT(bool64) // CHECK-64-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-64-NEXT: [[ENTRY:.*:]] // CHECK-64-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 8 -// CHECK-64-NEXT: [[TMP0:%.*]] = load <1 x i64>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6:![0-9]+]] +// CHECK-64-NEXT: [[TMP0:%.*]] = load <1 x i64>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10:![0-9]+]] // CHECK-64-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v1i64(<vscale x 1 x i64> poison, <1 x i64> [[TMP0]], i64 0) // CHECK-64-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]] // @@ -79,7 +79,7 @@ DEFINE_STRUCT(bool64) // CHECK-128-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16 -// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6:![0-9]+]] +// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10:![0-9]+]] // CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v2i64(<vscale x 1 x i64> poison, <2 x i64> [[TMP0]], i64 0) // CHECK-128-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]] // @@ -87,7 +87,7 @@ DEFINE_STRUCT(bool64) // CHECK-256-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32 -// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x i64>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6:![0-9]+]] +// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x i64>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10:![0-9]+]] // CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[TMP0]], i64 0) // CHECK-256-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]] // @@ -100,7 +100,7 @@ vint64m1_t read_int64m1(struct struct_int64m1 *s) { // CHECK-64-NEXT: [[ENTRY:.*:]] // CHECK-64-NEXT: [[CAST_FIXED:%.*]] = tail call <1 x i64> @llvm.vector.extract.v1i64.nxv1i64(<vscale x 1 x i64> [[X]], i64 0) // CHECK-64-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 8 -// CHECK-64-NEXT: store <1 x i64> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-64-NEXT: store <1 x i64> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-64-NEXT: ret void // // CHECK-128-LABEL: define dso_local void @write_int64m1( @@ -108,7 +108,7 @@ vint64m1_t read_int64m1(struct struct_int64m1 *s) { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x i64> @llvm.vector.extract.v2i64.nxv1i64(<vscale x 1 x i64> [[X]], i64 0) // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16 -// CHECK-128-NEXT: store <2 x i64> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-128-NEXT: store <2 x i64> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-128-NEXT: ret void // // CHECK-256-LABEL: define dso_local void @write_int64m1( @@ -116,7 +116,7 @@ vint64m1_t read_int64m1(struct struct_int64m1 *s) { // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[X]], i64 0) // CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32 -// CHECK-256-NEXT: store <4 x i64> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-256-NEXT: store <4 x i64> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-256-NEXT: ret void // void write_int64m1(struct struct_int64m1 *s, vint64m1_t x) { @@ -131,7 +131,7 @@ void write_int64m1(struct struct_int64m1 *s, vint64m1_t x) { // CHECK-64-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-64-NEXT: [[ENTRY:.*:]] // CHECK-64-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 8 -// CHECK-64-NEXT: [[TMP0:%.*]] = load <1 x double>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-64-NEXT: [[TMP0:%.*]] = load <1 x double>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-64-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v1f64(<vscale x 1 x double> poison, <1 x double> [[TMP0]], i64 0) // CHECK-64-NEXT: ret <vscale x 1 x double> [[CAST_SCALABLE]] // @@ -139,7 +139,7 @@ void write_int64m1(struct struct_int64m1 *s, vint64m1_t x) { // CHECK-128-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16 -// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v2f64(<vscale x 1 x double> poison, <2 x double> [[TMP0]], i64 0) // CHECK-128-NEXT: ret <vscale x 1 x double> [[CAST_SCALABLE]] // @@ -147,7 +147,7 @@ void write_int64m1(struct struct_int64m1 *s, vint64m1_t x) { // CHECK-256-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32 -// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x double>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x double>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> poison, <4 x double> [[TMP0]], i64 0) // CHECK-256-NEXT: ret <vscale x 1 x double> [[CAST_SCALABLE]] // @@ -160,7 +160,7 @@ vfloat64m1_t read_float64m1(struct struct_float64m1 *s) { // CHECK-64-NEXT: [[ENTRY:.*:]] // CHECK-64-NEXT: [[CAST_FIXED:%.*]] = tail call <1 x double> @llvm.vector.extract.v1f64.nxv1f64(<vscale x 1 x double> [[X]], i64 0) // CHECK-64-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 8 -// CHECK-64-NEXT: store <1 x double> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-64-NEXT: store <1 x double> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-64-NEXT: ret void // // CHECK-128-LABEL: define dso_local void @write_float64m1( @@ -168,7 +168,7 @@ vfloat64m1_t read_float64m1(struct struct_float64m1 *s) { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x double> @llvm.vector.extract.v2f64.nxv1f64(<vscale x 1 x double> [[X]], i64 0) // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16 -// CHECK-128-NEXT: store <2 x double> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-128-NEXT: store <2 x double> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-128-NEXT: ret void // // CHECK-256-LABEL: define dso_local void @write_float64m1( @@ -176,7 +176,7 @@ vfloat64m1_t read_float64m1(struct struct_float64m1 *s) { // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[X]], i64 0) // CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32 -// CHECK-256-NEXT: store <4 x double> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-256-NEXT: store <4 x double> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-256-NEXT: ret void // void write_float64m1(struct struct_float64m1 *s, vfloat64m1_t x) { @@ -191,7 +191,7 @@ void write_float64m1(struct struct_float64m1 *s, vfloat64m1_t x) { // CHECK-64-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-64-NEXT: [[ENTRY:.*:]] // CHECK-64-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 8 -// CHECK-64-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-64-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-64-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v8i8(<vscale x 8 x i8> poison, <8 x i8> [[TMP0]], i64 0) // CHECK-64-NEXT: [[TMP1:%.*]] = bitcast <vscale x 8 x i8> [[CAST_SCALABLE]] to <vscale x 64 x i1> // CHECK-64-NEXT: ret <vscale x 64 x i1> [[TMP1]] @@ -200,7 +200,7 @@ void write_float64m1(struct struct_float64m1 *s, vfloat64m1_t x) { // CHECK-128-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16 -// CHECK-128-NEXT: [[TMP0:%.*]] = load <16 x i8>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-128-NEXT: [[TMP0:%.*]] = load <16 x i8>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v16i8(<vscale x 8 x i8> poison, <16 x i8> [[TMP0]], i64 0) // CHECK-128-NEXT: [[TMP1:%.*]] = bitcast <vscale x 8 x i8> [[CAST_SCALABLE]] to <vscale x 64 x i1> // CHECK-128-NEXT: ret <vscale x 64 x i1> [[TMP1]] @@ -209,7 +209,7 @@ void write_float64m1(struct struct_float64m1 *s, vfloat64m1_t x) { // CHECK-256-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32 -// CHECK-256-NEXT: [[TMP0:%.*]] = load <32 x i8>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-256-NEXT: [[TMP0:%.*]] = load <32 x i8>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[TMP0]], i64 0) // CHECK-256-NEXT: [[TMP1:%.*]] = bitcast <vscale x 8 x i8> [[CAST_SCALABLE]] to <vscale x 64 x i1> // CHECK-256-NEXT: ret <vscale x 64 x i1> [[TMP1]] @@ -224,7 +224,7 @@ vbool1_t read_bool1(struct struct_bool1 *s) { // CHECK-64-NEXT: [[TMP0:%.*]] = bitcast <vscale x 64 x i1> [[X]] to <vscale x 8 x i8> // CHECK-64-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x i8> @llvm.vector.extract.v8i8.nxv8i8(<vscale x 8 x i8> [[TMP0]], i64 0) // CHECK-64-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 8 -// CHECK-64-NEXT: store <8 x i8> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-64-NEXT: store <8 x i8> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-64-NEXT: ret void // // CHECK-128-LABEL: define dso_local void @write_bool1( @@ -233,7 +233,7 @@ vbool1_t read_bool1(struct struct_bool1 *s) { // CHECK-128-NEXT: [[TMP0:%.*]] = bitcast <vscale x 64 x i1> [[X]] to <vscale x 8 x i8> // CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <16 x i8> @llvm.vector.extract.v16i8.nxv8i8(<vscale x 8 x i8> [[TMP0]], i64 0) // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16 -// CHECK-128-NEXT: store <16 x i8> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-128-NEXT: store <16 x i8> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-128-NEXT: ret void // // CHECK-256-LABEL: define dso_local void @write_bool1( @@ -242,22 +242,22 @@ vbool1_t read_bool1(struct struct_bool1 *s) { // CHECK-256-NEXT: [[TMP0:%.*]] = bitcast <vscale x 64 x i1> [[X]] to <vscale x 8 x i8> // CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[TMP0]], i64 0) // CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32 -// CHECK-256-NEXT: store <32 x i8> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-256-NEXT: store <32 x i8> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA10]] // CHECK-256-NEXT: ret void // void write_bool1(struct struct_bool1 *s, vbool1_t x) { s->y[0] = x; } //. -// CHECK-64: [[CHAR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK-64: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0} -// CHECK-64: [[META8]] = !{!"Simple C/C++ TBAA"} +// CHECK-64: [[META8:![0-9]+]] = !{!"omnipotent char", [[META9:![0-9]+]], i64 0} +// CHECK-64: [[META9]] = !{!"Simple C/C++ TBAA"} +// CHECK-64: [[CHAR_TBAA10]] = !{[[META8]], [[META8]], i64 0} //. -// CHECK-128: [[CHAR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK-128: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0} -// CHECK-128: [[META8]] = !{!"Simple C/C++ TBAA"} +// CHECK-128: [[META8:![0-9]+]] = !{!"omnipotent char", [[META9:![0-9]+]], i64 0} +// CHECK-128: [[META9]] = !{!"Simple C/C++ TBAA"} +// CHECK-128: [[CHAR_TBAA10]] = !{[[META8]], [[META8]], i64 0} //. -// CHECK-256: [[CHAR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK-256: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0} -// CHECK-256: [[META8]] = !{!"Simple C/C++ TBAA"} +// CHECK-256: [[META8:![0-9]+]] = !{!"omnipotent char", [[META9:![0-9]+]], i64 0} +// CHECK-256: [[META9]] = !{!"Simple C/C++ TBAA"} +// CHECK-256: [[CHAR_TBAA10]] = !{[[META8]], [[META8]], i64 0} //. diff --git a/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-cast.c b/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-cast.c index 4517b52..fd500f0 100644 --- a/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-cast.c +++ b/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-cast.c @@ -124,7 +124,7 @@ vbool32_t to_vbool32_t(fixed_bool32_t type) { // CHECK-LABEL: define dso_local <vscale x 2 x i32> @to_vint32m1_t__from_gnu_int32m1_t( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR2:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TYPE:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6:![0-9]+]] +// CHECK-NEXT: [[TYPE:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA10:![0-9]+]] // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[TYPE]], i64 0) // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]] // @@ -136,7 +136,7 @@ vint32m1_t to_vint32m1_t__from_gnu_int32m1_t(gnu_int32m1_t type) { // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], <vscale x 2 x i32> [[TYPE:%.*]]) local_unnamed_addr #[[ATTR3:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[TYPE]], i64 0) -// CHECK-NEXT: store <8 x i32> [[CAST_FIXED]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: store <8 x i32> [[CAST_FIXED]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA10]] // CHECK-NEXT: ret void // gnu_int32m1_t from_vint32m1_t__to_gnu_int32m1_t(vint32m1_t type) { @@ -146,7 +146,7 @@ gnu_int32m1_t from_vint32m1_t__to_gnu_int32m1_t(vint32m1_t type) { // CHECK-LABEL: define dso_local <vscale x 2 x i32> @to_fixed_int32m1_t__from_gnu_int32m1_t( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TYPE:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TYPE:%.*]] = load <8 x i32>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA10]] // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[TYPE]], i64 0) // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]] // @@ -158,14 +158,14 @@ fixed_int32m1_t to_fixed_int32m1_t__from_gnu_int32m1_t(gnu_int32m1_t type) { // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<8 x i32>) align 32 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], <vscale x 2 x i32> noundef [[TYPE_COERCE:%.*]]) local_unnamed_addr #[[ATTR3]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TYPE:%.*]] = tail call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[TYPE_COERCE]], i64 0) -// CHECK-NEXT: store <8 x i32> [[TYPE]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: store <8 x i32> [[TYPE]], ptr [[AGG_RESULT]], align 32, !tbaa [[CHAR_TBAA10]] // CHECK-NEXT: ret void // gnu_int32m1_t from_fixed_int32m1_t__to_gnu_int32m1_t(fixed_int32m1_t type) { return type; } //. -// CHECK: [[CHAR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0} -// CHECK: [[META8]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META8:![0-9]+]] = !{!"omnipotent char", [[META9:![0-9]+]], i64 0} +// CHECK: [[META9]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[CHAR_TBAA10]] = !{[[META8]], [[META8]], i64 0} //. diff --git a/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-globals.c b/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-globals.c index f3b91b2..f6c734f 100644 --- a/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-globals.c +++ b/clang/test/CodeGen/RISCV/attr-rvv-vector-bits-globals.c @@ -44,14 +44,14 @@ fixed_bool32_t global_bool32; // CHECK-64-SAME: <vscale x 1 x i64> [[V:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-64-NEXT: [[ENTRY:.*:]] // CHECK-64-NEXT: [[CAST_FIXED:%.*]] = tail call <1 x i64> @llvm.vector.extract.v1i64.nxv1i64(<vscale x 1 x i64> [[V]], i64 0) -// CHECK-64-NEXT: store <1 x i64> [[CAST_FIXED]], ptr @global_i64, align 8, !tbaa [[CHAR_TBAA6:![0-9]+]] +// CHECK-64-NEXT: store <1 x i64> [[CAST_FIXED]], ptr @global_i64, align 8, !tbaa [[CHAR_TBAA10:![0-9]+]] // CHECK-64-NEXT: ret void // // CHECK-256-LABEL: define dso_local void @write_global_i64( // CHECK-256-SAME: <vscale x 1 x i64> [[V:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[V]], i64 0) -// CHECK-256-NEXT: store <4 x i64> [[CAST_FIXED]], ptr @global_i64, align 8, !tbaa [[CHAR_TBAA6:![0-9]+]] +// CHECK-256-NEXT: store <4 x i64> [[CAST_FIXED]], ptr @global_i64, align 8, !tbaa [[CHAR_TBAA10:![0-9]+]] // CHECK-256-NEXT: ret void // void write_global_i64(vint64m1_t v) { global_i64 = v; } @@ -61,7 +61,7 @@ void write_global_i64(vint64m1_t v) { global_i64 = v; } // CHECK-64-NEXT: [[ENTRY:.*:]] // CHECK-64-NEXT: [[TMP0:%.*]] = bitcast <vscale x 64 x i1> [[V]] to <vscale x 8 x i8> // CHECK-64-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x i8> @llvm.vector.extract.v8i8.nxv8i8(<vscale x 8 x i8> [[TMP0]], i64 0) -// CHECK-64-NEXT: store <8 x i8> [[CAST_FIXED]], ptr @global_bool1, align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-64-NEXT: store <8 x i8> [[CAST_FIXED]], ptr @global_bool1, align 8, !tbaa [[CHAR_TBAA10]] // CHECK-64-NEXT: ret void // // CHECK-256-LABEL: define dso_local void @write_global_bool1( @@ -69,7 +69,7 @@ void write_global_i64(vint64m1_t v) { global_i64 = v; } // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[TMP0:%.*]] = bitcast <vscale x 64 x i1> [[V]] to <vscale x 8 x i8> // CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[TMP0]], i64 0) -// CHECK-256-NEXT: store <32 x i8> [[CAST_FIXED]], ptr @global_bool1, align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-256-NEXT: store <32 x i8> [[CAST_FIXED]], ptr @global_bool1, align 8, !tbaa [[CHAR_TBAA10]] // CHECK-256-NEXT: ret void // void write_global_bool1(vbool1_t v) { global_bool1 = v; } @@ -79,7 +79,7 @@ void write_global_bool1(vbool1_t v) { global_bool1 = v; } // CHECK-64-NEXT: [[ENTRY:.*:]] // CHECK-64-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1> [[V]] to <vscale x 2 x i8> // CHECK-64-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x i8> @llvm.vector.extract.v2i8.nxv2i8(<vscale x 2 x i8> [[TMP0]], i64 0) -// CHECK-64-NEXT: store <2 x i8> [[CAST_FIXED]], ptr @global_bool4, align 2, !tbaa [[CHAR_TBAA6]] +// CHECK-64-NEXT: store <2 x i8> [[CAST_FIXED]], ptr @global_bool4, align 2, !tbaa [[CHAR_TBAA10]] // CHECK-64-NEXT: ret void // // CHECK-256-LABEL: define dso_local void @write_global_bool4( @@ -87,7 +87,7 @@ void write_global_bool1(vbool1_t v) { global_bool1 = v; } // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1> [[V]] to <vscale x 2 x i8> // CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x i8> @llvm.vector.extract.v8i8.nxv2i8(<vscale x 2 x i8> [[TMP0]], i64 0) -// CHECK-256-NEXT: store <8 x i8> [[CAST_FIXED]], ptr @global_bool4, align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-256-NEXT: store <8 x i8> [[CAST_FIXED]], ptr @global_bool4, align 8, !tbaa [[CHAR_TBAA10]] // CHECK-256-NEXT: ret void // void write_global_bool4(vbool4_t v) { global_bool4 = v; } @@ -99,7 +99,7 @@ void write_global_bool4(vbool4_t v) { global_bool4 = v; } // CHECK-256-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.vector.insert.nxv8i1.nxv2i1(<vscale x 8 x i1> zeroinitializer, <vscale x 2 x i1> [[V]], i64 0) // CHECK-256-NEXT: [[TMP1:%.*]] = bitcast <vscale x 8 x i1> [[TMP0]] to <vscale x 1 x i8> // CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <1 x i8> @llvm.vector.extract.v1i8.nxv1i8(<vscale x 1 x i8> [[TMP1]], i64 0) -// CHECK-256-NEXT: store <1 x i8> [[CAST_FIXED]], ptr @global_bool32, align 1, !tbaa [[CHAR_TBAA6]] +// CHECK-256-NEXT: store <1 x i8> [[CAST_FIXED]], ptr @global_bool32, align 1, !tbaa [[CHAR_TBAA10]] // CHECK-256-NEXT: ret void // void write_global_bool32(vbool32_t v) { global_bool32 = v; } @@ -112,14 +112,14 @@ void write_global_bool32(vbool32_t v) { global_bool32 = v; } // CHECK-64-LABEL: define dso_local <vscale x 1 x i64> @read_global_i64( // CHECK-64-SAME: ) local_unnamed_addr #[[ATTR2:[0-9]+]] { // CHECK-64-NEXT: [[ENTRY:.*:]] -// CHECK-64-NEXT: [[TMP0:%.*]] = load <1 x i64>, ptr @global_i64, align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-64-NEXT: [[TMP0:%.*]] = load <1 x i64>, ptr @global_i64, align 8, !tbaa [[CHAR_TBAA10]] // CHECK-64-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v1i64(<vscale x 1 x i64> poison, <1 x i64> [[TMP0]], i64 0) // CHECK-64-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]] // // CHECK-256-LABEL: define dso_local <vscale x 1 x i64> @read_global_i64( // CHECK-256-SAME: ) local_unnamed_addr #[[ATTR2:[0-9]+]] { // CHECK-256-NEXT: [[ENTRY:.*:]] -// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x i64>, ptr @global_i64, align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x i64>, ptr @global_i64, align 8, !tbaa [[CHAR_TBAA10]] // CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[TMP0]], i64 0) // CHECK-256-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]] // @@ -128,7 +128,7 @@ vint64m1_t read_global_i64() { return global_i64; } // CHECK-64-LABEL: define dso_local <vscale x 64 x i1> @read_global_bool1( // CHECK-64-SAME: ) local_unnamed_addr #[[ATTR2]] { // CHECK-64-NEXT: [[ENTRY:.*:]] -// CHECK-64-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr @global_bool1, align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-64-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr @global_bool1, align 8, !tbaa [[CHAR_TBAA10]] // CHECK-64-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v8i8(<vscale x 8 x i8> poison, <8 x i8> [[TMP0]], i64 0) // CHECK-64-NEXT: [[TMP1:%.*]] = bitcast <vscale x 8 x i8> [[CAST_SCALABLE]] to <vscale x 64 x i1> // CHECK-64-NEXT: ret <vscale x 64 x i1> [[TMP1]] @@ -136,7 +136,7 @@ vint64m1_t read_global_i64() { return global_i64; } // CHECK-256-LABEL: define dso_local <vscale x 64 x i1> @read_global_bool1( // CHECK-256-SAME: ) local_unnamed_addr #[[ATTR2]] { // CHECK-256-NEXT: [[ENTRY:.*:]] -// CHECK-256-NEXT: [[TMP0:%.*]] = load <32 x i8>, ptr @global_bool1, align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-256-NEXT: [[TMP0:%.*]] = load <32 x i8>, ptr @global_bool1, align 8, !tbaa [[CHAR_TBAA10]] // CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[TMP0]], i64 0) // CHECK-256-NEXT: [[TMP1:%.*]] = bitcast <vscale x 8 x i8> [[CAST_SCALABLE]] to <vscale x 64 x i1> // CHECK-256-NEXT: ret <vscale x 64 x i1> [[TMP1]] @@ -146,7 +146,7 @@ vbool1_t read_global_bool1() { return global_bool1; } // CHECK-64-LABEL: define dso_local <vscale x 16 x i1> @read_global_bool4( // CHECK-64-SAME: ) local_unnamed_addr #[[ATTR2]] { // CHECK-64-NEXT: [[ENTRY:.*:]] -// CHECK-64-NEXT: [[TMP0:%.*]] = load <2 x i8>, ptr @global_bool4, align 2, !tbaa [[CHAR_TBAA6]] +// CHECK-64-NEXT: [[TMP0:%.*]] = load <2 x i8>, ptr @global_bool4, align 2, !tbaa [[CHAR_TBAA10]] // CHECK-64-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i8> @llvm.vector.insert.nxv2i8.v2i8(<vscale x 2 x i8> poison, <2 x i8> [[TMP0]], i64 0) // CHECK-64-NEXT: [[TMP1:%.*]] = bitcast <vscale x 2 x i8> [[CAST_SCALABLE]] to <vscale x 16 x i1> // CHECK-64-NEXT: ret <vscale x 16 x i1> [[TMP1]] @@ -154,7 +154,7 @@ vbool1_t read_global_bool1() { return global_bool1; } // CHECK-256-LABEL: define dso_local <vscale x 16 x i1> @read_global_bool4( // CHECK-256-SAME: ) local_unnamed_addr #[[ATTR2]] { // CHECK-256-NEXT: [[ENTRY:.*:]] -// CHECK-256-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr @global_bool4, align 8, !tbaa [[CHAR_TBAA6]] +// CHECK-256-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr @global_bool4, align 8, !tbaa [[CHAR_TBAA10]] // CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i8> @llvm.vector.insert.nxv2i8.v8i8(<vscale x 2 x i8> poison, <8 x i8> [[TMP0]], i64 0) // CHECK-256-NEXT: [[TMP1:%.*]] = bitcast <vscale x 2 x i8> [[CAST_SCALABLE]] to <vscale x 16 x i1> // CHECK-256-NEXT: ret <vscale x 16 x i1> [[TMP1]] @@ -165,7 +165,7 @@ vbool4_t read_global_bool4() { return global_bool4; } // CHECK-256-LABEL: define dso_local <vscale x 2 x i1> @read_global_bool32( // CHECK-256-SAME: ) local_unnamed_addr #[[ATTR2]] { // CHECK-256-NEXT: [[ENTRY:.*:]] -// CHECK-256-NEXT: [[TMP0:%.*]] = load <1 x i8>, ptr @global_bool32, align 1, !tbaa [[CHAR_TBAA6]] +// CHECK-256-NEXT: [[TMP0:%.*]] = load <1 x i8>, ptr @global_bool32, align 1, !tbaa [[CHAR_TBAA10]] // CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x i8> @llvm.vector.insert.nxv1i8.v1i8(<vscale x 1 x i8> poison, <1 x i8> [[TMP0]], i64 0) // CHECK-256-NEXT: [[TMP1:%.*]] = bitcast <vscale x 1 x i8> [[CAST_SCALABLE]] to <vscale x 8 x i1> // CHECK-256-NEXT: [[TMP2:%.*]] = tail call <vscale x 2 x i1> @llvm.vector.extract.nxv2i1.nxv8i1(<vscale x 8 x i1> [[TMP1]], i64 0) @@ -174,11 +174,11 @@ vbool4_t read_global_bool4() { return global_bool4; } vbool32_t read_global_bool32() { return global_bool32; } #endif //. -// CHECK-64: [[CHAR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK-64: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0} -// CHECK-64: [[META8]] = !{!"Simple C/C++ TBAA"} +// CHECK-64: [[META8:![0-9]+]] = !{!"omnipotent char", [[META9:![0-9]+]], i64 0} +// CHECK-64: [[META9]] = !{!"Simple C/C++ TBAA"} +// CHECK-64: [[CHAR_TBAA10]] = !{[[META8]], [[META8]], i64 0} //. -// CHECK-256: [[CHAR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK-256: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0} -// CHECK-256: [[META8]] = !{!"Simple C/C++ TBAA"} +// CHECK-256: [[META8:![0-9]+]] = !{!"omnipotent char", [[META9:![0-9]+]], i64 0} +// CHECK-256: [[META9]] = !{!"Simple C/C++ TBAA"} +// CHECK-256: [[CHAR_TBAA10]] = !{[[META8]], [[META8]], i64 0} //. diff --git a/clang/test/CodeGen/SystemZ/builtins-systemz-i128.c b/clang/test/CodeGen/SystemZ/builtins-systemz-i128.c index d25b8d8..2a2060712 100644 --- a/clang/test/CodeGen/SystemZ/builtins-systemz-i128.c +++ b/clang/test/CodeGen/SystemZ/builtins-systemz-i128.c @@ -14,124 +14,124 @@ volatile vector unsigned long long vul; // CHECK-LABEL: define dso_local void @test( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3:![0-9]+]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7:![0-9]+]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to i128 // CHECK-NEXT: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP1]] to i128 // CHECK-NEXT: [[ADD_I:%.*]] = add nsw i128 [[TMP3]], [[TMP2]] // CHECK-NEXT: [[TMP4:%.*]] = bitcast i128 [[ADD_I]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP4]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP5:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP6:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP4]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP5:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP6:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP7:%.*]] = bitcast <16 x i8> [[TMP5]] to i128 // CHECK-NEXT: [[TMP8:%.*]] = bitcast <16 x i8> [[TMP6]] to i128 // CHECK-NEXT: [[TMP9:%.*]] = tail call i128 @llvm.s390.vaccq(i128 [[TMP7]], i128 [[TMP8]]) // CHECK-NEXT: [[TMP10:%.*]] = bitcast i128 [[TMP9]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP10]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP11:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP12:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP13:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP10]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP11:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP12:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP13:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP14:%.*]] = bitcast <16 x i8> [[TMP11]] to i128 // CHECK-NEXT: [[TMP15:%.*]] = bitcast <16 x i8> [[TMP12]] to i128 // CHECK-NEXT: [[TMP16:%.*]] = bitcast <16 x i8> [[TMP13]] to i128 // CHECK-NEXT: [[TMP17:%.*]] = tail call i128 @llvm.s390.vacq(i128 [[TMP14]], i128 [[TMP15]], i128 [[TMP16]]) // CHECK-NEXT: [[TMP18:%.*]] = bitcast i128 [[TMP17]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP18]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP19:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP20:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP21:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP18]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP19:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP20:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP21:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP22:%.*]] = bitcast <16 x i8> [[TMP19]] to i128 // CHECK-NEXT: [[TMP23:%.*]] = bitcast <16 x i8> [[TMP20]] to i128 // CHECK-NEXT: [[TMP24:%.*]] = bitcast <16 x i8> [[TMP21]] to i128 // CHECK-NEXT: [[TMP25:%.*]] = tail call i128 @llvm.s390.vacccq(i128 [[TMP22]], i128 [[TMP23]], i128 [[TMP24]]) // CHECK-NEXT: [[TMP26:%.*]] = bitcast i128 [[TMP25]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP26]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP27:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP28:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP26]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP27:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP28:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP29:%.*]] = bitcast <16 x i8> [[TMP27]] to i128 // CHECK-NEXT: [[TMP30:%.*]] = bitcast <16 x i8> [[TMP28]] to i128 // CHECK-NEXT: [[SUB_I:%.*]] = sub nsw i128 [[TMP29]], [[TMP30]] // CHECK-NEXT: [[TMP31:%.*]] = bitcast i128 [[SUB_I]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP31]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP32:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP33:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP31]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP32:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP33:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP34:%.*]] = bitcast <16 x i8> [[TMP32]] to i128 // CHECK-NEXT: [[TMP35:%.*]] = bitcast <16 x i8> [[TMP33]] to i128 // CHECK-NEXT: [[TMP36:%.*]] = tail call i128 @llvm.s390.vscbiq(i128 [[TMP34]], i128 [[TMP35]]) // CHECK-NEXT: [[TMP37:%.*]] = bitcast i128 [[TMP36]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP37]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP38:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP39:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP40:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP37]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP38:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP39:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP40:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP41:%.*]] = bitcast <16 x i8> [[TMP38]] to i128 // CHECK-NEXT: [[TMP42:%.*]] = bitcast <16 x i8> [[TMP39]] to i128 // CHECK-NEXT: [[TMP43:%.*]] = bitcast <16 x i8> [[TMP40]] to i128 // CHECK-NEXT: [[TMP44:%.*]] = tail call i128 @llvm.s390.vsbiq(i128 [[TMP41]], i128 [[TMP42]], i128 [[TMP43]]) // CHECK-NEXT: [[TMP45:%.*]] = bitcast i128 [[TMP44]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP45]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP46:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP47:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP48:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP45]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP46:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP47:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP48:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP49:%.*]] = bitcast <16 x i8> [[TMP46]] to i128 // CHECK-NEXT: [[TMP50:%.*]] = bitcast <16 x i8> [[TMP47]] to i128 // CHECK-NEXT: [[TMP51:%.*]] = bitcast <16 x i8> [[TMP48]] to i128 // CHECK-NEXT: [[TMP52:%.*]] = tail call i128 @llvm.s390.vsbcbiq(i128 [[TMP49]], i128 [[TMP50]], i128 [[TMP51]]) // CHECK-NEXT: [[TMP53:%.*]] = bitcast i128 [[TMP52]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP53]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP54:%.*]] = load volatile <4 x i32>, ptr @vui, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP55:%.*]] = load volatile <4 x i32>, ptr @vui, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP53]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP54:%.*]] = load volatile <4 x i32>, ptr @vui, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP55:%.*]] = load volatile <4 x i32>, ptr @vui, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP56:%.*]] = tail call i128 @llvm.s390.vsumqf(<4 x i32> [[TMP54]], <4 x i32> [[TMP55]]) // CHECK-NEXT: [[TMP57:%.*]] = bitcast i128 [[TMP56]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP57]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP58:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP59:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP57]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP58:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP59:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP60:%.*]] = tail call i128 @llvm.s390.vsumqg(<2 x i64> [[TMP58]], <2 x i64> [[TMP59]]) // CHECK-NEXT: [[TMP61:%.*]] = bitcast i128 [[TMP60]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP61]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP62:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP63:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP61]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP62:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP63:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP64:%.*]] = tail call i128 @llvm.s390.vgfmg(<2 x i64> [[TMP62]], <2 x i64> [[TMP63]]) // CHECK-NEXT: [[TMP65:%.*]] = bitcast i128 [[TMP64]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP65]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP66:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP67:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP68:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP65]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP66:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP67:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP68:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP69:%.*]] = bitcast <16 x i8> [[TMP68]] to i128 // CHECK-NEXT: [[TMP70:%.*]] = tail call i128 @llvm.s390.vgfmag(<2 x i64> [[TMP66]], <2 x i64> [[TMP67]], i128 [[TMP69]]) // CHECK-NEXT: [[TMP71:%.*]] = bitcast i128 [[TMP70]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP71]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP72:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP73:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP74:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP71]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP72:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP73:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP74:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP75:%.*]] = bitcast <16 x i8> [[TMP74]] to i128 // CHECK-NEXT: [[TMP76:%.*]] = tail call i128 @llvm.s390.vmslg(<2 x i64> [[TMP72]], <2 x i64> [[TMP73]], i128 [[TMP75]], i32 0) // CHECK-NEXT: [[TMP77:%.*]] = bitcast i128 [[TMP76]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP77]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP78:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP79:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP80:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP77]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP78:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP79:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP80:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP81:%.*]] = bitcast <16 x i8> [[TMP80]] to i128 // CHECK-NEXT: [[TMP82:%.*]] = tail call i128 @llvm.s390.vmslg(<2 x i64> [[TMP78]], <2 x i64> [[TMP79]], i128 [[TMP81]], i32 4) // CHECK-NEXT: [[TMP83:%.*]] = bitcast i128 [[TMP82]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP83]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP84:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP85:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP86:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP83]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP84:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP85:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP86:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP87:%.*]] = bitcast <16 x i8> [[TMP86]] to i128 // CHECK-NEXT: [[TMP88:%.*]] = tail call i128 @llvm.s390.vmslg(<2 x i64> [[TMP84]], <2 x i64> [[TMP85]], i128 [[TMP87]], i32 8) // CHECK-NEXT: [[TMP89:%.*]] = bitcast i128 [[TMP88]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP89]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP90:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP91:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP92:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP89]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP90:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP91:%.*]] = load volatile <2 x i64>, ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP92:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP93:%.*]] = bitcast <16 x i8> [[TMP92]] to i128 // CHECK-NEXT: [[TMP94:%.*]] = tail call i128 @llvm.s390.vmslg(<2 x i64> [[TMP90]], <2 x i64> [[TMP91]], i128 [[TMP93]], i32 12) // CHECK-NEXT: [[TMP95:%.*]] = bitcast i128 [[TMP94]] to <16 x i8> -// CHECK-NEXT: store volatile <16 x i8> [[TMP95]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP96:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP97:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <16 x i8> [[TMP95]], ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP96:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP97:%.*]] = load volatile <16 x i8>, ptr @vuc, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[TMP98:%.*]] = tail call <2 x i64> @llvm.s390.vbperm(<16 x i8> [[TMP96]], <16 x i8> [[TMP97]]) -// CHECK-NEXT: store volatile <2 x i64> [[TMP98]], ptr @vul, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <2 x i64> [[TMP98]], ptr @vul, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test(void) { @@ -159,7 +159,7 @@ void test(void) { vul = vec_bperm_u128(vuc, vuc); } //. -// CHECK: [[CHAR_TBAA3]] = !{[[META4:![0-9]+]], [[META4]], i64 0} -// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} -// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META5:![0-9]+]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} +// CHECK: [[META6]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[CHAR_TBAA7]] = !{[[META5]], [[META5]], i64 0} //. diff --git a/clang/test/CodeGen/SystemZ/gnu-atomic-builtins-i128-16Al.c b/clang/test/CodeGen/SystemZ/gnu-atomic-builtins-i128-16Al.c index 5f3b0ec..6bd61f9 100644 --- a/clang/test/CodeGen/SystemZ/gnu-atomic-builtins-i128-16Al.c +++ b/clang/test/CodeGen/SystemZ/gnu-atomic-builtins-i128-16Al.c @@ -17,7 +17,7 @@ __int128 Des __attribute__((aligned(16))); // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = load atomic i128, ptr @Ptr seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP0]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2:![0-9]+]] +// CHECK-NEXT: store i128 [[TMP0]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6:![0-9]+]] // CHECK-NEXT: ret void // __int128 f1() { @@ -29,7 +29,7 @@ __int128 f1() { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = load atomic i128, ptr @Ptr seq_cst, align 16 // CHECK-NEXT: store i128 [[TMP0]], ptr @Ret, align 16 -// CHECK-NEXT: store i128 [[TMP0]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP0]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f2() { @@ -40,7 +40,7 @@ __int128 f2() { // CHECK-LABEL: define dso_local void @f3( // CHECK-SAME: ) local_unnamed_addr #[[ATTR1:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: store atomic i128 [[TMP0]], ptr @Ptr seq_cst, align 16 // CHECK-NEXT: ret void // @@ -62,9 +62,9 @@ void f4() { // CHECK-LABEL: define dso_local void @f5( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw xchg ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f5() { @@ -77,7 +77,7 @@ __int128 f5() { // CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16 // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw xchg ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 // CHECK-NEXT: store i128 [[TMP1]], ptr @Ret, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f6() { @@ -88,7 +88,7 @@ __int128 f6() { // CHECK-LABEL: define dso_local noundef zeroext i1 @f7( // CHECK-SAME: ) local_unnamed_addr #[[ATTR1]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Des, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Des, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = load i128, ptr @Exp, align 16 // CHECK-NEXT: [[TMP2:%.*]] = cmpxchg ptr @Ptr, i128 [[TMP1]], i128 [[TMP0]] seq_cst seq_cst, align 16 // CHECK-NEXT: [[TMP3:%.*]] = extractvalue { i128, i1 } [[TMP2]], 1 @@ -128,10 +128,10 @@ _Bool f8() { // CHECK-LABEL: define dso_local void @f9( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw add ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 // CHECK-NEXT: [[TMP2:%.*]] = add i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f9() { @@ -141,10 +141,10 @@ __int128 f9() { // CHECK-LABEL: define dso_local void @f10( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw sub ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 // CHECK-NEXT: [[TMP2:%.*]] = sub i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f10() { @@ -154,10 +154,10 @@ __int128 f10() { // CHECK-LABEL: define dso_local void @f11( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw and ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 // CHECK-NEXT: [[TMP2:%.*]] = and i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f11() { @@ -167,10 +167,10 @@ __int128 f11() { // CHECK-LABEL: define dso_local void @f12( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw xor ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 // CHECK-NEXT: [[TMP2:%.*]] = xor i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f12() { @@ -180,10 +180,10 @@ __int128 f12() { // CHECK-LABEL: define dso_local void @f13( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw or ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 // CHECK-NEXT: [[TMP2:%.*]] = or i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f13() { @@ -193,11 +193,11 @@ __int128 f13() { // CHECK-LABEL: define dso_local void @f14( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw nand ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 // CHECK-NEXT: [[TMP2:%.*]] = and i128 [[TMP1]], [[TMP0]] // CHECK-NEXT: [[TMP3:%.*]] = xor i128 [[TMP2]], -1 -// CHECK-NEXT: store i128 [[TMP3]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP3]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f14() { @@ -207,9 +207,9 @@ __int128 f14() { // CHECK-LABEL: define dso_local void @f15( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw add ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f15() { @@ -219,9 +219,9 @@ __int128 f15() { // CHECK-LABEL: define dso_local void @f16( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw sub ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f16() { @@ -231,9 +231,9 @@ __int128 f16() { // CHECK-LABEL: define dso_local void @f17( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw and ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f17() { @@ -243,9 +243,9 @@ __int128 f17() { // CHECK-LABEL: define dso_local void @f18( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw xor ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f18() { @@ -255,9 +255,9 @@ __int128 f18() { // CHECK-LABEL: define dso_local void @f19( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw or ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f19() { @@ -267,17 +267,17 @@ __int128 f19() { // CHECK-LABEL: define dso_local void @f20( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw nand ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f20() { return __atomic_fetch_nand(&Ptr, Val, memory_order_seq_cst); } //. -// CHECK: [[__INT128_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"__int128", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[__INT128_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK: [[META7]] = !{!"__int128", [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/SystemZ/gnu-atomic-builtins-i128-8Al.c b/clang/test/CodeGen/SystemZ/gnu-atomic-builtins-i128-8Al.c index 3ac5959..332d7bd 100644 --- a/clang/test/CodeGen/SystemZ/gnu-atomic-builtins-i128-8Al.c +++ b/clang/test/CodeGen/SystemZ/gnu-atomic-builtins-i128-8Al.c @@ -22,7 +22,7 @@ __int128 Des; // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = load atomic i128, ptr @Ptr seq_cst, align 8 -// CHECK-NEXT: store i128 [[TMP0]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2:![0-9]+]] +// CHECK-NEXT: store i128 [[TMP0]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6:![0-9]+]] // CHECK-NEXT: ret void // __int128 f1() { @@ -34,7 +34,7 @@ __int128 f1() { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = load atomic i128, ptr @Ptr seq_cst, align 8 // CHECK-NEXT: store i128 [[TMP0]], ptr @Ret, align 8 -// CHECK-NEXT: store i128 [[TMP0]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP0]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f2() { @@ -45,7 +45,7 @@ __int128 f2() { // CHECK-LABEL: define dso_local void @f3( // CHECK-SAME: ) local_unnamed_addr #[[ATTR1:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: store atomic i128 [[TMP0]], ptr @Ptr seq_cst, align 8 // CHECK-NEXT: ret void // @@ -67,9 +67,9 @@ void f4() { // CHECK-LABEL: define dso_local void @f5( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw xchg ptr @Ptr, i128 [[TMP0]] seq_cst, align 8 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f5() { @@ -82,7 +82,7 @@ __int128 f5() { // CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8 // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw xchg ptr @Ptr, i128 [[TMP0]] seq_cst, align 8 // CHECK-NEXT: store i128 [[TMP1]], ptr @Ret, align 8 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f6() { @@ -93,7 +93,7 @@ __int128 f6() { // CHECK-LABEL: define dso_local noundef zeroext i1 @f7( // CHECK-SAME: ) local_unnamed_addr #[[ATTR1]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Des, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Des, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = load i128, ptr @Exp, align 8 // CHECK-NEXT: [[TMP2:%.*]] = cmpxchg ptr @Ptr, i128 [[TMP1]], i128 [[TMP0]] seq_cst seq_cst, align 8 // CHECK-NEXT: [[TMP3:%.*]] = extractvalue { i128, i1 } [[TMP2]], 1 @@ -133,10 +133,10 @@ _Bool f8() { // CHECK-LABEL: define dso_local void @f9( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw add ptr @Ptr, i128 [[TMP0]] seq_cst, align 8 // CHECK-NEXT: [[TMP2:%.*]] = add i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f9() { @@ -146,10 +146,10 @@ __int128 f9() { // CHECK-LABEL: define dso_local void @f10( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw sub ptr @Ptr, i128 [[TMP0]] seq_cst, align 8 // CHECK-NEXT: [[TMP2:%.*]] = sub i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f10() { @@ -159,10 +159,10 @@ __int128 f10() { // CHECK-LABEL: define dso_local void @f11( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw and ptr @Ptr, i128 [[TMP0]] seq_cst, align 8 // CHECK-NEXT: [[TMP2:%.*]] = and i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f11() { @@ -172,10 +172,10 @@ __int128 f11() { // CHECK-LABEL: define dso_local void @f12( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw xor ptr @Ptr, i128 [[TMP0]] seq_cst, align 8 // CHECK-NEXT: [[TMP2:%.*]] = xor i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f12() { @@ -185,10 +185,10 @@ __int128 f12() { // CHECK-LABEL: define dso_local void @f13( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw or ptr @Ptr, i128 [[TMP0]] seq_cst, align 8 // CHECK-NEXT: [[TMP2:%.*]] = or i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f13() { @@ -198,11 +198,11 @@ __int128 f13() { // CHECK-LABEL: define dso_local void @f14( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw nand ptr @Ptr, i128 [[TMP0]] seq_cst, align 8 // CHECK-NEXT: [[TMP2:%.*]] = and i128 [[TMP1]], [[TMP0]] // CHECK-NEXT: [[TMP3:%.*]] = xor i128 [[TMP2]], -1 -// CHECK-NEXT: store i128 [[TMP3]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP3]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f14() { @@ -212,9 +212,9 @@ __int128 f14() { // CHECK-LABEL: define dso_local void @f15( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw add ptr @Ptr, i128 [[TMP0]] seq_cst, align 8 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f15() { @@ -224,9 +224,9 @@ __int128 f15() { // CHECK-LABEL: define dso_local void @f16( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw sub ptr @Ptr, i128 [[TMP0]] seq_cst, align 8 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f16() { @@ -236,9 +236,9 @@ __int128 f16() { // CHECK-LABEL: define dso_local void @f17( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw and ptr @Ptr, i128 [[TMP0]] seq_cst, align 8 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f17() { @@ -248,9 +248,9 @@ __int128 f17() { // CHECK-LABEL: define dso_local void @f18( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw xor ptr @Ptr, i128 [[TMP0]] seq_cst, align 8 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f18() { @@ -260,9 +260,9 @@ __int128 f18() { // CHECK-LABEL: define dso_local void @f19( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw or ptr @Ptr, i128 [[TMP0]] seq_cst, align 8 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f19() { @@ -272,17 +272,17 @@ __int128 f19() { // CHECK-LABEL: define dso_local void @f20( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw nand ptr @Ptr, i128 [[TMP0]] seq_cst, align 8 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f20() { return __atomic_fetch_nand(&Ptr, Val, memory_order_seq_cst); } //. -// CHECK: [[__INT128_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"__int128", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[__INT128_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK: [[META7]] = !{!"__int128", [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/SystemZ/sync-builtins-i128-16Al.c b/clang/test/CodeGen/SystemZ/sync-builtins-i128-16Al.c index 601bd7f..ba77cdb 100644 --- a/clang/test/CodeGen/SystemZ/sync-builtins-i128-16Al.c +++ b/clang/test/CodeGen/SystemZ/sync-builtins-i128-16Al.c @@ -13,9 +13,9 @@ __int128 OldVal __attribute__((aligned(16))); // CHECK-LABEL: define dso_local void @f1( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6:![0-9]+]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw add ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f1() { @@ -25,9 +25,9 @@ __int128 f1() { // CHECK-LABEL: define dso_local void @f2( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw sub ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f2() { @@ -37,9 +37,9 @@ __int128 f2() { // CHECK-LABEL: define dso_local void @f3( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw or ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f3() { @@ -49,9 +49,9 @@ __int128 f3() { // CHECK-LABEL: define dso_local void @f4( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw and ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f4() { @@ -61,9 +61,9 @@ __int128 f4() { // CHECK-LABEL: define dso_local void @f5( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw xor ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f5() { @@ -73,9 +73,9 @@ __int128 f5() { // CHECK-LABEL: define dso_local void @f6( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw nand ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f6() { @@ -85,10 +85,10 @@ __int128 f6() { // CHECK-LABEL: define dso_local void @f7( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw add ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 // CHECK-NEXT: [[TMP2:%.*]] = add i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f7() { @@ -98,10 +98,10 @@ __int128 f7() { // CHECK-LABEL: define dso_local void @f8( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw sub ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 // CHECK-NEXT: [[TMP2:%.*]] = sub i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f8() { @@ -111,10 +111,10 @@ __int128 f8() { // CHECK-LABEL: define dso_local void @f9( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw or ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 // CHECK-NEXT: [[TMP2:%.*]] = or i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f9() { @@ -124,10 +124,10 @@ __int128 f9() { // CHECK-LABEL: define dso_local void @f10( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw and ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 // CHECK-NEXT: [[TMP2:%.*]] = and i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f10() { @@ -137,10 +137,10 @@ __int128 f10() { // CHECK-LABEL: define dso_local void @f11( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw xor ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 // CHECK-NEXT: [[TMP2:%.*]] = xor i128 [[TMP1]], [[TMP0]] -// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP2]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f11() { @@ -150,11 +150,11 @@ __int128 f11() { // CHECK-LABEL: define dso_local void @f12( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw nand ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 // CHECK-NEXT: [[TMP2:%.*]] = and i128 [[TMP1]], [[TMP0]] // CHECK-NEXT: [[TMP3:%.*]] = xor i128 [[TMP2]], -1 -// CHECK-NEXT: store i128 [[TMP3]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP3]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f12() { @@ -164,8 +164,8 @@ __int128 f12() { // CHECK-LABEL: define dso_local zeroext i1 @f13( // CHECK-SAME: ) local_unnamed_addr #[[ATTR1:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @OldVal, align 16, !tbaa [[__INT128_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @OldVal, align 16, !tbaa [[__INT128_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = cmpxchg ptr @Ptr, i128 [[TMP0]], i128 [[TMP1]] seq_cst seq_cst, align 16 // CHECK-NEXT: [[TMP3:%.*]] = extractvalue { i128, i1 } [[TMP2]], 1 // CHECK-NEXT: ret i1 [[TMP3]] @@ -177,11 +177,11 @@ _Bool f13() { // CHECK-LABEL: define dso_local void @f14( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @OldVal, align 16, !tbaa [[__INT128_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @OldVal, align 16, !tbaa [[__INT128_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP2:%.*]] = cmpxchg ptr @Ptr, i128 [[TMP0]], i128 [[TMP1]] seq_cst seq_cst, align 16 // CHECK-NEXT: [[TMP3:%.*]] = extractvalue { i128, i1 } [[TMP2]], 0 -// CHECK-NEXT: store i128 [[TMP3]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP3]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f14() { @@ -191,9 +191,9 @@ __int128 f14() { // CHECK-LABEL: define dso_local void @f15( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw xchg ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f15() { @@ -213,9 +213,9 @@ void f16() { // CHECK-LABEL: define dso_local void @f17( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(i128) align 8 captures(none) initializes((0, 16)) [[AGG_RESULT:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i128, ptr @Val, align 16, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = atomicrmw xchg ptr @Ptr, i128 [[TMP0]] seq_cst, align 16 -// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: store i128 [[TMP1]], ptr [[AGG_RESULT]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: ret void // __int128 f17() { @@ -227,8 +227,8 @@ __int128 f17() { // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR2:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[T_ADDR:%.*]] = alloca i128, align 8 -// CHECK-NEXT: [[T:%.*]] = load i128, ptr [[TMP0]], align 8, !tbaa [[__INT128_TBAA2]] -// CHECK-NEXT: store i128 [[T]], ptr [[T_ADDR]], align 8, !tbaa [[__INT128_TBAA2]] +// CHECK-NEXT: [[T:%.*]] = load i128, ptr [[TMP0]], align 8, !tbaa [[__INT128_TBAA6]] +// CHECK-NEXT: store i128 [[T]], ptr [[T_ADDR]], align 8, !tbaa [[__INT128_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = cmpxchg ptr [[T_ADDR]], i128 [[T]], i128 [[T]] seq_cst seq_cst, align 16 // CHECK-NEXT: ret void // @@ -236,8 +236,8 @@ void f18(__int128 t) { __sync_bool_compare_and_swap(({int x = 1; &t;}), t, t); } //. -// CHECK: [[__INT128_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"__int128", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[__INT128_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK: [[META7]] = !{!"__int128", [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/SystemZ/zvector2.c b/clang/test/CodeGen/SystemZ/zvector2.c index f00fcdd..b799467 100644 --- a/clang/test/CodeGen/SystemZ/zvector2.c +++ b/clang/test/CodeGen/SystemZ/zvector2.c @@ -8,8 +8,8 @@ volatile vector bool int bi; // CHECK-LABEL: define dso_local void @test_assign( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3:![0-9]+]] -// CHECK-NEXT: store volatile <4 x float> [[TMP0]], ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7:![0-9]+]] +// CHECK-NEXT: store volatile <4 x float> [[TMP0]], ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_assign (void) @@ -20,8 +20,8 @@ void test_assign (void) // CHECK-LABEL: define dso_local void @test_pos( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: store volatile <4 x float> [[TMP0]], ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: store volatile <4 x float> [[TMP0]], ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_pos (void) @@ -32,9 +32,9 @@ void test_pos (void) // CHECK-LABEL: define dso_local void @test_neg( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[FNEG:%.*]] = fneg <4 x float> [[TMP0]] -// CHECK-NEXT: store volatile <4 x float> [[FNEG]], ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x float> [[FNEG]], ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_neg (void) @@ -45,9 +45,9 @@ void test_neg (void) // CHECK-LABEL: define dso_local void @test_preinc( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[INC:%.*]] = fadd <4 x float> [[TMP0]], splat (float 1.000000e+00) -// CHECK-NEXT: store volatile <4 x float> [[INC]], ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x float> [[INC]], ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_preinc (void) @@ -58,9 +58,9 @@ void test_preinc (void) // CHECK-LABEL: define dso_local void @test_postinc( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[INC:%.*]] = fadd <4 x float> [[TMP0]], splat (float 1.000000e+00) -// CHECK-NEXT: store volatile <4 x float> [[INC]], ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x float> [[INC]], ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_postinc (void) @@ -71,9 +71,9 @@ void test_postinc (void) // CHECK-LABEL: define dso_local void @test_predec( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[DEC:%.*]] = fadd <4 x float> [[TMP0]], splat (float -1.000000e+00) -// CHECK-NEXT: store volatile <4 x float> [[DEC]], ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x float> [[DEC]], ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_predec (void) @@ -84,9 +84,9 @@ void test_predec (void) // CHECK-LABEL: define dso_local void @test_postdec( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[DEC:%.*]] = fadd <4 x float> [[TMP0]], splat (float -1.000000e+00) -// CHECK-NEXT: store volatile <4 x float> [[DEC]], ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x float> [[DEC]], ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_postdec (void) @@ -97,10 +97,10 @@ void test_postdec (void) // CHECK-LABEL: define dso_local void @test_add( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[ADD:%.*]] = fadd <4 x float> [[TMP0]], [[TMP1]] -// CHECK-NEXT: store volatile <4 x float> [[ADD]], ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x float> [[ADD]], ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_add (void) @@ -111,10 +111,10 @@ void test_add (void) // CHECK-LABEL: define dso_local void @test_add_assign( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[ADD:%.*]] = fadd <4 x float> [[TMP0]], [[TMP1]] -// CHECK-NEXT: store volatile <4 x float> [[ADD]], ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x float> [[ADD]], ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_add_assign (void) @@ -125,10 +125,10 @@ void test_add_assign (void) // CHECK-LABEL: define dso_local void @test_sub( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[SUB:%.*]] = fsub <4 x float> [[TMP0]], [[TMP1]] -// CHECK-NEXT: store volatile <4 x float> [[SUB]], ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x float> [[SUB]], ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_sub (void) @@ -139,10 +139,10 @@ void test_sub (void) // CHECK-LABEL: define dso_local void @test_sub_assign( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[SUB:%.*]] = fsub <4 x float> [[TMP1]], [[TMP0]] -// CHECK-NEXT: store volatile <4 x float> [[SUB]], ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x float> [[SUB]], ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_sub_assign (void) @@ -153,10 +153,10 @@ void test_sub_assign (void) // CHECK-LABEL: define dso_local void @test_mul( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[MUL:%.*]] = fmul <4 x float> [[TMP0]], [[TMP1]] -// CHECK-NEXT: store volatile <4 x float> [[MUL]], ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x float> [[MUL]], ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_mul (void) @@ -167,10 +167,10 @@ void test_mul (void) // CHECK-LABEL: define dso_local void @test_mul_assign( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[MUL:%.*]] = fmul <4 x float> [[TMP0]], [[TMP1]] -// CHECK-NEXT: store volatile <4 x float> [[MUL]], ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x float> [[MUL]], ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_mul_assign (void) @@ -181,10 +181,10 @@ void test_mul_assign (void) // CHECK-LABEL: define dso_local void @test_div( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[DIV:%.*]] = fdiv <4 x float> [[TMP0]], [[TMP1]] -// CHECK-NEXT: store volatile <4 x float> [[DIV]], ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x float> [[DIV]], ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_div (void) @@ -195,10 +195,10 @@ void test_div (void) // CHECK-LABEL: define dso_local void @test_div_assign( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[DIV:%.*]] = fdiv <4 x float> [[TMP1]], [[TMP0]] -// CHECK-NEXT: store volatile <4 x float> [[DIV]], ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x float> [[DIV]], ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_div_assign (void) @@ -209,11 +209,11 @@ void test_div_assign (void) // CHECK-LABEL: define dso_local void @test_cmpeq( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[CMP:%.*]] = fcmp oeq <4 x float> [[TMP0]], [[TMP1]] // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i32> -// CHECK-NEXT: store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_cmpeq (void) @@ -224,11 +224,11 @@ void test_cmpeq (void) // CHECK-LABEL: define dso_local void @test_cmpne( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[CMP:%.*]] = fcmp une <4 x float> [[TMP0]], [[TMP1]] // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i32> -// CHECK-NEXT: store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_cmpne (void) @@ -239,11 +239,11 @@ void test_cmpne (void) // CHECK-LABEL: define dso_local void @test_cmpge( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[CMP:%.*]] = fcmp oge <4 x float> [[TMP0]], [[TMP1]] // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i32> -// CHECK-NEXT: store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_cmpge (void) @@ -254,11 +254,11 @@ void test_cmpge (void) // CHECK-LABEL: define dso_local void @test_cmpgt( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[CMP:%.*]] = fcmp ogt <4 x float> [[TMP0]], [[TMP1]] // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i32> -// CHECK-NEXT: store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_cmpgt (void) @@ -269,11 +269,11 @@ void test_cmpgt (void) // CHECK-LABEL: define dso_local void @test_cmple( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[CMP:%.*]] = fcmp ole <4 x float> [[TMP0]], [[TMP1]] // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i32> -// CHECK-NEXT: store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_cmple (void) @@ -284,11 +284,11 @@ void test_cmple (void) // CHECK-LABEL: define dso_local void @test_cmplt( // CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA3]] -// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[CHAR_TBAA7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: [[CMP:%.*]] = fcmp olt <4 x float> [[TMP0]], [[TMP1]] // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i32> -// CHECK-NEXT: store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[CHAR_TBAA3]] +// CHECK-NEXT: store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[CHAR_TBAA7]] // CHECK-NEXT: ret void // void test_cmplt (void) @@ -297,7 +297,7 @@ void test_cmplt (void) } //. -// CHECK: [[CHAR_TBAA3]] = !{[[META4:![0-9]+]], [[META4]], i64 0} -// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} -// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META5:![0-9]+]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} +// CHECK: [[META6]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[CHAR_TBAA7]] = !{[[META5]], [[META5]], i64 0} //. diff --git a/clang/test/CodeGen/allow-ubsan-check.c b/clang/test/CodeGen/allow-ubsan-check.c index 8d30e29..1e12885 100644 --- a/clang/test/CodeGen/allow-ubsan-check.c +++ b/clang/test/CodeGen/allow-ubsan-check.c @@ -10,27 +10,27 @@ // CHECK-LABEL: define dso_local noundef i32 @div( // CHECK-SAME: i32 noundef [[X:%.*]], i32 noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = icmp ne i32 [[Y]], 0, !nosanitize [[META2:![0-9]+]] -// CHECK-NEXT: [[TMP1:%.*]] = icmp ne i32 [[X]], -2147483648, !nosanitize [[META2]] -// CHECK-NEXT: [[TMP2:%.*]] = icmp ne i32 [[Y]], -1, !nosanitize [[META2]] -// CHECK-NEXT: [[OR:%.*]] = or i1 [[TMP1]], [[TMP2]], !nosanitize [[META2]] +// CHECK-NEXT: [[TMP0:%.*]] = icmp ne i32 [[Y]], 0, !nosanitize [[META6:![0-9]+]] +// CHECK-NEXT: [[TMP1:%.*]] = icmp ne i32 [[X]], -2147483648, !nosanitize [[META6]] +// CHECK-NEXT: [[TMP2:%.*]] = icmp ne i32 [[Y]], -1, !nosanitize [[META6]] +// CHECK-NEXT: [[OR:%.*]] = or i1 [[TMP1]], [[TMP2]], !nosanitize [[META6]] // // 27 == SO_IntegerDivideByZero -// CHECK-NEXT: [[TMP3:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 27), !nosanitize [[META2]] -// CHECK-NEXT: [[TMP4:%.*]] = xor i1 [[TMP3]], true, !nosanitize [[META2]] -// CHECK-NEXT: [[TMP5:%.*]] = or i1 [[TMP0]], [[TMP4]], !nosanitize [[META2]] +// CHECK-NEXT: [[TMP3:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 27), !nosanitize [[META6]] +// CHECK-NEXT: [[TMP4:%.*]] = xor i1 [[TMP3]], true, !nosanitize [[META6]] +// CHECK-NEXT: [[TMP5:%.*]] = or i1 [[TMP0]], [[TMP4]], !nosanitize [[META6]] // // 41 == SO_SignedIntegerOverflow -// CHECK-NEXT: [[TMP6:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 41), !nosanitize [[META2]] -// CHECK-NEXT: [[TMP7:%.*]] = xor i1 [[TMP6]], true, !nosanitize [[META2]] -// CHECK-NEXT: [[TMP8:%.*]] = or i1 [[OR]], [[TMP7]], !nosanitize [[META2]] -// CHECK-NEXT: [[TMP9:%.*]] = and i1 [[TMP5]], [[TMP8]], !nosanitize [[META2]] -// CHECK-NEXT: br i1 [[TMP9]], label %[[CONT:.*]], label %[[HANDLER_DIVREM_OVERFLOW:.*]], !prof [[PROF3:![0-9]+]], !nosanitize [[META2]] +// CHECK-NEXT: [[TMP6:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 41), !nosanitize [[META6]] +// CHECK-NEXT: [[TMP7:%.*]] = xor i1 [[TMP6]], true, !nosanitize [[META6]] +// CHECK-NEXT: [[TMP8:%.*]] = or i1 [[OR]], [[TMP7]], !nosanitize [[META6]] +// CHECK-NEXT: [[TMP9:%.*]] = and i1 [[TMP5]], [[TMP8]], !nosanitize [[META6]] +// CHECK-NEXT: br i1 [[TMP9]], label %[[CONT:.*]], label %[[HANDLER_DIVREM_OVERFLOW:.*]], !prof [[PROF7:![0-9]+]], !nosanitize [[META6]] // CHECK: [[HANDLER_DIVREM_OVERFLOW]]: -// CHECK-NEXT: [[TMP10:%.*]] = zext i32 [[X]] to i64, !nosanitize [[META2]] -// CHECK-NEXT: [[TMP11:%.*]] = zext i32 [[Y]] to i64, !nosanitize [[META2]] -// CHECK-NEXT: tail call void @__ubsan_handle_divrem_overflow_abort(ptr nonnull @[[GLOB1:[0-9]+]], i64 [[TMP10]], i64 [[TMP11]]) #[[ATTR6:[0-9]+]], !nosanitize [[META2]] -// CHECK-NEXT: unreachable, !nosanitize [[META2]] +// CHECK-NEXT: [[TMP10:%.*]] = zext i32 [[X]] to i64, !nosanitize [[META6]] +// CHECK-NEXT: [[TMP11:%.*]] = zext i32 [[Y]] to i64, !nosanitize [[META6]] +// CHECK-NEXT: tail call void @__ubsan_handle_divrem_overflow_abort(ptr nonnull @[[GLOB1:[0-9]+]], i64 [[TMP10]], i64 [[TMP11]]) #[[ATTR6:[0-9]+]], !nosanitize [[META6]] +// CHECK-NEXT: unreachable, !nosanitize [[META6]] // CHECK: [[CONT]]: // CHECK-NEXT: [[DIV:%.*]] = sdiv i32 [[X]], [[Y]] // CHECK-NEXT: ret i32 [[DIV]] @@ -38,21 +38,21 @@ // TR-LABEL: define dso_local noundef i32 @div( // TR-SAME: i32 noundef [[X:%.*]], i32 noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // TR-NEXT: [[ENTRY:.*:]] -// TR-NEXT: [[TMP0:%.*]] = icmp ne i32 [[Y]], 0, !nosanitize [[META2:![0-9]+]] -// TR-NEXT: [[TMP1:%.*]] = icmp ne i32 [[X]], -2147483648, !nosanitize [[META2]] -// TR-NEXT: [[TMP2:%.*]] = icmp ne i32 [[Y]], -1, !nosanitize [[META2]] -// TR-NEXT: [[OR:%.*]] = or i1 [[TMP1]], [[TMP2]], !nosanitize [[META2]] -// TR-NEXT: [[TMP3:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 27), !nosanitize [[META2]] -// TR-NEXT: [[TMP4:%.*]] = xor i1 [[TMP3]], true, !nosanitize [[META2]] -// TR-NEXT: [[TMP5:%.*]] = or i1 [[TMP0]], [[TMP4]], !nosanitize [[META2]] -// TR-NEXT: [[TMP6:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 41), !nosanitize [[META2]] -// TR-NEXT: [[TMP7:%.*]] = xor i1 [[TMP6]], true, !nosanitize [[META2]] -// TR-NEXT: [[TMP8:%.*]] = or i1 [[OR]], [[TMP7]], !nosanitize [[META2]] -// TR-NEXT: [[TMP9:%.*]] = and i1 [[TMP5]], [[TMP8]], !nosanitize [[META2]] -// TR-NEXT: br i1 [[TMP9]], label %[[CONT:.*]], label %[[TRAP:.*]], !prof [[PROF3:![0-9]+]], !nosanitize [[META2]] +// TR-NEXT: [[TMP0:%.*]] = icmp ne i32 [[Y]], 0, !nosanitize [[META6:![0-9]+]] +// TR-NEXT: [[TMP1:%.*]] = icmp ne i32 [[X]], -2147483648, !nosanitize [[META6]] +// TR-NEXT: [[TMP2:%.*]] = icmp ne i32 [[Y]], -1, !nosanitize [[META6]] +// TR-NEXT: [[OR:%.*]] = or i1 [[TMP1]], [[TMP2]], !nosanitize [[META6]] +// TR-NEXT: [[TMP3:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 27), !nosanitize [[META6]] +// TR-NEXT: [[TMP4:%.*]] = xor i1 [[TMP3]], true, !nosanitize [[META6]] +// TR-NEXT: [[TMP5:%.*]] = or i1 [[TMP0]], [[TMP4]], !nosanitize [[META6]] +// TR-NEXT: [[TMP6:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 41), !nosanitize [[META6]] +// TR-NEXT: [[TMP7:%.*]] = xor i1 [[TMP6]], true, !nosanitize [[META6]] +// TR-NEXT: [[TMP8:%.*]] = or i1 [[OR]], [[TMP7]], !nosanitize [[META6]] +// TR-NEXT: [[TMP9:%.*]] = and i1 [[TMP5]], [[TMP8]], !nosanitize [[META6]] +// TR-NEXT: br i1 [[TMP9]], label %[[CONT:.*]], label %[[TRAP:.*]], !prof [[PROF7:![0-9]+]], !nosanitize [[META6]] // TR: [[TRAP]]: -// TR-NEXT: tail call void @llvm.ubsantrap(i8 3) #[[ATTR7:[0-9]+]], !nosanitize [[META2]] -// TR-NEXT: unreachable, !nosanitize [[META2]] +// TR-NEXT: tail call void @llvm.ubsantrap(i8 3) #[[ATTR7:[0-9]+]], !nosanitize [[META6]] +// TR-NEXT: unreachable, !nosanitize [[META6]] // TR: [[CONT]]: // TR-NEXT: [[DIV:%.*]] = sdiv i32 [[X]], [[Y]] // TR-NEXT: ret i32 [[DIV]] @@ -60,23 +60,23 @@ // REC-LABEL: define dso_local noundef i32 @div( // REC-SAME: i32 noundef [[X:%.*]], i32 noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // REC-NEXT: [[ENTRY:.*:]] -// REC-NEXT: [[TMP0:%.*]] = icmp ne i32 [[Y]], 0, !nosanitize [[META2:![0-9]+]] -// REC-NEXT: [[TMP1:%.*]] = icmp ne i32 [[X]], -2147483648, !nosanitize [[META2]] -// REC-NEXT: [[TMP2:%.*]] = icmp ne i32 [[Y]], -1, !nosanitize [[META2]] -// REC-NEXT: [[OR:%.*]] = or i1 [[TMP1]], [[TMP2]], !nosanitize [[META2]] -// REC-NEXT: [[TMP3:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 27), !nosanitize [[META2]] -// REC-NEXT: [[TMP4:%.*]] = xor i1 [[TMP3]], true, !nosanitize [[META2]] -// REC-NEXT: [[TMP5:%.*]] = or i1 [[TMP0]], [[TMP4]], !nosanitize [[META2]] -// REC-NEXT: [[TMP6:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 41), !nosanitize [[META2]] -// REC-NEXT: [[TMP7:%.*]] = xor i1 [[TMP6]], true, !nosanitize [[META2]] -// REC-NEXT: [[TMP8:%.*]] = or i1 [[OR]], [[TMP7]], !nosanitize [[META2]] -// REC-NEXT: [[TMP9:%.*]] = and i1 [[TMP5]], [[TMP8]], !nosanitize [[META2]] -// REC-NEXT: br i1 [[TMP9]], label %[[CONT:.*]], label %[[HANDLER_DIVREM_OVERFLOW:.*]], !prof [[PROF3:![0-9]+]], !nosanitize [[META2]] +// REC-NEXT: [[TMP0:%.*]] = icmp ne i32 [[Y]], 0, !nosanitize [[META6:![0-9]+]] +// REC-NEXT: [[TMP1:%.*]] = icmp ne i32 [[X]], -2147483648, !nosanitize [[META6]] +// REC-NEXT: [[TMP2:%.*]] = icmp ne i32 [[Y]], -1, !nosanitize [[META6]] +// REC-NEXT: [[OR:%.*]] = or i1 [[TMP1]], [[TMP2]], !nosanitize [[META6]] +// REC-NEXT: [[TMP3:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 27), !nosanitize [[META6]] +// REC-NEXT: [[TMP4:%.*]] = xor i1 [[TMP3]], true, !nosanitize [[META6]] +// REC-NEXT: [[TMP5:%.*]] = or i1 [[TMP0]], [[TMP4]], !nosanitize [[META6]] +// REC-NEXT: [[TMP6:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 41), !nosanitize [[META6]] +// REC-NEXT: [[TMP7:%.*]] = xor i1 [[TMP6]], true, !nosanitize [[META6]] +// REC-NEXT: [[TMP8:%.*]] = or i1 [[OR]], [[TMP7]], !nosanitize [[META6]] +// REC-NEXT: [[TMP9:%.*]] = and i1 [[TMP5]], [[TMP8]], !nosanitize [[META6]] +// REC-NEXT: br i1 [[TMP9]], label %[[CONT:.*]], label %[[HANDLER_DIVREM_OVERFLOW:.*]], !prof [[PROF7:![0-9]+]], !nosanitize [[META6]] // REC: [[HANDLER_DIVREM_OVERFLOW]]: -// REC-NEXT: [[TMP10:%.*]] = zext i32 [[X]] to i64, !nosanitize [[META2]] -// REC-NEXT: [[TMP11:%.*]] = zext i32 [[Y]] to i64, !nosanitize [[META2]] -// REC-NEXT: tail call void @__ubsan_handle_divrem_overflow(ptr nonnull @[[GLOB1:[0-9]+]], i64 [[TMP10]], i64 [[TMP11]]) #[[ATTR6:[0-9]+]], !nosanitize [[META2]] -// REC-NEXT: br label %[[CONT]], !nosanitize [[META2]] +// REC-NEXT: [[TMP10:%.*]] = zext i32 [[X]] to i64, !nosanitize [[META6]] +// REC-NEXT: [[TMP11:%.*]] = zext i32 [[Y]] to i64, !nosanitize [[META6]] +// REC-NEXT: tail call void @__ubsan_handle_divrem_overflow(ptr nonnull @[[GLOB1:[0-9]+]], i64 [[TMP10]], i64 [[TMP11]]) #[[ATTR6:[0-9]+]], !nosanitize [[META6]] +// REC-NEXT: br label %[[CONT]], !nosanitize [[META6]] // REC: [[CONT]]: // REC-NEXT: [[DIV:%.*]] = sdiv i32 [[X]], [[Y]] // REC-NEXT: ret i32 [[DIV]] @@ -88,45 +88,45 @@ int div(int x, int y) { // CHECK-LABEL: define dso_local i32 @null( // CHECK-SAME: ptr noundef readonly captures(address_is_null) [[X:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = icmp eq ptr [[X]], null, !nosanitize [[META2]] +// CHECK-NEXT: [[TMP0:%.*]] = icmp eq ptr [[X]], null, !nosanitize [[META6]] // // 29 == SO_Null -// CHECK-NEXT: [[TMP1:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 29), !nosanitize [[META2]] +// CHECK-NEXT: [[TMP1:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 29), !nosanitize [[META6]] // CHECK-NEXT: [[DOTNOT1:%.*]] = and i1 [[TMP0]], [[TMP1]] -// CHECK-NEXT: br i1 [[DOTNOT1]], label %[[HANDLER_TYPE_MISMATCH:.*]], label %[[CONT:.*]], !prof [[PROF4:![0-9]+]], !nosanitize [[META2]] +// CHECK-NEXT: br i1 [[DOTNOT1]], label %[[HANDLER_TYPE_MISMATCH:.*]], label %[[CONT:.*]], !prof [[PROF8:![0-9]+]], !nosanitize [[META6]] // CHECK: [[HANDLER_TYPE_MISMATCH]]: -// CHECK-NEXT: tail call void @__ubsan_handle_type_mismatch_v1_abort(ptr nonnull @[[GLOB2:[0-9]+]], i64 0) #[[ATTR6]], !nosanitize [[META2]] -// CHECK-NEXT: unreachable, !nosanitize [[META2]] +// CHECK-NEXT: tail call void @__ubsan_handle_type_mismatch_v1_abort(ptr nonnull @[[GLOB2:[0-9]+]], i64 0) #[[ATTR6]], !nosanitize [[META6]] +// CHECK-NEXT: unreachable, !nosanitize [[META6]] // CHECK: [[CONT]]: -// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[X]], align 4, !tbaa [[INT_TBAA5:![0-9]+]] +// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[X]], align 4, !tbaa [[INT_TBAA2:![0-9]+]] // CHECK-NEXT: ret i32 [[TMP2]] // // TR-LABEL: define dso_local i32 @null( // TR-SAME: ptr noundef readonly captures(address_is_null) [[X:%.*]]) local_unnamed_addr #[[ATTR3:[0-9]+]] { // TR-NEXT: [[ENTRY:.*:]] -// TR-NEXT: [[TMP0:%.*]] = icmp eq ptr [[X]], null, !nosanitize [[META2]] -// TR-NEXT: [[TMP1:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 29), !nosanitize [[META2]] +// TR-NEXT: [[TMP0:%.*]] = icmp eq ptr [[X]], null, !nosanitize [[META6]] +// TR-NEXT: [[TMP1:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 29), !nosanitize [[META6]] // TR-NEXT: [[DOTNOT1:%.*]] = and i1 [[TMP0]], [[TMP1]] -// TR-NEXT: br i1 [[DOTNOT1]], label %[[TRAP:.*]], label %[[CONT:.*]], !prof [[PROF4:![0-9]+]], !nosanitize [[META2]] +// TR-NEXT: br i1 [[DOTNOT1]], label %[[TRAP:.*]], label %[[CONT:.*]], !prof [[PROF8:![0-9]+]], !nosanitize [[META6]] // TR: [[TRAP]]: -// TR-NEXT: tail call void @llvm.ubsantrap(i8 22) #[[ATTR7]], !nosanitize [[META2]] -// TR-NEXT: unreachable, !nosanitize [[META2]] +// TR-NEXT: tail call void @llvm.ubsantrap(i8 22) #[[ATTR7]], !nosanitize [[META6]] +// TR-NEXT: unreachable, !nosanitize [[META6]] // TR: [[CONT]]: -// TR-NEXT: [[TMP2:%.*]] = load i32, ptr [[X]], align 4, !tbaa [[INT_TBAA5:![0-9]+]] +// TR-NEXT: [[TMP2:%.*]] = load i32, ptr [[X]], align 4, !tbaa [[INT_TBAA2:![0-9]+]] // TR-NEXT: ret i32 [[TMP2]] // // REC-LABEL: define dso_local i32 @null( // REC-SAME: ptr noundef readonly captures(address_is_null) [[X:%.*]]) local_unnamed_addr #[[ATTR0]] { // REC-NEXT: [[ENTRY:.*:]] -// REC-NEXT: [[TMP0:%.*]] = icmp eq ptr [[X]], null, !nosanitize [[META2]] -// REC-NEXT: [[TMP1:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 29), !nosanitize [[META2]] +// REC-NEXT: [[TMP0:%.*]] = icmp eq ptr [[X]], null, !nosanitize [[META6]] +// REC-NEXT: [[TMP1:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 29), !nosanitize [[META6]] // REC-NEXT: [[DOTNOT1:%.*]] = and i1 [[TMP0]], [[TMP1]] -// REC-NEXT: br i1 [[DOTNOT1]], label %[[HANDLER_TYPE_MISMATCH:.*]], label %[[CONT:.*]], !prof [[PROF4:![0-9]+]], !nosanitize [[META2]] +// REC-NEXT: br i1 [[DOTNOT1]], label %[[HANDLER_TYPE_MISMATCH:.*]], label %[[CONT:.*]], !prof [[PROF8:![0-9]+]], !nosanitize [[META6]] // REC: [[HANDLER_TYPE_MISMATCH]]: -// REC-NEXT: tail call void @__ubsan_handle_type_mismatch_v1(ptr nonnull @[[GLOB2:[0-9]+]], i64 0) #[[ATTR6]], !nosanitize [[META2]] -// REC-NEXT: br label %[[CONT]], !nosanitize [[META2]] +// REC-NEXT: tail call void @__ubsan_handle_type_mismatch_v1(ptr nonnull @[[GLOB2:[0-9]+]], i64 0) #[[ATTR6]], !nosanitize [[META6]] +// REC-NEXT: br label %[[CONT]], !nosanitize [[META6]] // REC: [[CONT]]: -// REC-NEXT: [[TMP2:%.*]] = load i32, ptr [[X]], align 4, !tbaa [[INT_TBAA5:![0-9]+]] +// REC-NEXT: [[TMP2:%.*]] = load i32, ptr [[X]], align 4, !tbaa [[INT_TBAA2:![0-9]+]] // REC-NEXT: ret i32 [[TMP2]] // int null(int* x) { @@ -136,52 +136,52 @@ int null(int* x) { // CHECK-LABEL: define dso_local noundef i32 @overflow( // CHECK-SAME: i32 noundef [[X:%.*]], i32 noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = tail call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 [[X]], i32 [[Y]]), !nosanitize [[META2]] -// CHECK-NEXT: [[TMP1:%.*]] = extractvalue { i32, i1 } [[TMP0]], 1, !nosanitize [[META2]] +// CHECK-NEXT: [[TMP0:%.*]] = tail call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 [[X]], i32 [[Y]]), !nosanitize [[META6]] +// CHECK-NEXT: [[TMP1:%.*]] = extractvalue { i32, i1 } [[TMP0]], 1, !nosanitize [[META6]] // // 41 == SO_SignedIntegerOverflow -// CHECK-NEXT: [[TMP2:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 41), !nosanitize [[META2]] +// CHECK-NEXT: [[TMP2:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 41), !nosanitize [[META6]] // CHECK-NEXT: [[DOTDEMORGAN:%.*]] = and i1 [[TMP1]], [[TMP2]] -// CHECK-NEXT: br i1 [[DOTDEMORGAN]], label %[[HANDLER_ADD_OVERFLOW:.*]], label %[[CONT:.*]], !prof [[PROF4]], !nosanitize [[META2]] +// CHECK-NEXT: br i1 [[DOTDEMORGAN]], label %[[HANDLER_ADD_OVERFLOW:.*]], label %[[CONT:.*]], !prof [[PROF8]], !nosanitize [[META6]] // CHECK: [[HANDLER_ADD_OVERFLOW]]: -// CHECK-NEXT: [[TMP3:%.*]] = zext i32 [[X]] to i64, !nosanitize [[META2]] -// CHECK-NEXT: [[TMP4:%.*]] = zext i32 [[Y]] to i64, !nosanitize [[META2]] -// CHECK-NEXT: tail call void @__ubsan_handle_add_overflow_abort(ptr nonnull @[[GLOB3:[0-9]+]], i64 [[TMP3]], i64 [[TMP4]]) #[[ATTR6]], !nosanitize [[META2]] -// CHECK-NEXT: unreachable, !nosanitize [[META2]] +// CHECK-NEXT: [[TMP3:%.*]] = zext i32 [[X]] to i64, !nosanitize [[META6]] +// CHECK-NEXT: [[TMP4:%.*]] = zext i32 [[Y]] to i64, !nosanitize [[META6]] +// CHECK-NEXT: tail call void @__ubsan_handle_add_overflow_abort(ptr nonnull @[[GLOB3:[0-9]+]], i64 [[TMP3]], i64 [[TMP4]]) #[[ATTR6]], !nosanitize [[META6]] +// CHECK-NEXT: unreachable, !nosanitize [[META6]] // CHECK: [[CONT]]: -// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP0]], 0, !nosanitize [[META2]] +// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP0]], 0, !nosanitize [[META6]] // CHECK-NEXT: ret i32 [[TMP5]] // // TR-LABEL: define dso_local noundef i32 @overflow( // TR-SAME: i32 noundef [[X:%.*]], i32 noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR0]] { // TR-NEXT: [[ENTRY:.*:]] -// TR-NEXT: [[TMP0:%.*]] = tail call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 [[X]], i32 [[Y]]), !nosanitize [[META2]] -// TR-NEXT: [[TMP1:%.*]] = extractvalue { i32, i1 } [[TMP0]], 1, !nosanitize [[META2]] -// TR-NEXT: [[TMP2:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 41), !nosanitize [[META2]] +// TR-NEXT: [[TMP0:%.*]] = tail call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 [[X]], i32 [[Y]]), !nosanitize [[META6]] +// TR-NEXT: [[TMP1:%.*]] = extractvalue { i32, i1 } [[TMP0]], 1, !nosanitize [[META6]] +// TR-NEXT: [[TMP2:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 41), !nosanitize [[META6]] // TR-NEXT: [[DOTDEMORGAN:%.*]] = and i1 [[TMP1]], [[TMP2]] -// TR-NEXT: br i1 [[DOTDEMORGAN]], label %[[TRAP:.*]], label %[[CONT:.*]], !prof [[PROF4]], !nosanitize [[META2]] +// TR-NEXT: br i1 [[DOTDEMORGAN]], label %[[TRAP:.*]], label %[[CONT:.*]], !prof [[PROF8]], !nosanitize [[META6]] // TR: [[TRAP]]: -// TR-NEXT: tail call void @llvm.ubsantrap(i8 0) #[[ATTR7]], !nosanitize [[META2]] -// TR-NEXT: unreachable, !nosanitize [[META2]] +// TR-NEXT: tail call void @llvm.ubsantrap(i8 0) #[[ATTR7]], !nosanitize [[META6]] +// TR-NEXT: unreachable, !nosanitize [[META6]] // TR: [[CONT]]: -// TR-NEXT: [[TMP3:%.*]] = extractvalue { i32, i1 } [[TMP0]], 0, !nosanitize [[META2]] +// TR-NEXT: [[TMP3:%.*]] = extractvalue { i32, i1 } [[TMP0]], 0, !nosanitize [[META6]] // TR-NEXT: ret i32 [[TMP3]] // // REC-LABEL: define dso_local noundef i32 @overflow( // REC-SAME: i32 noundef [[X:%.*]], i32 noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR0]] { // REC-NEXT: [[ENTRY:.*:]] -// REC-NEXT: [[TMP0:%.*]] = tail call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 [[X]], i32 [[Y]]), !nosanitize [[META2]] -// REC-NEXT: [[TMP1:%.*]] = extractvalue { i32, i1 } [[TMP0]], 1, !nosanitize [[META2]] -// REC-NEXT: [[TMP2:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 41), !nosanitize [[META2]] +// REC-NEXT: [[TMP0:%.*]] = tail call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 [[X]], i32 [[Y]]), !nosanitize [[META6]] +// REC-NEXT: [[TMP1:%.*]] = extractvalue { i32, i1 } [[TMP0]], 1, !nosanitize [[META6]] +// REC-NEXT: [[TMP2:%.*]] = tail call i1 @llvm.allow.ubsan.check(i8 41), !nosanitize [[META6]] // REC-NEXT: [[DOTDEMORGAN:%.*]] = and i1 [[TMP1]], [[TMP2]] -// REC-NEXT: br i1 [[DOTDEMORGAN]], label %[[HANDLER_ADD_OVERFLOW:.*]], label %[[CONT:.*]], !prof [[PROF4]], !nosanitize [[META2]] +// REC-NEXT: br i1 [[DOTDEMORGAN]], label %[[HANDLER_ADD_OVERFLOW:.*]], label %[[CONT:.*]], !prof [[PROF8]], !nosanitize [[META6]] // REC: [[HANDLER_ADD_OVERFLOW]]: -// REC-NEXT: [[TMP3:%.*]] = zext i32 [[X]] to i64, !nosanitize [[META2]] -// REC-NEXT: [[TMP4:%.*]] = zext i32 [[Y]] to i64, !nosanitize [[META2]] -// REC-NEXT: tail call void @__ubsan_handle_add_overflow(ptr nonnull @[[GLOB3:[0-9]+]], i64 [[TMP3]], i64 [[TMP4]]) #[[ATTR6]], !nosanitize [[META2]] -// REC-NEXT: br label %[[CONT]], !nosanitize [[META2]] +// REC-NEXT: [[TMP3:%.*]] = zext i32 [[X]] to i64, !nosanitize [[META6]] +// REC-NEXT: [[TMP4:%.*]] = zext i32 [[Y]] to i64, !nosanitize [[META6]] +// REC-NEXT: tail call void @__ubsan_handle_add_overflow(ptr nonnull @[[GLOB3:[0-9]+]], i64 [[TMP3]], i64 [[TMP4]]) #[[ATTR6]], !nosanitize [[META6]] +// REC-NEXT: br label %[[CONT]], !nosanitize [[META6]] // REC: [[CONT]]: -// REC-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP0]], 0, !nosanitize [[META2]] +// REC-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP0]], 0, !nosanitize [[META6]] // REC-NEXT: ret i32 [[TMP5]] // int overflow(int x, int y) { @@ -200,16 +200,16 @@ void use(double*); // CHECK-NEXT: [[TMP1:%.*]] = icmp ule i64 [[TMP0]], [[IDXPROM]] // // 71 == SO_LocalBounds -// CHECK-NEXT: [[TMP2:%.*]] = call i1 @llvm.allow.ubsan.check(i8 71), !nosanitize [[META2]] -// CHECK-NEXT: [[TMP3:%.*]] = and i1 [[TMP1]], [[TMP2]], !nosanitize [[META2]] +// CHECK-NEXT: [[TMP2:%.*]] = call i1 @llvm.allow.ubsan.check(i8 71), !nosanitize [[META6]] +// CHECK-NEXT: [[TMP3:%.*]] = and i1 [[TMP1]], [[TMP2]], !nosanitize [[META6]] // CHECK-NEXT: br i1 [[TMP3]], label %[[TRAP:.*]], label %[[BB4:.*]] // CHECK: [[BB4]]: // CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, ptr [[VLA]], i64 [[IDXPROM]] // CHECK-NEXT: [[TMP5:%.*]] = load double, ptr [[ARRAYIDX]], align 8, !tbaa [[DOUBLE_TBAA9:![0-9]+]] // CHECK-NEXT: ret double [[TMP5]] // CHECK: [[TRAP]]: -// CHECK-NEXT: call void @__ubsan_handle_local_out_of_bounds_abort() #[[ATTR6]], !nosanitize [[META2]] -// CHECK-NEXT: unreachable, !nosanitize [[META2]] +// CHECK-NEXT: call void @__ubsan_handle_local_out_of_bounds_abort() #[[ATTR6]], !nosanitize [[META6]] +// CHECK-NEXT: unreachable, !nosanitize [[META6]] // // TR-LABEL: define dso_local double @lbounds( // TR-SAME: i32 noundef [[B:%.*]], i32 noundef [[I:%.*]]) local_unnamed_addr #[[ATTR5:[0-9]+]] { @@ -219,16 +219,16 @@ void use(double*); // TR-NEXT: call void @use(ptr noundef nonnull [[VLA]]) #[[ATTR8:[0-9]+]] // TR-NEXT: [[IDXPROM:%.*]] = sext i32 [[I]] to i64 // TR-NEXT: [[TMP1:%.*]] = icmp ule i64 [[TMP0]], [[IDXPROM]] -// TR-NEXT: [[TMP2:%.*]] = call i1 @llvm.allow.ubsan.check(i8 71), !nosanitize [[META2]] -// TR-NEXT: [[TMP3:%.*]] = and i1 [[TMP1]], [[TMP2]], !nosanitize [[META2]] +// TR-NEXT: [[TMP2:%.*]] = call i1 @llvm.allow.ubsan.check(i8 71), !nosanitize [[META6]] +// TR-NEXT: [[TMP3:%.*]] = and i1 [[TMP1]], [[TMP2]], !nosanitize [[META6]] // TR-NEXT: br i1 [[TMP3]], label %[[TRAP:.*]], label %[[BB4:.*]] // TR: [[BB4]]: // TR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, ptr [[VLA]], i64 [[IDXPROM]] // TR-NEXT: [[TMP5:%.*]] = load double, ptr [[ARRAYIDX]], align 8, !tbaa [[DOUBLE_TBAA9:![0-9]+]] // TR-NEXT: ret double [[TMP5]] // TR: [[TRAP]]: -// TR-NEXT: call void @llvm.ubsantrap(i8 71) #[[ATTR7]], !nosanitize [[META2]] -// TR-NEXT: unreachable, !nosanitize [[META2]] +// TR-NEXT: call void @llvm.ubsantrap(i8 71) #[[ATTR7]], !nosanitize [[META6]] +// TR-NEXT: unreachable, !nosanitize [[META6]] // // REC-LABEL: define dso_local double @lbounds( // REC-SAME: i32 noundef [[B:%.*]], i32 noundef [[I:%.*]]) local_unnamed_addr #[[ATTR0]] { @@ -238,16 +238,16 @@ void use(double*); // REC-NEXT: call void @use(ptr noundef nonnull [[VLA]]) #[[ATTR5:[0-9]+]] // REC-NEXT: [[IDXPROM:%.*]] = sext i32 [[I]] to i64 // REC-NEXT: [[TMP1:%.*]] = icmp ule i64 [[TMP0]], [[IDXPROM]] -// REC-NEXT: [[TMP2:%.*]] = call i1 @llvm.allow.ubsan.check(i8 71), !nosanitize [[META2]] -// REC-NEXT: [[TMP3:%.*]] = and i1 [[TMP1]], [[TMP2]], !nosanitize [[META2]] +// REC-NEXT: [[TMP2:%.*]] = call i1 @llvm.allow.ubsan.check(i8 71), !nosanitize [[META6]] +// REC-NEXT: [[TMP3:%.*]] = and i1 [[TMP1]], [[TMP2]], !nosanitize [[META6]] // REC-NEXT: br i1 [[TMP3]], label %[[TRAP:.*]], label %[[BB4:.*]] // REC: [[BB4]]: // REC-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, ptr [[VLA]], i64 [[IDXPROM]] // REC-NEXT: [[TMP5:%.*]] = load double, ptr [[ARRAYIDX]], align 8, !tbaa [[DOUBLE_TBAA9:![0-9]+]] // REC-NEXT: ret double [[TMP5]] // REC: [[TRAP]]: -// REC-NEXT: call void @__ubsan_handle_local_out_of_bounds() #[[ATTR6]], !nosanitize [[META2]] -// REC-NEXT: br label %[[BB4]], !nosanitize [[META2]] +// REC-NEXT: call void @__ubsan_handle_local_out_of_bounds() #[[ATTR6]], !nosanitize [[META6]] +// REC-NEXT: br label %[[BB4]], !nosanitize [[META6]] // double lbounds(int b, int i) { double a[b]; @@ -256,33 +256,33 @@ double lbounds(int b, int i) { } //. -// CHECK: [[META2]] = !{} -// CHECK: [[PROF3]] = !{!"branch_weights", i32 1048575, i32 1} -// CHECK: [[PROF4]] = !{!"branch_weights", i32 1, i32 1048575} -// CHECK: [[INT_TBAA5]] = !{[[META6:![0-9]+]], [[META6]], i64 0} -// CHECK: [[META6]] = !{!"int", [[META7:![0-9]+]], i64 0} -// CHECK: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0} -// CHECK: [[META8]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[INT_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// CHECK: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META6]] = !{} +// CHECK: [[PROF7]] = !{!"branch_weights", i32 1048575, i32 1} +// CHECK: [[PROF8]] = !{!"branch_weights", i32 1, i32 1048575} // CHECK: [[DOUBLE_TBAA9]] = !{[[META10:![0-9]+]], [[META10]], i64 0} -// CHECK: [[META10]] = !{!"double", [[META7]], i64 0} +// CHECK: [[META10]] = !{!"double", [[META4]], i64 0} //. -// TR: [[META2]] = !{} -// TR: [[PROF3]] = !{!"branch_weights", i32 1048575, i32 1} -// TR: [[PROF4]] = !{!"branch_weights", i32 1, i32 1048575} -// TR: [[INT_TBAA5]] = !{[[META6:![0-9]+]], [[META6]], i64 0} -// TR: [[META6]] = !{!"int", [[META7:![0-9]+]], i64 0} -// TR: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0} -// TR: [[META8]] = !{!"Simple C/C++ TBAA"} +// TR: [[INT_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// TR: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// TR: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// TR: [[META5]] = !{!"Simple C/C++ TBAA"} +// TR: [[META6]] = !{} +// TR: [[PROF7]] = !{!"branch_weights", i32 1048575, i32 1} +// TR: [[PROF8]] = !{!"branch_weights", i32 1, i32 1048575} // TR: [[DOUBLE_TBAA9]] = !{[[META10:![0-9]+]], [[META10]], i64 0} -// TR: [[META10]] = !{!"double", [[META7]], i64 0} +// TR: [[META10]] = !{!"double", [[META4]], i64 0} //. -// REC: [[META2]] = !{} -// REC: [[PROF3]] = !{!"branch_weights", i32 1048575, i32 1} -// REC: [[PROF4]] = !{!"branch_weights", i32 1, i32 1048575} -// REC: [[INT_TBAA5]] = !{[[META6:![0-9]+]], [[META6]], i64 0} -// REC: [[META6]] = !{!"int", [[META7:![0-9]+]], i64 0} -// REC: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0} -// REC: [[META8]] = !{!"Simple C/C++ TBAA"} +// REC: [[INT_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// REC: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// REC: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// REC: [[META5]] = !{!"Simple C/C++ TBAA"} +// REC: [[META6]] = !{} +// REC: [[PROF7]] = !{!"branch_weights", i32 1048575, i32 1} +// REC: [[PROF8]] = !{!"branch_weights", i32 1, i32 1048575} // REC: [[DOUBLE_TBAA9]] = !{[[META10:![0-9]+]], [[META10]], i64 0} -// REC: [[META10]] = !{!"double", [[META7]], i64 0} +// REC: [[META10]] = !{!"double", [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/arm-branch-protection-attr-2.c b/clang/test/CodeGen/arm-branch-protection-attr-2.c index fad5dc0..5391537 100644 --- a/clang/test/CodeGen/arm-branch-protection-attr-2.c +++ b/clang/test/CodeGen/arm-branch-protection-attr-2.c @@ -23,16 +23,16 @@ // NONE-NOT: !"branch-target-enforcement" // PART-NOT: !"branch-target-enforcement" // ALL-NOT: !"branch-target-enforcement" -// BTE: !{i32 8, !"branch-target-enforcement", i32 1} +// BTE: !{i32 8, !"branch-target-enforcement", i32 2} // NONE-NOT: !"sign-return-address" -// PART: !{i32 8, !"sign-return-address", i32 1} -// ALL: !{i32 8, !"sign-return-address", i32 1} +// PART: !{i32 8, !"sign-return-address", i32 2} +// ALL: !{i32 8, !"sign-return-address", i32 2} // BTE-NOT: !"sign-return-address" // NONE-NOT: !"sign-return-address-all", i32 0} // PART-NOT: !"sign-return-address-all", i32 0} -// ALL: !{i32 8, !"sign-return-address-all", i32 1} +// ALL: !{i32 8, !"sign-return-address-all", i32 2} // BTE-NOT: !"sign-return-address-all", i32 0} void foo() {} diff --git a/clang/test/CodeGen/attr-arm-sve-vector-bits-bitcast.c b/clang/test/CodeGen/attr-arm-sve-vector-bits-bitcast.c index 847ce67f..29425a03 100644 --- a/clang/test/CodeGen/attr-arm-sve-vector-bits-bitcast.c +++ b/clang/test/CodeGen/attr-arm-sve-vector-bits-bitcast.c @@ -32,7 +32,7 @@ DEFINE_STRUCT(bool) // CHECK-128-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16 -// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2:![0-9]+]] +// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6:![0-9]+]] // CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.insert.nxv2i64.v2i64(<vscale x 2 x i64> poison, <2 x i64> [[TMP0]], i64 0) // CHECK-128-NEXT: ret <vscale x 2 x i64> [[CAST_SCALABLE]] // @@ -40,7 +40,7 @@ DEFINE_STRUCT(bool) // CHECK-256-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32 -// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x i64>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2:![0-9]+]] +// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x i64>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6:![0-9]+]] // CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.insert.nxv2i64.v4i64(<vscale x 2 x i64> poison, <4 x i64> [[TMP0]], i64 0) // CHECK-256-NEXT: ret <vscale x 2 x i64> [[CAST_SCALABLE]] // @@ -48,7 +48,7 @@ DEFINE_STRUCT(bool) // CHECK-512-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-512-NEXT: [[ENTRY:.*:]] // CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 64 -// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x i64>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2:![0-9]+]] +// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x i64>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6:![0-9]+]] // CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.insert.nxv2i64.v8i64(<vscale x 2 x i64> poison, <8 x i64> [[TMP0]], i64 0) // CHECK-512-NEXT: ret <vscale x 2 x i64> [[CAST_SCALABLE]] // @@ -61,7 +61,7 @@ svint64_t read_int64(struct struct_int64 *s) { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x i64> @llvm.vector.extract.v2i64.nxv2i64(<vscale x 2 x i64> [[X]], i64 0) // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16 -// CHECK-128-NEXT: store <2 x i64> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-128-NEXT: store <2 x i64> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-128-NEXT: ret void // // CHECK-256-LABEL: define dso_local void @write_int64( @@ -69,7 +69,7 @@ svint64_t read_int64(struct struct_int64 *s) { // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <4 x i64> @llvm.vector.extract.v4i64.nxv2i64(<vscale x 2 x i64> [[X]], i64 0) // CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32 -// CHECK-256-NEXT: store <4 x i64> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-256-NEXT: store <4 x i64> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-256-NEXT: ret void // // CHECK-512-LABEL: define dso_local void @write_int64( @@ -77,7 +77,7 @@ svint64_t read_int64(struct struct_int64 *s) { // CHECK-512-NEXT: [[ENTRY:.*:]] // CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x i64> @llvm.vector.extract.v8i64.nxv2i64(<vscale x 2 x i64> [[X]], i64 0) // CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 64 -// CHECK-512-NEXT: store <8 x i64> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-512-NEXT: store <8 x i64> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-512-NEXT: ret void // void write_int64(struct struct_int64 *s, svint64_t x) { @@ -92,7 +92,7 @@ void write_int64(struct struct_int64 *s, svint64_t x) { // CHECK-128-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16 -// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x double> @llvm.vector.insert.nxv2f64.v2f64(<vscale x 2 x double> poison, <2 x double> [[TMP0]], i64 0) // CHECK-128-NEXT: ret <vscale x 2 x double> [[CAST_SCALABLE]] // @@ -100,7 +100,7 @@ void write_int64(struct struct_int64 *s, svint64_t x) { // CHECK-256-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32 -// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x double>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x double>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x double> @llvm.vector.insert.nxv2f64.v4f64(<vscale x 2 x double> poison, <4 x double> [[TMP0]], i64 0) // CHECK-256-NEXT: ret <vscale x 2 x double> [[CAST_SCALABLE]] // @@ -108,7 +108,7 @@ void write_int64(struct struct_int64 *s, svint64_t x) { // CHECK-512-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-512-NEXT: [[ENTRY:.*:]] // CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 64 -// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x double>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x double>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x double> @llvm.vector.insert.nxv2f64.v8f64(<vscale x 2 x double> poison, <8 x double> [[TMP0]], i64 0) // CHECK-512-NEXT: ret <vscale x 2 x double> [[CAST_SCALABLE]] // @@ -121,7 +121,7 @@ svfloat64_t read_float64(struct struct_float64 *s) { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x double> @llvm.vector.extract.v2f64.nxv2f64(<vscale x 2 x double> [[X]], i64 0) // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16 -// CHECK-128-NEXT: store <2 x double> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-128-NEXT: store <2 x double> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-128-NEXT: ret void // // CHECK-256-LABEL: define dso_local void @write_float64( @@ -129,7 +129,7 @@ svfloat64_t read_float64(struct struct_float64 *s) { // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <4 x double> @llvm.vector.extract.v4f64.nxv2f64(<vscale x 2 x double> [[X]], i64 0) // CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32 -// CHECK-256-NEXT: store <4 x double> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-256-NEXT: store <4 x double> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-256-NEXT: ret void // // CHECK-512-LABEL: define dso_local void @write_float64( @@ -137,7 +137,7 @@ svfloat64_t read_float64(struct struct_float64 *s) { // CHECK-512-NEXT: [[ENTRY:.*:]] // CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x double> @llvm.vector.extract.v8f64.nxv2f64(<vscale x 2 x double> [[X]], i64 0) // CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 64 -// CHECK-512-NEXT: store <8 x double> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-512-NEXT: store <8 x double> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-512-NEXT: ret void // void write_float64(struct struct_float64 *s, svfloat64_t x) { @@ -152,7 +152,7 @@ void write_float64(struct struct_float64 *s, svfloat64_t x) { // CHECK-128-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16 -// CHECK-128-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-128-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.insert.nxv8bf16.v8bf16(<vscale x 8 x bfloat> poison, <8 x bfloat> [[TMP0]], i64 0) // CHECK-128-NEXT: ret <vscale x 8 x bfloat> [[CAST_SCALABLE]] // @@ -160,7 +160,7 @@ void write_float64(struct struct_float64 *s, svfloat64_t x) { // CHECK-256-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32 -// CHECK-256-NEXT: [[TMP0:%.*]] = load <16 x bfloat>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-256-NEXT: [[TMP0:%.*]] = load <16 x bfloat>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.insert.nxv8bf16.v16bf16(<vscale x 8 x bfloat> poison, <16 x bfloat> [[TMP0]], i64 0) // CHECK-256-NEXT: ret <vscale x 8 x bfloat> [[CAST_SCALABLE]] // @@ -168,7 +168,7 @@ void write_float64(struct struct_float64 *s, svfloat64_t x) { // CHECK-512-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-512-NEXT: [[ENTRY:.*:]] // CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 64 -// CHECK-512-NEXT: [[TMP0:%.*]] = load <32 x bfloat>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-512-NEXT: [[TMP0:%.*]] = load <32 x bfloat>, ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.insert.nxv8bf16.v32bf16(<vscale x 8 x bfloat> poison, <32 x bfloat> [[TMP0]], i64 0) // CHECK-512-NEXT: ret <vscale x 8 x bfloat> [[CAST_SCALABLE]] // @@ -181,7 +181,7 @@ svbfloat16_t read_bfloat16(struct struct_bfloat16 *s) { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x bfloat> @llvm.vector.extract.v8bf16.nxv8bf16(<vscale x 8 x bfloat> [[X]], i64 0) // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16 -// CHECK-128-NEXT: store <8 x bfloat> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-128-NEXT: store <8 x bfloat> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-128-NEXT: ret void // // CHECK-256-LABEL: define dso_local void @write_bfloat16( @@ -189,7 +189,7 @@ svbfloat16_t read_bfloat16(struct struct_bfloat16 *s) { // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <16 x bfloat> @llvm.vector.extract.v16bf16.nxv8bf16(<vscale x 8 x bfloat> [[X]], i64 0) // CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32 -// CHECK-256-NEXT: store <16 x bfloat> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-256-NEXT: store <16 x bfloat> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-256-NEXT: ret void // // CHECK-512-LABEL: define dso_local void @write_bfloat16( @@ -197,7 +197,7 @@ svbfloat16_t read_bfloat16(struct struct_bfloat16 *s) { // CHECK-512-NEXT: [[ENTRY:.*:]] // CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <32 x bfloat> @llvm.vector.extract.v32bf16.nxv8bf16(<vscale x 8 x bfloat> [[X]], i64 0) // CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 64 -// CHECK-512-NEXT: store <32 x bfloat> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-512-NEXT: store <32 x bfloat> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-512-NEXT: ret void // void write_bfloat16(struct struct_bfloat16 *s, svbfloat16_t x) { @@ -212,7 +212,7 @@ void write_bfloat16(struct struct_bfloat16 *s, svbfloat16_t x) { // CHECK-128-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 2 -// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i8>, ptr [[Y]], align 2, !tbaa [[CHAR_TBAA2]] +// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i8>, ptr [[Y]], align 2, !tbaa [[CHAR_TBAA6]] // CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i8> @llvm.vector.insert.nxv2i8.v2i8(<vscale x 2 x i8> poison, <2 x i8> [[TMP0]], i64 0) // CHECK-128-NEXT: [[TMP1:%.*]] = bitcast <vscale x 2 x i8> [[CAST_SCALABLE]] to <vscale x 16 x i1> // CHECK-128-NEXT: ret <vscale x 16 x i1> [[TMP1]] @@ -221,7 +221,7 @@ void write_bfloat16(struct struct_bfloat16 *s, svbfloat16_t x) { // CHECK-256-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-256-NEXT: [[ENTRY:.*:]] // CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 4 -// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x i8>, ptr [[Y]], align 2, !tbaa [[CHAR_TBAA2]] +// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x i8>, ptr [[Y]], align 2, !tbaa [[CHAR_TBAA6]] // CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i8> @llvm.vector.insert.nxv2i8.v4i8(<vscale x 2 x i8> poison, <4 x i8> [[TMP0]], i64 0) // CHECK-256-NEXT: [[TMP1:%.*]] = bitcast <vscale x 2 x i8> [[CAST_SCALABLE]] to <vscale x 16 x i1> // CHECK-256-NEXT: ret <vscale x 16 x i1> [[TMP1]] @@ -230,7 +230,7 @@ void write_bfloat16(struct struct_bfloat16 *s, svbfloat16_t x) { // CHECK-512-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-512-NEXT: [[ENTRY:.*:]] // CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 8 -// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr [[Y]], align 2, !tbaa [[CHAR_TBAA2]] +// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr [[Y]], align 2, !tbaa [[CHAR_TBAA6]] // CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i8> @llvm.vector.insert.nxv2i8.v8i8(<vscale x 2 x i8> poison, <8 x i8> [[TMP0]], i64 0) // CHECK-512-NEXT: [[TMP1:%.*]] = bitcast <vscale x 2 x i8> [[CAST_SCALABLE]] to <vscale x 16 x i1> // CHECK-512-NEXT: ret <vscale x 16 x i1> [[TMP1]] @@ -245,7 +245,7 @@ svbool_t read_bool(struct struct_bool *s) { // CHECK-128-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1> [[X]] to <vscale x 2 x i8> // CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x i8> @llvm.vector.extract.v2i8.nxv2i8(<vscale x 2 x i8> [[TMP0]], i64 0) // CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 2 -// CHECK-128-NEXT: store <2 x i8> [[CAST_FIXED]], ptr [[Y]], align 2, !tbaa [[CHAR_TBAA2]] +// CHECK-128-NEXT: store <2 x i8> [[CAST_FIXED]], ptr [[Y]], align 2, !tbaa [[CHAR_TBAA6]] // CHECK-128-NEXT: ret void // // CHECK-256-LABEL: define dso_local void @write_bool( @@ -254,7 +254,7 @@ svbool_t read_bool(struct struct_bool *s) { // CHECK-256-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1> [[X]] to <vscale x 2 x i8> // CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <4 x i8> @llvm.vector.extract.v4i8.nxv2i8(<vscale x 2 x i8> [[TMP0]], i64 0) // CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 4 -// CHECK-256-NEXT: store <4 x i8> [[CAST_FIXED]], ptr [[Y]], align 2, !tbaa [[CHAR_TBAA2]] +// CHECK-256-NEXT: store <4 x i8> [[CAST_FIXED]], ptr [[Y]], align 2, !tbaa [[CHAR_TBAA6]] // CHECK-256-NEXT: ret void // // CHECK-512-LABEL: define dso_local void @write_bool( @@ -263,22 +263,22 @@ svbool_t read_bool(struct struct_bool *s) { // CHECK-512-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1> [[X]] to <vscale x 2 x i8> // CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x i8> @llvm.vector.extract.v8i8.nxv2i8(<vscale x 2 x i8> [[TMP0]], i64 0) // CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 8 -// CHECK-512-NEXT: store <8 x i8> [[CAST_FIXED]], ptr [[Y]], align 2, !tbaa [[CHAR_TBAA2]] +// CHECK-512-NEXT: store <8 x i8> [[CAST_FIXED]], ptr [[Y]], align 2, !tbaa [[CHAR_TBAA6]] // CHECK-512-NEXT: ret void // void write_bool(struct struct_bool *s, svbool_t x) { s->y[0] = x; } //. -// CHECK-128: [[CHAR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK-128: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK-128: [[META4]] = !{!"Simple C/C++ TBAA"} +// CHECK-128: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK-128: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK-128: [[CHAR_TBAA6]] = !{[[META4]], [[META4]], i64 0} //. -// CHECK-256: [[CHAR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK-256: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK-256: [[META4]] = !{!"Simple C/C++ TBAA"} +// CHECK-256: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK-256: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK-256: [[CHAR_TBAA6]] = !{[[META4]], [[META4]], i64 0} //. -// CHECK-512: [[CHAR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK-512: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK-512: [[META4]] = !{!"Simple C/C++ TBAA"} +// CHECK-512: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK-512: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK-512: [[CHAR_TBAA6]] = !{[[META4]], [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/attr-arm-sve-vector-bits-cast.c b/clang/test/CodeGen/attr-arm-sve-vector-bits-cast.c index bdaebf7..7e3223f 100644 --- a/clang/test/CodeGen/attr-arm-sve-vector-bits-cast.c +++ b/clang/test/CodeGen/attr-arm-sve-vector-bits-cast.c @@ -79,7 +79,7 @@ svint64_t lax_cast(fixed_int32_t type) { // CHECK-LABEL: define dso_local <vscale x 4 x i32> @to_svint32_t__from_gnu_int32_t( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR2:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TYPE:%.*]] = load <16 x i32>, ptr [[TMP0]], align 16, !tbaa [[CHAR_TBAA2:![0-9]+]] +// CHECK-NEXT: [[TYPE:%.*]] = load <16 x i32>, ptr [[TMP0]], align 16, !tbaa [[CHAR_TBAA6:![0-9]+]] // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.insert.nxv4i32.v16i32(<vscale x 4 x i32> poison, <16 x i32> [[TYPE]], i64 0) // CHECK-NEXT: ret <vscale x 4 x i32> [[CAST_SCALABLE]] // @@ -91,7 +91,7 @@ svint32_t to_svint32_t__from_gnu_int32_t(gnu_int32_t type) { // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i32>) align 16 captures(none) initializes((0, 64)) [[AGG_RESULT:%.*]], <vscale x 4 x i32> [[TYPE:%.*]]) local_unnamed_addr #[[ATTR3:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[CAST_FIXED:%.*]] = tail call <16 x i32> @llvm.vector.extract.v16i32.nxv4i32(<vscale x 4 x i32> [[TYPE]], i64 0) -// CHECK-NEXT: store <16 x i32> [[CAST_FIXED]], ptr [[AGG_RESULT]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i32> [[CAST_FIXED]], ptr [[AGG_RESULT]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // gnu_int32_t from_svint32_t__to_gnu_int32_t(svint32_t type) { @@ -101,7 +101,7 @@ gnu_int32_t from_svint32_t__to_gnu_int32_t(svint32_t type) { // CHECK-LABEL: define dso_local <vscale x 4 x i32> @to_fixed_int32_t__from_gnu_int32_t( // CHECK-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TYPE:%.*]] = load <16 x i32>, ptr [[TMP0]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TYPE:%.*]] = load <16 x i32>, ptr [[TMP0]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.insert.nxv4i32.v16i32(<vscale x 4 x i32> poison, <16 x i32> [[TYPE]], i64 0) // CHECK-NEXT: ret <vscale x 4 x i32> [[CAST_SCALABLE]] // @@ -113,14 +113,14 @@ fixed_int32_t to_fixed_int32_t__from_gnu_int32_t(gnu_int32_t type) { // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<16 x i32>) align 16 captures(none) initializes((0, 64)) [[AGG_RESULT:%.*]], <vscale x 4 x i32> noundef [[TYPE_COERCE:%.*]]) local_unnamed_addr #[[ATTR3]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TYPE:%.*]] = tail call <16 x i32> @llvm.vector.extract.v16i32.nxv4i32(<vscale x 4 x i32> [[TYPE_COERCE]], i64 0) -// CHECK-NEXT: store <16 x i32> [[TYPE]], ptr [[AGG_RESULT]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <16 x i32> [[TYPE]], ptr [[AGG_RESULT]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // gnu_int32_t from_fixed_int32_t__to_gnu_int32_t(fixed_int32_t type) { return type; } //. -// CHECK: [[CHAR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[CHAR_TBAA6]] = !{[[META4]], [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/attr-arm-sve-vector-bits-globals.c b/clang/test/CodeGen/attr-arm-sve-vector-bits-globals.c index b604a06..ae011db 100644 --- a/clang/test/CodeGen/attr-arm-sve-vector-bits-globals.c +++ b/clang/test/CodeGen/attr-arm-sve-vector-bits-globals.c @@ -24,14 +24,14 @@ fixed_bool_t global_bool; // CHECK-128-SAME: <vscale x 2 x i64> [[V:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x i64> @llvm.vector.extract.v2i64.nxv2i64(<vscale x 2 x i64> [[V]], i64 0) -// CHECK-128-NEXT: store <2 x i64> [[CAST_FIXED]], ptr @global_i64, align 16, !tbaa [[CHAR_TBAA2:![0-9]+]] +// CHECK-128-NEXT: store <2 x i64> [[CAST_FIXED]], ptr @global_i64, align 16, !tbaa [[CHAR_TBAA6:![0-9]+]] // CHECK-128-NEXT: ret void // // CHECK-512-LABEL: define void @write_global_i64( // CHECK-512-SAME: <vscale x 2 x i64> [[V:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-512-NEXT: [[ENTRY:.*:]] // CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x i64> @llvm.vector.extract.v8i64.nxv2i64(<vscale x 2 x i64> [[V]], i64 0) -// CHECK-512-NEXT: store <8 x i64> [[CAST_FIXED]], ptr @global_i64, align 16, !tbaa [[CHAR_TBAA2:![0-9]+]] +// CHECK-512-NEXT: store <8 x i64> [[CAST_FIXED]], ptr @global_i64, align 16, !tbaa [[CHAR_TBAA6:![0-9]+]] // CHECK-512-NEXT: ret void // void write_global_i64(svint64_t v) { global_i64 = v; } @@ -40,14 +40,14 @@ void write_global_i64(svint64_t v) { global_i64 = v; } // CHECK-128-SAME: <vscale x 8 x bfloat> [[V:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x bfloat> @llvm.vector.extract.v8bf16.nxv8bf16(<vscale x 8 x bfloat> [[V]], i64 0) -// CHECK-128-NEXT: store <8 x bfloat> [[CAST_FIXED]], ptr @global_bf16, align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-128-NEXT: store <8 x bfloat> [[CAST_FIXED]], ptr @global_bf16, align 16, !tbaa [[CHAR_TBAA6]] // CHECK-128-NEXT: ret void // // CHECK-512-LABEL: define void @write_global_bf16( // CHECK-512-SAME: <vscale x 8 x bfloat> [[V:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-512-NEXT: [[ENTRY:.*:]] // CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <32 x bfloat> @llvm.vector.extract.v32bf16.nxv8bf16(<vscale x 8 x bfloat> [[V]], i64 0) -// CHECK-512-NEXT: store <32 x bfloat> [[CAST_FIXED]], ptr @global_bf16, align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-512-NEXT: store <32 x bfloat> [[CAST_FIXED]], ptr @global_bf16, align 16, !tbaa [[CHAR_TBAA6]] // CHECK-512-NEXT: ret void // void write_global_bf16(svbfloat16_t v) { global_bf16 = v; } @@ -57,7 +57,7 @@ void write_global_bf16(svbfloat16_t v) { global_bf16 = v; } // CHECK-128-NEXT: [[ENTRY:.*:]] // CHECK-128-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1> [[V]] to <vscale x 2 x i8> // CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x i8> @llvm.vector.extract.v2i8.nxv2i8(<vscale x 2 x i8> [[TMP0]], i64 0) -// CHECK-128-NEXT: store <2 x i8> [[CAST_FIXED]], ptr @global_bool, align 2, !tbaa [[CHAR_TBAA2]] +// CHECK-128-NEXT: store <2 x i8> [[CAST_FIXED]], ptr @global_bool, align 2, !tbaa [[CHAR_TBAA6]] // CHECK-128-NEXT: ret void // // CHECK-512-LABEL: define void @write_global_bool( @@ -65,7 +65,7 @@ void write_global_bf16(svbfloat16_t v) { global_bf16 = v; } // CHECK-512-NEXT: [[ENTRY:.*:]] // CHECK-512-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1> [[V]] to <vscale x 2 x i8> // CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x i8> @llvm.vector.extract.v8i8.nxv2i8(<vscale x 2 x i8> [[TMP0]], i64 0) -// CHECK-512-NEXT: store <8 x i8> [[CAST_FIXED]], ptr @global_bool, align 2, !tbaa [[CHAR_TBAA2]] +// CHECK-512-NEXT: store <8 x i8> [[CAST_FIXED]], ptr @global_bool, align 2, !tbaa [[CHAR_TBAA6]] // CHECK-512-NEXT: ret void // void write_global_bool(svbool_t v) { global_bool = v; } @@ -77,14 +77,14 @@ void write_global_bool(svbool_t v) { global_bool = v; } // CHECK-128-LABEL: define <vscale x 2 x i64> @read_global_i64( // CHECK-128-SAME: ) local_unnamed_addr #[[ATTR2:[0-9]+]] { // CHECK-128-NEXT: [[ENTRY:.*:]] -// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr @global_i64, align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr @global_i64, align 16, !tbaa [[CHAR_TBAA6]] // CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.insert.nxv2i64.v2i64(<vscale x 2 x i64> poison, <2 x i64> [[TMP0]], i64 0) // CHECK-128-NEXT: ret <vscale x 2 x i64> [[CAST_SCALABLE]] // // CHECK-512-LABEL: define <vscale x 2 x i64> @read_global_i64( // CHECK-512-SAME: ) local_unnamed_addr #[[ATTR2:[0-9]+]] { // CHECK-512-NEXT: [[ENTRY:.*:]] -// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x i64>, ptr @global_i64, align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x i64>, ptr @global_i64, align 16, !tbaa [[CHAR_TBAA6]] // CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.insert.nxv2i64.v8i64(<vscale x 2 x i64> poison, <8 x i64> [[TMP0]], i64 0) // CHECK-512-NEXT: ret <vscale x 2 x i64> [[CAST_SCALABLE]] // @@ -93,14 +93,14 @@ svint64_t read_global_i64() { return global_i64; } // CHECK-128-LABEL: define <vscale x 8 x bfloat> @read_global_bf16( // CHECK-128-SAME: ) local_unnamed_addr #[[ATTR2]] { // CHECK-128-NEXT: [[ENTRY:.*:]] -// CHECK-128-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr @global_bf16, align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-128-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr @global_bf16, align 16, !tbaa [[CHAR_TBAA6]] // CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.insert.nxv8bf16.v8bf16(<vscale x 8 x bfloat> poison, <8 x bfloat> [[TMP0]], i64 0) // CHECK-128-NEXT: ret <vscale x 8 x bfloat> [[CAST_SCALABLE]] // // CHECK-512-LABEL: define <vscale x 8 x bfloat> @read_global_bf16( // CHECK-512-SAME: ) local_unnamed_addr #[[ATTR2]] { // CHECK-512-NEXT: [[ENTRY:.*:]] -// CHECK-512-NEXT: [[TMP0:%.*]] = load <32 x bfloat>, ptr @global_bf16, align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-512-NEXT: [[TMP0:%.*]] = load <32 x bfloat>, ptr @global_bf16, align 16, !tbaa [[CHAR_TBAA6]] // CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.insert.nxv8bf16.v32bf16(<vscale x 8 x bfloat> poison, <32 x bfloat> [[TMP0]], i64 0) // CHECK-512-NEXT: ret <vscale x 8 x bfloat> [[CAST_SCALABLE]] // @@ -109,7 +109,7 @@ svbfloat16_t read_global_bf16() { return global_bf16; } // CHECK-128-LABEL: define <vscale x 16 x i1> @read_global_bool( // CHECK-128-SAME: ) local_unnamed_addr #[[ATTR2]] { // CHECK-128-NEXT: [[ENTRY:.*:]] -// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i8>, ptr @global_bool, align 2, !tbaa [[CHAR_TBAA2]] +// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i8>, ptr @global_bool, align 2, !tbaa [[CHAR_TBAA6]] // CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i8> @llvm.vector.insert.nxv2i8.v2i8(<vscale x 2 x i8> poison, <2 x i8> [[TMP0]], i64 0) // CHECK-128-NEXT: [[TMP1:%.*]] = bitcast <vscale x 2 x i8> [[CAST_SCALABLE]] to <vscale x 16 x i1> // CHECK-128-NEXT: ret <vscale x 16 x i1> [[TMP1]] @@ -117,18 +117,18 @@ svbfloat16_t read_global_bf16() { return global_bf16; } // CHECK-512-LABEL: define <vscale x 16 x i1> @read_global_bool( // CHECK-512-SAME: ) local_unnamed_addr #[[ATTR2]] { // CHECK-512-NEXT: [[ENTRY:.*:]] -// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr @global_bool, align 2, !tbaa [[CHAR_TBAA2]] +// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr @global_bool, align 2, !tbaa [[CHAR_TBAA6]] // CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i8> @llvm.vector.insert.nxv2i8.v8i8(<vscale x 2 x i8> poison, <8 x i8> [[TMP0]], i64 0) // CHECK-512-NEXT: [[TMP1:%.*]] = bitcast <vscale x 2 x i8> [[CAST_SCALABLE]] to <vscale x 16 x i1> // CHECK-512-NEXT: ret <vscale x 16 x i1> [[TMP1]] // svbool_t read_global_bool() { return global_bool; } //. -// CHECK-128: [[CHAR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK-128: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK-128: [[META4]] = !{!"Simple C/C++ TBAA"} +// CHECK-128: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK-128: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK-128: [[CHAR_TBAA6]] = !{[[META4]], [[META4]], i64 0} //. -// CHECK-512: [[CHAR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK-512: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK-512: [[META4]] = !{!"Simple C/C++ TBAA"} +// CHECK-512: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK-512: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK-512: [[CHAR_TBAA6]] = !{[[META4]], [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/attr-counted-by-for-pointers.c b/clang/test/CodeGen/attr-counted-by-for-pointers.c index f7b737d..c5729fd 100644 --- a/clang/test/CodeGen/attr-counted-by-for-pointers.c +++ b/clang/test/CodeGen/attr-counted-by-for-pointers.c @@ -33,47 +33,47 @@ struct annotated_ptr { // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 16 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 8 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2:![0-9]+]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT10:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3:![0-9]+]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6:![0-9]+]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT10:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7:![0-9]+]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB2:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR3:[0-9]+]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB2:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR3:[0-9]+]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT10]]: // SANITIZE-WITH-ATTR-NEXT: [[BUF:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 -// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA4:![0-9]+]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA8:![0-9]+]] // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 [[IDXPROM]] -// SANITIZE-WITH-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA13:![0-9]+]] +// SANITIZE-WITH-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA14:![0-9]+]] // SANITIZE-WITH-ATTR-NEXT: ret void // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local void @test1( // NO-SANITIZE-WITH-ATTR-SAME: ptr noundef readonly captures(none) [[P:%.*]], i32 noundef [[INDEX:%.*]], ptr noundef [[VALUE:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // NO-SANITIZE-WITH-ATTR-NEXT: [[ENTRY:.*:]] // NO-SANITIZE-WITH-ATTR-NEXT: [[BUF:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 -// NO-SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA2:![0-9]+]] +// NO-SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA6:![0-9]+]] // NO-SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // NO-SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i64 [[IDXPROM]] -// NO-SANITIZE-WITH-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA11:![0-9]+]] +// NO-SANITIZE-WITH-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA12:![0-9]+]] // NO-SANITIZE-WITH-ATTR-NEXT: ret void // // SANITIZE-WITHOUT-ATTR-LABEL: define dso_local void @test1( // SANITIZE-WITHOUT-ATTR-SAME: ptr noundef [[P:%.*]], i32 noundef [[INDEX:%.*]], ptr noundef [[VALUE:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // SANITIZE-WITHOUT-ATTR-NEXT: [[ENTRY:.*:]] // SANITIZE-WITHOUT-ATTR-NEXT: [[BUF:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 -// SANITIZE-WITHOUT-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA2:![0-9]+]] +// SANITIZE-WITHOUT-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA6:![0-9]+]] // SANITIZE-WITHOUT-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITHOUT-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i64 [[IDXPROM]] -// SANITIZE-WITHOUT-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA11:![0-9]+]] +// SANITIZE-WITHOUT-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA12:![0-9]+]] // SANITIZE-WITHOUT-ATTR-NEXT: ret void // // NO-SANITIZE-WITHOUT-ATTR-LABEL: define dso_local void @test1( // NO-SANITIZE-WITHOUT-ATTR-SAME: ptr noundef readonly captures(none) [[P:%.*]], i32 noundef [[INDEX:%.*]], ptr noundef [[VALUE:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // NO-SANITIZE-WITHOUT-ATTR-NEXT: [[ENTRY:.*:]] // NO-SANITIZE-WITHOUT-ATTR-NEXT: [[BUF:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 -// NO-SANITIZE-WITHOUT-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA2:![0-9]+]] +// NO-SANITIZE-WITHOUT-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA6:![0-9]+]] // NO-SANITIZE-WITHOUT-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // NO-SANITIZE-WITHOUT-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i64 [[IDXPROM]] -// NO-SANITIZE-WITHOUT-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA11:![0-9]+]] +// NO-SANITIZE-WITHOUT-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA12:![0-9]+]] // NO-SANITIZE-WITHOUT-ATTR-NEXT: ret void // void test1(struct annotated_ptr *p, int index, struct foo *value) { @@ -86,47 +86,47 @@ void test1(struct annotated_ptr *p, int index, struct foo *value) { // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 16 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 8 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT10:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT10:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB4:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR3]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB4:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR3]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT10]]: // SANITIZE-WITH-ATTR-NEXT: [[BUF:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 -// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA8]] // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 [[IDXPROM]] -// SANITIZE-WITH-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA13]] +// SANITIZE-WITH-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA14]] // SANITIZE-WITH-ATTR-NEXT: ret void // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local void @test2( // NO-SANITIZE-WITH-ATTR-SAME: ptr noundef readonly captures(none) [[P:%.*]], i32 noundef [[INDEX:%.*]], ptr noundef [[VALUE:%.*]]) local_unnamed_addr #[[ATTR0]] { // NO-SANITIZE-WITH-ATTR-NEXT: [[ENTRY:.*:]] // NO-SANITIZE-WITH-ATTR-NEXT: [[BUF:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 -// NO-SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA2]] +// NO-SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA6]] // NO-SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // NO-SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i64 [[IDXPROM]] -// NO-SANITIZE-WITH-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA11]] +// NO-SANITIZE-WITH-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA12]] // NO-SANITIZE-WITH-ATTR-NEXT: ret void // // SANITIZE-WITHOUT-ATTR-LABEL: define dso_local void @test2( // SANITIZE-WITHOUT-ATTR-SAME: ptr noundef [[P:%.*]], i32 noundef [[INDEX:%.*]], ptr noundef [[VALUE:%.*]]) local_unnamed_addr #[[ATTR0]] { // SANITIZE-WITHOUT-ATTR-NEXT: [[ENTRY:.*:]] // SANITIZE-WITHOUT-ATTR-NEXT: [[BUF:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 -// SANITIZE-WITHOUT-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA2]] +// SANITIZE-WITHOUT-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA6]] // SANITIZE-WITHOUT-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITHOUT-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i64 [[IDXPROM]] -// SANITIZE-WITHOUT-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA11]] +// SANITIZE-WITHOUT-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA12]] // SANITIZE-WITHOUT-ATTR-NEXT: ret void // // NO-SANITIZE-WITHOUT-ATTR-LABEL: define dso_local void @test2( // NO-SANITIZE-WITHOUT-ATTR-SAME: ptr noundef readonly captures(none) [[P:%.*]], i32 noundef [[INDEX:%.*]], ptr noundef [[VALUE:%.*]]) local_unnamed_addr #[[ATTR0]] { // NO-SANITIZE-WITHOUT-ATTR-NEXT: [[ENTRY:.*:]] // NO-SANITIZE-WITHOUT-ATTR-NEXT: [[BUF:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 -// NO-SANITIZE-WITHOUT-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA2]] +// NO-SANITIZE-WITHOUT-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA6]] // NO-SANITIZE-WITHOUT-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // NO-SANITIZE-WITHOUT-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i64 [[IDXPROM]] -// NO-SANITIZE-WITHOUT-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA11]] +// NO-SANITIZE-WITHOUT-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA12]] // NO-SANITIZE-WITHOUT-ATTR-NEXT: ret void // void test2(struct annotated_ptr *p, int index, struct foo *value) { @@ -139,47 +139,47 @@ void test2(struct annotated_ptr *p, int index, struct foo *value) { // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 16 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 8 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[DOTNOT:%.*]] = icmp ugt i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], label %[[CONT10:.*]], !prof [[PROF15:![0-9]+]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[DOTNOT:%.*]] = icmp ugt i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], label %[[CONT10:.*]], !prof [[PROF16:![0-9]+]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB5:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR3]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB5:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR3]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT10]]: // SANITIZE-WITH-ATTR-NEXT: [[BUF:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA8]] // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i64 [[IDXPROM]] -// SANITIZE-WITH-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA13]] +// SANITIZE-WITH-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA14]] // SANITIZE-WITH-ATTR-NEXT: ret void // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local void @test3( // NO-SANITIZE-WITH-ATTR-SAME: ptr noundef readonly captures(none) [[P:%.*]], i32 noundef [[INDEX:%.*]], ptr noundef [[VALUE:%.*]]) local_unnamed_addr #[[ATTR0]] { // NO-SANITIZE-WITH-ATTR-NEXT: [[ENTRY:.*:]] // NO-SANITIZE-WITH-ATTR-NEXT: [[BUF:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 -// NO-SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA2]] +// NO-SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA6]] // NO-SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // NO-SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i64 [[IDXPROM]] -// NO-SANITIZE-WITH-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA11]] +// NO-SANITIZE-WITH-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA12]] // NO-SANITIZE-WITH-ATTR-NEXT: ret void // // SANITIZE-WITHOUT-ATTR-LABEL: define dso_local void @test3( // SANITIZE-WITHOUT-ATTR-SAME: ptr noundef [[P:%.*]], i32 noundef [[INDEX:%.*]], ptr noundef [[VALUE:%.*]]) local_unnamed_addr #[[ATTR0]] { // SANITIZE-WITHOUT-ATTR-NEXT: [[ENTRY:.*:]] // SANITIZE-WITHOUT-ATTR-NEXT: [[BUF:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 -// SANITIZE-WITHOUT-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA2]] +// SANITIZE-WITHOUT-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA6]] // SANITIZE-WITHOUT-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITHOUT-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i64 [[IDXPROM]] -// SANITIZE-WITHOUT-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA11]] +// SANITIZE-WITHOUT-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA12]] // SANITIZE-WITHOUT-ATTR-NEXT: ret void // // NO-SANITIZE-WITHOUT-ATTR-LABEL: define dso_local void @test3( // NO-SANITIZE-WITHOUT-ATTR-SAME: ptr noundef readonly captures(none) [[P:%.*]], i32 noundef [[INDEX:%.*]], ptr noundef [[VALUE:%.*]]) local_unnamed_addr #[[ATTR0]] { // NO-SANITIZE-WITHOUT-ATTR-NEXT: [[ENTRY:.*:]] // NO-SANITIZE-WITHOUT-ATTR-NEXT: [[BUF:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 -// NO-SANITIZE-WITHOUT-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA2]] +// NO-SANITIZE-WITHOUT-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[BUF]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA6]] // NO-SANITIZE-WITHOUT-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // NO-SANITIZE-WITHOUT-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i64 [[IDXPROM]] -// NO-SANITIZE-WITHOUT-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA11]] +// NO-SANITIZE-WITHOUT-ATTR-NEXT: store ptr [[VALUE]], ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS3FOOPTR_TBAA12]] // NO-SANITIZE-WITHOUT-ATTR-NEXT: ret void // void test3(struct annotated_ptr *p, int index, struct foo *value) { @@ -264,12 +264,12 @@ size_t test5(struct annotated_ptr *p, int index) { // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 16 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[DOTNOT:%.*]] = icmp ugt i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], label %[[CONT8:.*]], !prof [[PROF15]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[DOTNOT:%.*]] = icmp ugt i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], label %[[CONT8:.*]], !prof [[PROF16]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB6:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR3]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB6:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR3]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT8]]: // SANITIZE-WITH-ATTR-NEXT: [[COUNT:%.*]] = sext i32 [[DOTCOUNTED_BY_LOAD]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = sub nsw i64 [[COUNT]], [[IDXPROM]] @@ -312,12 +312,12 @@ size_t test6(struct annotated_ptr *p, int index) { // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 16 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 8 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT10:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT10:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB7:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR3]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB7:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR3]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT10]]: // SANITIZE-WITH-ATTR-NEXT: ret i64 -1 // @@ -384,12 +384,12 @@ size_t test8(struct annotated_sized_ptr *p, int index) { // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 16 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[DOTNOT:%.*]] = icmp ugt i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], label %[[CONT8:.*]], !prof [[PROF15]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[DOTNOT:%.*]] = icmp ugt i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], label %[[CONT8:.*]], !prof [[PROF16]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB9:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR3]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB9:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR3]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT8]]: // SANITIZE-WITH-ATTR-NEXT: [[COUNT:%.*]] = sext i32 [[DOTCOUNTED_BY_LOAD]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[RESULT:%.*]] = sub nsw i64 [[COUNT]], [[IDXPROM]] @@ -430,12 +430,12 @@ size_t test9(struct annotated_sized_ptr *p, int index) { // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 16 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[DOTNOT:%.*]] = icmp ugt i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], label %[[CONT8:.*]], !prof [[PROF15]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[DOTNOT:%.*]] = icmp ugt i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], label %[[CONT8:.*]], !prof [[PROF16]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB11:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR3]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB11:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR3]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT8]]: // SANITIZE-WITH-ATTR-NEXT: [[COUNT:%.*]] = sext i32 [[DOTCOUNTED_BY_LOAD]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[INDEX_SIZE:%.*]] = shl nuw nsw i64 [[IDXPROM]], 2 @@ -549,54 +549,54 @@ size_t test12(struct pr151236_struct *p) { return __bdos(p->a) + __bdos(((int *)p->a)); } //. -// SANITIZE-WITH-ATTR: [[META2]] = !{} -// SANITIZE-WITH-ATTR: [[PROF3]] = !{!"branch_weights", i32 1048575, i32 1} -// SANITIZE-WITH-ATTR: [[_ZTS3FOOPTR_TBAA4]] = !{[[META5:![0-9]+]], [[META9:![0-9]+]], i64 8} -// SANITIZE-WITH-ATTR: [[META5]] = !{!"annotated_ptr", [[META6:![0-9]+]], i64 0, [[META9]], i64 8, [[META12:![0-9]+]], i64 16} -// SANITIZE-WITH-ATTR: [[META6]] = !{!"long", [[META7:![0-9]+]], i64 0} -// SANITIZE-WITH-ATTR: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0} -// SANITIZE-WITH-ATTR: [[META8]] = !{!"Simple C/C++ TBAA"} -// SANITIZE-WITH-ATTR: [[META9]] = !{!"p2 _ZTS3foo", [[META10:![0-9]+]], i64 0} -// SANITIZE-WITH-ATTR: [[META10]] = !{!"any p2 pointer", [[META11:![0-9]+]], i64 0} -// SANITIZE-WITH-ATTR: [[META11]] = !{!"any pointer", [[META7]], i64 0} -// SANITIZE-WITH-ATTR: [[META12]] = !{!"int", [[META7]], i64 0} -// SANITIZE-WITH-ATTR: [[_ZTS3FOOPTR_TBAA13]] = !{[[META14:![0-9]+]], [[META14]], i64 0} -// SANITIZE-WITH-ATTR: [[META14]] = !{!"p1 _ZTS3foo", [[META11]], i64 0} -// SANITIZE-WITH-ATTR: [[PROF15]] = !{!"branch_weights", i32 1, i32 1048575} +// SANITIZE-WITH-ATTR: [[META3:![0-9]+]] = !{!"int", [[META4:![0-9]+]], i64 0} +// SANITIZE-WITH-ATTR: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// SANITIZE-WITH-ATTR: [[META5]] = !{!"Simple C/C++ TBAA"} +// SANITIZE-WITH-ATTR: [[META6]] = !{} +// SANITIZE-WITH-ATTR: [[PROF7]] = !{!"branch_weights", i32 1048575, i32 1} +// SANITIZE-WITH-ATTR: [[_ZTS3FOOPTR_TBAA8]] = !{[[META9:![0-9]+]], [[META11:![0-9]+]], i64 8} +// SANITIZE-WITH-ATTR: [[META9]] = !{!"annotated_ptr", [[META10:![0-9]+]], i64 0, [[META11]], i64 8, [[META3]], i64 16} +// SANITIZE-WITH-ATTR: [[META10]] = !{!"long", [[META4]], i64 0} +// SANITIZE-WITH-ATTR: [[META11]] = !{!"p2 _ZTS3foo", [[META12:![0-9]+]], i64 0} +// SANITIZE-WITH-ATTR: [[META12]] = !{!"any p2 pointer", [[META13:![0-9]+]], i64 0} +// SANITIZE-WITH-ATTR: [[META13]] = !{!"any pointer", [[META4]], i64 0} +// SANITIZE-WITH-ATTR: [[_ZTS3FOOPTR_TBAA14]] = !{[[META15:![0-9]+]], [[META15]], i64 0} +// SANITIZE-WITH-ATTR: [[META15]] = !{!"p1 _ZTS3foo", [[META13]], i64 0} +// SANITIZE-WITH-ATTR: [[PROF16]] = !{!"branch_weights", i32 1, i32 1048575} //. -// NO-SANITIZE-WITH-ATTR: [[_ZTS3FOOPTR_TBAA2]] = !{[[META3:![0-9]+]], [[META7:![0-9]+]], i64 8} -// NO-SANITIZE-WITH-ATTR: [[META3]] = !{!"annotated_ptr", [[META4:![0-9]+]], i64 0, [[META7]], i64 8, [[META10:![0-9]+]], i64 16} -// NO-SANITIZE-WITH-ATTR: [[META4]] = !{!"long", [[META5:![0-9]+]], i64 0} -// NO-SANITIZE-WITH-ATTR: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// NO-SANITIZE-WITH-ATTR: [[META6]] = !{!"Simple C/C++ TBAA"} -// NO-SANITIZE-WITH-ATTR: [[META7]] = !{!"p2 _ZTS3foo", [[META8:![0-9]+]], i64 0} -// NO-SANITIZE-WITH-ATTR: [[META8]] = !{!"any p2 pointer", [[META9:![0-9]+]], i64 0} -// NO-SANITIZE-WITH-ATTR: [[META9]] = !{!"any pointer", [[META5]], i64 0} -// NO-SANITIZE-WITH-ATTR: [[META10]] = !{!"int", [[META5]], i64 0} -// NO-SANITIZE-WITH-ATTR: [[_ZTS3FOOPTR_TBAA11]] = !{[[META12:![0-9]+]], [[META12]], i64 0} -// NO-SANITIZE-WITH-ATTR: [[META12]] = !{!"p1 _ZTS3foo", [[META9]], i64 0} +// NO-SANITIZE-WITH-ATTR: [[META3:![0-9]+]] = !{!"int", [[META4:![0-9]+]], i64 0} +// NO-SANITIZE-WITH-ATTR: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// NO-SANITIZE-WITH-ATTR: [[META5]] = !{!"Simple C/C++ TBAA"} +// NO-SANITIZE-WITH-ATTR: [[_ZTS3FOOPTR_TBAA6]] = !{[[META7:![0-9]+]], [[META9:![0-9]+]], i64 8} +// NO-SANITIZE-WITH-ATTR: [[META7]] = !{!"annotated_ptr", [[META8:![0-9]+]], i64 0, [[META9]], i64 8, [[META3]], i64 16} +// NO-SANITIZE-WITH-ATTR: [[META8]] = !{!"long", [[META4]], i64 0} +// NO-SANITIZE-WITH-ATTR: [[META9]] = !{!"p2 _ZTS3foo", [[META10:![0-9]+]], i64 0} +// NO-SANITIZE-WITH-ATTR: [[META10]] = !{!"any p2 pointer", [[META11:![0-9]+]], i64 0} +// NO-SANITIZE-WITH-ATTR: [[META11]] = !{!"any pointer", [[META4]], i64 0} +// NO-SANITIZE-WITH-ATTR: [[_ZTS3FOOPTR_TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} +// NO-SANITIZE-WITH-ATTR: [[META13]] = !{!"p1 _ZTS3foo", [[META11]], i64 0} //. -// SANITIZE-WITHOUT-ATTR: [[_ZTS3FOOPTR_TBAA2]] = !{[[META3:![0-9]+]], [[META7:![0-9]+]], i64 8} -// SANITIZE-WITHOUT-ATTR: [[META3]] = !{!"annotated_ptr", [[META4:![0-9]+]], i64 0, [[META7]], i64 8, [[META10:![0-9]+]], i64 16} -// SANITIZE-WITHOUT-ATTR: [[META4]] = !{!"long", [[META5:![0-9]+]], i64 0} -// SANITIZE-WITHOUT-ATTR: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// SANITIZE-WITHOUT-ATTR: [[META6]] = !{!"Simple C/C++ TBAA"} -// SANITIZE-WITHOUT-ATTR: [[META7]] = !{!"p2 _ZTS3foo", [[META8:![0-9]+]], i64 0} -// SANITIZE-WITHOUT-ATTR: [[META8]] = !{!"any p2 pointer", [[META9:![0-9]+]], i64 0} -// SANITIZE-WITHOUT-ATTR: [[META9]] = !{!"any pointer", [[META5]], i64 0} -// SANITIZE-WITHOUT-ATTR: [[META10]] = !{!"int", [[META5]], i64 0} -// SANITIZE-WITHOUT-ATTR: [[_ZTS3FOOPTR_TBAA11]] = !{[[META12:![0-9]+]], [[META12]], i64 0} -// SANITIZE-WITHOUT-ATTR: [[META12]] = !{!"p1 _ZTS3foo", [[META9]], i64 0} +// SANITIZE-WITHOUT-ATTR: [[META3:![0-9]+]] = !{!"int", [[META4:![0-9]+]], i64 0} +// SANITIZE-WITHOUT-ATTR: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// SANITIZE-WITHOUT-ATTR: [[META5]] = !{!"Simple C/C++ TBAA"} +// SANITIZE-WITHOUT-ATTR: [[_ZTS3FOOPTR_TBAA6]] = !{[[META7:![0-9]+]], [[META9:![0-9]+]], i64 8} +// SANITIZE-WITHOUT-ATTR: [[META7]] = !{!"annotated_ptr", [[META8:![0-9]+]], i64 0, [[META9]], i64 8, [[META3]], i64 16} +// SANITIZE-WITHOUT-ATTR: [[META8]] = !{!"long", [[META4]], i64 0} +// SANITIZE-WITHOUT-ATTR: [[META9]] = !{!"p2 _ZTS3foo", [[META10:![0-9]+]], i64 0} +// SANITIZE-WITHOUT-ATTR: [[META10]] = !{!"any p2 pointer", [[META11:![0-9]+]], i64 0} +// SANITIZE-WITHOUT-ATTR: [[META11]] = !{!"any pointer", [[META4]], i64 0} +// SANITIZE-WITHOUT-ATTR: [[_ZTS3FOOPTR_TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} +// SANITIZE-WITHOUT-ATTR: [[META13]] = !{!"p1 _ZTS3foo", [[META11]], i64 0} //. -// NO-SANITIZE-WITHOUT-ATTR: [[_ZTS3FOOPTR_TBAA2]] = !{[[META3:![0-9]+]], [[META7:![0-9]+]], i64 8} -// NO-SANITIZE-WITHOUT-ATTR: [[META3]] = !{!"annotated_ptr", [[META4:![0-9]+]], i64 0, [[META7]], i64 8, [[META10:![0-9]+]], i64 16} -// NO-SANITIZE-WITHOUT-ATTR: [[META4]] = !{!"long", [[META5:![0-9]+]], i64 0} -// NO-SANITIZE-WITHOUT-ATTR: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// NO-SANITIZE-WITHOUT-ATTR: [[META6]] = !{!"Simple C/C++ TBAA"} -// NO-SANITIZE-WITHOUT-ATTR: [[META7]] = !{!"p2 _ZTS3foo", [[META8:![0-9]+]], i64 0} -// NO-SANITIZE-WITHOUT-ATTR: [[META8]] = !{!"any p2 pointer", [[META9:![0-9]+]], i64 0} -// NO-SANITIZE-WITHOUT-ATTR: [[META9]] = !{!"any pointer", [[META5]], i64 0} -// NO-SANITIZE-WITHOUT-ATTR: [[META10]] = !{!"int", [[META5]], i64 0} -// NO-SANITIZE-WITHOUT-ATTR: [[_ZTS3FOOPTR_TBAA11]] = !{[[META12:![0-9]+]], [[META12]], i64 0} -// NO-SANITIZE-WITHOUT-ATTR: [[META12]] = !{!"p1 _ZTS3foo", [[META9]], i64 0} +// NO-SANITIZE-WITHOUT-ATTR: [[META3:![0-9]+]] = !{!"int", [[META4:![0-9]+]], i64 0} +// NO-SANITIZE-WITHOUT-ATTR: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// NO-SANITIZE-WITHOUT-ATTR: [[META5]] = !{!"Simple C/C++ TBAA"} +// NO-SANITIZE-WITHOUT-ATTR: [[_ZTS3FOOPTR_TBAA6]] = !{[[META7:![0-9]+]], [[META9:![0-9]+]], i64 8} +// NO-SANITIZE-WITHOUT-ATTR: [[META7]] = !{!"annotated_ptr", [[META8:![0-9]+]], i64 0, [[META9]], i64 8, [[META3]], i64 16} +// NO-SANITIZE-WITHOUT-ATTR: [[META8]] = !{!"long", [[META4]], i64 0} +// NO-SANITIZE-WITHOUT-ATTR: [[META9]] = !{!"p2 _ZTS3foo", [[META10:![0-9]+]], i64 0} +// NO-SANITIZE-WITHOUT-ATTR: [[META10]] = !{!"any p2 pointer", [[META11:![0-9]+]], i64 0} +// NO-SANITIZE-WITHOUT-ATTR: [[META11]] = !{!"any pointer", [[META4]], i64 0} +// NO-SANITIZE-WITHOUT-ATTR: [[_ZTS3FOOPTR_TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} +// NO-SANITIZE-WITHOUT-ATTR: [[META13]] = !{!"p1 _ZTS3foo", [[META11]], i64 0} //. diff --git a/clang/test/CodeGen/attr-counted-by-pr110385.c b/clang/test/CodeGen/attr-counted-by-pr110385.c index 32ee1c8..e854ed4 100644 --- a/clang/test/CodeGen/attr-counted-by-pr110385.c +++ b/clang/test/CodeGen/attr-counted-by-pr110385.c @@ -29,7 +29,7 @@ void init(void * __attribute__((pass_dynamic_object_size(0)))); // CHECK-SAME: ptr noundef readonly captures(none) [[FOO:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[GROWABLE:%.*]] = getelementptr inbounds nuw i8, ptr [[FOO]], i64 8 -// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[GROWABLE]], align 8, !tbaa [[_ZTS8VARIABLEPTR_TBAA2:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[GROWABLE]], align 8, !tbaa [[_ZTS8VARIABLEPTR_TBAA6:![0-9]+]] // CHECK-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 12 // CHECK-NEXT: [[COUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 8 // CHECK-NEXT: [[COUNTED_BY_LOAD:%.*]] = load i32, ptr [[COUNTED_BY_GEP]], align 4 @@ -61,11 +61,11 @@ void test2(struct bucket2 *foo) { init(foo->growable.array); } //. -// CHECK: [[_ZTS8VARIABLEPTR_TBAA2]] = !{[[META3:![0-9]+]], [[META7:![0-9]+]], i64 8} -// CHECK: [[META3]] = !{!"bucket", [[META4:![0-9]+]], i64 0, [[META7]], i64 8, [[META4]], i64 16} -// CHECK: [[META4]] = !{!"int", [[META5:![0-9]+]], i64 0} -// CHECK: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// CHECK: [[META6]] = !{!"Simple C/C++ TBAA"} -// CHECK: [[META7]] = !{!"p1 _ZTS8variable", [[META8:![0-9]+]], i64 0} -// CHECK: [[META8]] = !{!"any pointer", [[META5]], i64 0} +// CHECK: [[META3:![0-9]+]] = !{!"int", [[META4:![0-9]+]], i64 0} +// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[_ZTS8VARIABLEPTR_TBAA6]] = !{[[META7:![0-9]+]], [[META8:![0-9]+]], i64 8} +// CHECK: [[META7]] = !{!"bucket", [[META3]], i64 0, [[META8]], i64 8, [[META3]], i64 16} +// CHECK: [[META8]] = !{!"p1 _ZTS8variable", [[META9:![0-9]+]], i64 0} +// CHECK: [[META9]] = !{!"any pointer", [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/attr-counted-by.c b/clang/test/CodeGen/attr-counted-by.c index 9675fe2..86c59fb 100644 --- a/clang/test/CodeGen/attr-counted-by.c +++ b/clang/test/CodeGen/attr-counted-by.c @@ -64,16 +64,16 @@ struct anon_struct { // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2:![0-9]+]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3:![0-9]+]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6:![0-9]+]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7:![0-9]+]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB1:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8:[0-9]+]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB1:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8:[0-9]+]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT3]]: // SANITIZE-WITH-ATTR-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 12 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAY]], i64 [[IDXPROM]] -// SANITIZE-WITH-ATTR-NEXT: store i32 [[VAL]], ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA4:![0-9]+]] +// SANITIZE-WITH-ATTR-NEXT: store i32 [[VAL]], ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA2:![0-9]+]] // SANITIZE-WITH-ATTR-NEXT: ret void // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local void @test1( @@ -112,18 +112,18 @@ void test1(struct annotated *p, int index, int val) { // SANITIZE-WITH-ATTR-NEXT: [[ENTRY:.*:]] // SANITIZE-WITH-ATTR-NEXT: [[COUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[COUNTED_BY_LOAD:%.*]] = load i32, ptr [[COUNTED_BY_GEP]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[COUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[INDEX]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT6:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[COUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[INDEX]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT6:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB3:[0-9]+]], i64 [[INDEX]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB3:[0-9]+]], i64 [[INDEX]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT6]]: // SANITIZE-WITH-ATTR-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 12 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAY]], i64 [[INDEX]] // SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = tail call i32 @llvm.smax.i32(i32 [[COUNTED_BY_LOAD]], i32 0) // SANITIZE-WITH-ATTR-NEXT: [[CONV:%.*]] = shl i32 [[TMP2]], 2 -// SANITIZE-WITH-ATTR-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: ret void // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local void @test2( @@ -235,16 +235,16 @@ size_t test2_bdos_cast(struct annotated *p) { // SANITIZE-WITH-ATTR-NEXT: [[ENTRY:.*:]] // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[INDEX]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[INDEX]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB4:[0-9]+]], i64 [[INDEX]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB4:[0-9]+]], i64 [[INDEX]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT3]]: // SANITIZE-WITH-ATTR-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 12 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAY]], i64 [[INDEX]] -// SANITIZE-WITH-ATTR-NEXT: store i32 -1, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: store i32 -1, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: ret void // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local void @test3( @@ -331,63 +331,63 @@ size_t test3_bdos_cast(struct annotated *p) { // SANITIZE-WITH-ATTR-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 12 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] // SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ugt i32 [[DOTCOUNTED_BY_LOAD]], 2 -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT1:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT1:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB5:[0-9]+]], i64 3) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB5:[0-9]+]], i64 3) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT1]]: // SANITIZE-WITH-ATTR-NEXT: [[FLEXIBLE_ARRAY_MEMBER_SIZE:%.*]] = shl i32 [[DOTCOUNTED_BY_LOAD]], 2 // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 -// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP2]], label %[[CONT12:.*]], label %[[HANDLER_OUT_OF_BOUNDS8:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP2]], label %[[CONT12:.*]], label %[[HANDLER_OUT_OF_BOUNDS8:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS8]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB6:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB6:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT12]]: // SANITIZE-WITH-ATTR-NEXT: [[TMP3:%.*]] = icmp sgt i32 [[DOTCOUNTED_BY_LOAD]], 2 // SANITIZE-WITH-ATTR-NEXT: [[RESULT:%.*]] = add i32 [[FLEXIBLE_ARRAY_MEMBER_SIZE]], 244 // SANITIZE-WITH-ATTR-NEXT: [[TMP4:%.*]] = and i32 [[RESULT]], 252 // SANITIZE-WITH-ATTR-NEXT: [[CONV2:%.*]] = select i1 [[TMP3]], i32 [[TMP4]], i32 0 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAY]], i64 [[IDXPROM]] -// SANITIZE-WITH-ATTR-NEXT: store i32 [[CONV2]], ptr [[ARRAYIDX10]], align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: store i32 [[CONV2]], ptr [[ARRAYIDX10]], align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: [[DOTNOT81:%.*]] = icmp eq i32 [[DOTCOUNTED_BY_LOAD]], 3 -// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT81]], label %[[HANDLER_OUT_OF_BOUNDS18:.*]], label %[[CONT19:.*]], !prof [[PROF8:![0-9]+]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT81]], label %[[HANDLER_OUT_OF_BOUNDS18:.*]], label %[[CONT19:.*]], !prof [[PROF8:![0-9]+]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS18]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB7:[0-9]+]], i64 4) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB7:[0-9]+]], i64 4) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT19]]: // SANITIZE-WITH-ATTR-NEXT: [[ADD:%.*]] = add nsw i32 [[INDEX]], 1 // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM31:%.*]] = sext i32 [[ADD]] to i64 -// SANITIZE-WITH-ATTR-NEXT: [[TMP5:%.*]] = icmp ult i64 [[IDXPROM31]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP5]], label %[[CONT38:.*]], label %[[HANDLER_OUT_OF_BOUNDS34:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP5:%.*]] = icmp ult i64 [[IDXPROM31]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP5]], label %[[CONT38:.*]], label %[[HANDLER_OUT_OF_BOUNDS34:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS34]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB8:[0-9]+]], i64 [[IDXPROM31]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB8:[0-9]+]], i64 [[IDXPROM31]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT38]]: // SANITIZE-WITH-ATTR-NEXT: [[TMP6:%.*]] = icmp sgt i32 [[DOTCOUNTED_BY_LOAD]], 3 // SANITIZE-WITH-ATTR-NEXT: [[RESULT25:%.*]] = add i32 [[FLEXIBLE_ARRAY_MEMBER_SIZE]], 240 // SANITIZE-WITH-ATTR-NEXT: [[TMP7:%.*]] = and i32 [[RESULT25]], 252 // SANITIZE-WITH-ATTR-NEXT: [[CONV27:%.*]] = select i1 [[TMP6]], i32 [[TMP7]], i32 0 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX36:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAY]], i64 [[IDXPROM31]] -// SANITIZE-WITH-ATTR-NEXT: store i32 [[CONV27]], ptr [[ARRAYIDX36]], align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: store i32 [[CONV27]], ptr [[ARRAYIDX36]], align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM42:%.*]] = sext i32 [[FAM_IDX]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD44:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP8:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD44]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[DOTNOT:%.*]] = icmp ugt i64 [[IDXPROM42]], [[TMP8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT]], label %[[HANDLER_OUT_OF_BOUNDS45:.*]], label %[[CONT46:.*]], !prof [[PROF8]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP8:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD44]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[DOTNOT:%.*]] = icmp ugt i64 [[IDXPROM42]], [[TMP8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT]], label %[[HANDLER_OUT_OF_BOUNDS45:.*]], label %[[CONT46:.*]], !prof [[PROF8]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS45]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB9:[0-9]+]], i64 [[IDXPROM42]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB9:[0-9]+]], i64 [[IDXPROM42]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT46]]: // SANITIZE-WITH-ATTR-NEXT: [[ADD59:%.*]] = add nsw i32 [[INDEX]], 2 // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM60:%.*]] = sext i32 [[ADD59]] to i64 -// SANITIZE-WITH-ATTR-NEXT: [[TMP9:%.*]] = icmp ult i64 [[IDXPROM60]], [[TMP8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP9]], label %[[CONT67:.*]], label %[[HANDLER_OUT_OF_BOUNDS63:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP9:%.*]] = icmp ult i64 [[IDXPROM60]], [[TMP8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP9]], label %[[CONT67:.*]], label %[[HANDLER_OUT_OF_BOUNDS63:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS63]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB10:[0-9]+]], i64 [[IDXPROM60]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB10:[0-9]+]], i64 [[IDXPROM60]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT67]]: // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX65:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAY]], i64 [[IDXPROM60]] // SANITIZE-WITH-ATTR-NEXT: [[COUNT50:%.*]] = sext i32 [[DOTCOUNTED_BY_LOAD44]] to i64 @@ -396,7 +396,7 @@ size_t test3_bdos_cast(struct annotated *p) { // SANITIZE-WITH-ATTR-NEXT: [[DOTTR:%.*]] = trunc nuw nsw i64 [[TMP11]] to i32 // SANITIZE-WITH-ATTR-NEXT: [[CONV54:%.*]] = shl i32 [[DOTTR]], 2 // SANITIZE-WITH-ATTR-NEXT: [[CONV55:%.*]] = and i32 [[CONV54]], 252 -// SANITIZE-WITH-ATTR-NEXT: store i32 [[CONV55]], ptr [[ARRAYIDX65]], align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: store i32 [[CONV55]], ptr [[ARRAYIDX65]], align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: ret void // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local void @test4( @@ -475,12 +475,12 @@ void test4(struct annotated *p, int index, int fam_idx) { // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[DOTNOT:%.*]] = icmp ugt i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], label %[[CONT1:.*]], !prof [[PROF8]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[DOTNOT:%.*]] = icmp ugt i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], label %[[CONT1:.*]], !prof [[PROF8]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB11:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB11:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT1]]: // SANITIZE-WITH-ATTR-NEXT: [[COUNT:%.*]] = sext i32 [[DOTCOUNTED_BY_LOAD]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = sub nsw i64 [[COUNT]], [[IDXPROM]] @@ -613,15 +613,15 @@ size_t test4_bdos_cast2(struct annotated *p, int index) { // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i64, ptr [[DOTCOUNTED_BY_GEP]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = icmp ugt i64 [[DOTCOUNTED_BY_LOAD]], [[IDXPROM]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = icmp ugt i64 [[DOTCOUNTED_BY_LOAD]], [[IDXPROM]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB12:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB12:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT3]]: // SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 16 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[IDXPROM]] -// SANITIZE-WITH-ATTR-NEXT: store i32 -1, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: store i32 -1, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: ret void // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local void @test5( @@ -685,18 +685,18 @@ size_t test5_bdos(struct anon_struct *p) { // SANITIZE-WITH-ATTR-NEXT: [[COUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[COUNTED_BY_LOAD:%.*]] = load i64, ptr [[COUNTED_BY_GEP]], align 4 // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = icmp ugt i64 [[COUNTED_BY_LOAD]], [[IDXPROM]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT6:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = icmp ugt i64 [[COUNTED_BY_LOAD]], [[IDXPROM]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT6:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB13:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB13:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT6]]: // SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 16 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[IDXPROM]] // SANITIZE-WITH-ATTR-NEXT: [[FLEXIBLE_ARRAY_MEMBER_SIZE:%.*]] = shl nuw i64 [[COUNTED_BY_LOAD]], 2 // SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = tail call i64 @llvm.smax.i64(i64 [[FLEXIBLE_ARRAY_MEMBER_SIZE]], i64 0) // SANITIZE-WITH-ATTR-NEXT: [[CONV:%.*]] = trunc i64 [[TMP2]] to i32 -// SANITIZE-WITH-ATTR-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: ret void // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local void @test6( @@ -773,12 +773,12 @@ size_t test6_bdos(struct anon_struct *p) { // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i8, ptr [[TMP0]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i8 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP1]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP2]], label %[[CONT7:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i8 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP1]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP2]], label %[[CONT7:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB15:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB15:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT7]]: // SANITIZE-WITH-ATTR-NEXT: [[INTS:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 9 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[INTS]], i64 [[IDXPROM]] @@ -846,12 +846,12 @@ size_t test7_bdos(struct union_of_fams *p) { // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[COUNTED_BY_LOAD:%.*]] = load i8, ptr [[TMP0]], align 4 // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i8 [[COUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP1]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP2]], label %[[CONT14:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i8 [[COUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP1]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP2]], label %[[CONT14:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB16:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB16:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT14]]: // SANITIZE-WITH-ATTR-NEXT: [[INTS:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 9 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[INTS]], i64 [[IDXPROM]] @@ -927,12 +927,12 @@ size_t test8_bdos(struct union_of_fams *p) { // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[TMP0]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP1]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP2]], label %[[CONT7:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP1]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP2]], label %[[CONT7:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB18:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB18:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT7]]: // SANITIZE-WITH-ATTR-NEXT: [[BYTES:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 12 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[BYTES]], i64 [[IDXPROM]] @@ -1000,12 +1000,12 @@ size_t test9_bdos(struct union_of_fams *p) { // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[COUNTED_BY_LOAD:%.*]] = load i32, ptr [[TMP0]], align 4 // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i32 [[COUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP1]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP2]], label %[[CONT14:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i32 [[COUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP1]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP2]], label %[[CONT14:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB19:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB19:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT14]]: // SANITIZE-WITH-ATTR-NEXT: [[BYTES:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 12 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[BYTES]], i64 [[IDXPROM]] @@ -1087,12 +1087,12 @@ size_t test10_bdos(struct union_of_fams *p) { // SANITIZE-WITH-ATTR-NEXT: [[COUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[COUNTED_BY_LOAD:%.*]] = load i32, ptr [[COUNTED_BY_GEP]], align 4 // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[INDEX]] to i64 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[COUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT6:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[COUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT6:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB20:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB20:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT6]]: // SANITIZE-WITH-ATTR-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 12 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAY]], i64 [[IDXPROM]] @@ -1100,7 +1100,7 @@ size_t test10_bdos(struct union_of_fams *p) { // SANITIZE-WITH-ATTR-NEXT: [[FLEXIBLE_ARRAY_MEMBER_SIZE:%.*]] = shl i32 [[COUNTED_BY_LOAD]], 2 // SANITIZE-WITH-ATTR-NEXT: [[RESULT:%.*]] = add i32 [[FLEXIBLE_ARRAY_MEMBER_SIZE]], 8 // SANITIZE-WITH-ATTR-NEXT: [[CONV:%.*]] = select i1 [[TMP2]], i32 [[RESULT]], i32 0 -// SANITIZE-WITH-ATTR-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: ret void // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local void @test11( @@ -1201,23 +1201,23 @@ int test12_a, test12_b; // SANITIZE-WITH-ATTR-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 4 dereferenceable(24) [[BAZ]], ptr noundef nonnull align 4 dereferenceable(24) @test12_bar, i64 24, i1 false), !tbaa.struct [[TBAA_STRUCT10:![0-9]+]] // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = icmp ult i32 [[INDEX]], 6 // SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i32 [[INDEX]] to i64 -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB22:[0-9]+]], i64 [[TMP1]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB22:[0-9]+]], i64 [[TMP1]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT]]: // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[BAZ]], i64 [[TMP1]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA4]] -// SANITIZE-WITH-ATTR-NEXT: store i32 [[TMP2]], ptr @test12_b, align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA2]] +// SANITIZE-WITH-ATTR-NEXT: store i32 [[TMP2]], ptr @test12_b, align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr @test12_foo, align 4 // SANITIZE-WITH-ATTR-NEXT: [[DOTNOT:%.*]] = icmp eq i32 [[DOTCOUNTED_BY_LOAD]], 0 -// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT]], label %[[HANDLER_OUT_OF_BOUNDS4:.*]], label %[[HANDLER_TYPE_MISMATCH6:.*]], !prof [[PROF8]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[DOTNOT]], label %[[HANDLER_OUT_OF_BOUNDS4:.*]], label %[[HANDLER_TYPE_MISMATCH6:.*]], !prof [[PROF8]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS4]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB24:[0-9]+]], i64 0) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB24:[0-9]+]], i64 0) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_TYPE_MISMATCH6]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_type_mismatch_v1_abort(ptr nonnull @[[GLOB25:[0-9]+]], i64 ptrtoint (ptr getelementptr inbounds nuw (i8, ptr @test12_foo, i64 4) to i64)) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_type_mismatch_v1_abort(ptr nonnull @[[GLOB25:[0-9]+]], i64 ptrtoint (ptr getelementptr inbounds nuw (i8, ptr @test12_foo, i64 4) to i64)) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local noundef i32 @test12( // NO-SANITIZE-WITH-ATTR-SAME: i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR4:[0-9]+]] { @@ -1302,12 +1302,12 @@ struct test13_bar { // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr @test13_f, align 8, !tbaa [[_ZTS10TEST13_BARPTR_TBAA11:![0-9]+]] // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = icmp ult i64 [[INDEX]], [[TMP1]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP2]], label %[[CONT5:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = icmp ult i64 [[INDEX]], [[TMP1]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP2]], label %[[CONT5:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB28:[0-9]+]], i64 [[INDEX]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB28:[0-9]+]], i64 [[INDEX]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT5]]: // SANITIZE-WITH-ATTR-NEXT: [[REVMAP:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 16 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw ptr, ptr [[REVMAP]], i64 [[INDEX]] @@ -1364,11 +1364,11 @@ struct test14_foo { // SANITIZE-WITH-ATTR-SAME: i32 noundef [[IDX:%.*]]) local_unnamed_addr #[[ATTR0]] { // SANITIZE-WITH-ATTR-NEXT: [[ENTRY:.*:]] // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = icmp eq i32 [[IDX]], 0 -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[IDX]] to i64 -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB29:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB29:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT3]]: // SANITIZE-WITH-ATTR-NEXT: ret i32 undef // @@ -1418,11 +1418,11 @@ int test14(int idx) { // SANITIZE-WITH-ATTR-SAME: i32 noundef [[IDX:%.*]]) local_unnamed_addr #[[ATTR0]] { // SANITIZE-WITH-ATTR-NEXT: [[ENTRY:.*:]] // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = icmp eq i32 [[IDX]], 0 -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT1:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT1:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[IDX]] to i64 -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB31:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB31:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT1]]: // SANITIZE-WITH-ATTR-NEXT: ret i32 undef // @@ -1469,10 +1469,10 @@ int test15(int idx) { // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 680 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 4 // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = icmp ugt i32 [[DOTCOUNTED_BY_LOAD]], 1 -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT1:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT1:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB32:[0-9]+]], i64 2) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB32:[0-9]+]], i64 2) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT1]]: // SANITIZE-WITH-ATTR-NEXT: ret i64 -1 // @@ -1607,13 +1607,13 @@ struct tests_foo { // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[VAR]], i64 40 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = icmp ugt i32 [[DOTCOUNTED_BY_LOAD]], 10 -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT4:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT4:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB33:[0-9]+]], i64 10) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB33:[0-9]+]], i64 10) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT4]]: // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[VAR]], i64 84 -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: ret i32 [[TMP1]] // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local i32 @test24( @@ -1648,13 +1648,13 @@ int test24(int c, struct tests_foo *var) { // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = load ptr, ptr [[VAR]], align 8, !tbaa [[_ZTS9TESTS_FOOPTR_TBAA17:![0-9]+]] // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[TMP0]], align 4 // SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ugt i32 [[DOTCOUNTED_BY_LOAD]], 10 -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT5:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT5:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB34:[0-9]+]], i64 10) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB34:[0-9]+]], i64 10) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT5]]: // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 44 -// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: ret i32 [[TMP2]] // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local i32 @test25( @@ -1698,16 +1698,16 @@ struct test26_foo { // SANITIZE-WITH-ATTR-NEXT: [[S:%.*]] = getelementptr inbounds nuw i8, ptr [[FOO]], i64 4 // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[C]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[S]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT5:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT5:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB35:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB35:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT5]]: // SANITIZE-WITH-ATTR-NEXT: [[ARR:%.*]] = getelementptr inbounds nuw i8, ptr [[FOO]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[ARR]], i64 [[IDXPROM]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: ret i32 [[TMP2]] // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local i32 @test26( @@ -1769,12 +1769,12 @@ struct test27_foo { // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[I]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 12 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB37:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB37:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT3]]: // SANITIZE-WITH-ATTR-NEXT: [[ENTRIES:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 24 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw ptr, ptr [[ENTRIES]], i64 [[IDXPROM]] @@ -1835,16 +1835,16 @@ struct test28_foo { // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM:%.*]] = sext i32 [[I]] to i64 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP2]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP3:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP4:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP3]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP4]], label %[[CONT17:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP3:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP4:%.*]] = icmp ult i64 [[IDXPROM]], [[TMP3]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP4]], label %[[CONT17:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB39:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB39:[0-9]+]], i64 [[IDXPROM]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT17]]: // SANITIZE-WITH-ATTR-NEXT: [[ARR:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP2]], i64 12 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[ARR]], i64 [[IDXPROM]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: ret i32 [[TMP5]] // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local i32 @test28( @@ -1899,28 +1899,28 @@ struct annotated_struct_array { // SANITIZE-WITH-ATTR-NEXT: [[ENTRY:.*:]] // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = icmp ult i32 [[IDX1]], 10 // SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i32 [[IDX1]] to i64 -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB41:[0-9]+]], i64 [[TMP1]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB41:[0-9]+]], i64 [[TMP1]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT3]]: // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw ptr, ptr [[ANN]], i64 [[TMP1]] // SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = load ptr, ptr [[ARRAYIDX]], align 8, !tbaa [[_ZTS9ANNOTATEDPTR_TBAA23:![0-9]+]] // SANITIZE-WITH-ATTR-NEXT: [[COUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP2]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[COUNTED_BY_LOAD:%.*]] = load i32, ptr [[COUNTED_BY_GEP]], align 4 // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM27:%.*]] = sext i32 [[IDX2]] to i64 -// SANITIZE-WITH-ATTR-NEXT: [[TMP3:%.*]] = zext i32 [[COUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP4:%.*]] = icmp ult i64 [[IDXPROM27]], [[TMP3]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP4]], label %[[CONT32:.*]], label %[[HANDLER_OUT_OF_BOUNDS28:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP3:%.*]] = zext i32 [[COUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP4:%.*]] = icmp ult i64 [[IDXPROM27]], [[TMP3]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP4]], label %[[CONT32:.*]], label %[[HANDLER_OUT_OF_BOUNDS28:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS28]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB42:[0-9]+]], i64 [[IDXPROM27]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB42:[0-9]+]], i64 [[IDXPROM27]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT32]]: // SANITIZE-WITH-ATTR-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP2]], i64 12 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX30:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAY]], i64 [[IDXPROM27]] // SANITIZE-WITH-ATTR-NEXT: [[TMP5:%.*]] = tail call i32 @llvm.smax.i32(i32 [[COUNTED_BY_LOAD]], i32 0) // SANITIZE-WITH-ATTR-NEXT: [[CONV:%.*]] = shl i32 [[TMP5]], 2 -// SANITIZE-WITH-ATTR-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX30]], align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX30]], align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: ret void // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local void @test29( @@ -1987,9 +1987,9 @@ struct test30_struct { // SANITIZE-WITH-ATTR-LABEL: define dso_local void @test30( // SANITIZE-WITH-ATTR-SAME: ptr noundef [[PTR:%.*]], i32 noundef [[IDX:%.*]]) local_unnamed_addr #[[ATTR3]] { // SANITIZE-WITH-ATTR-NEXT: [[ENTRY:.*:]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[IDX]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB44:[0-9]+]], i64 [[TMP0]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[IDX]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB44:[0-9]+]], i64 [[TMP0]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local void @test30( // NO-SANITIZE-WITH-ATTR-SAME: ptr noundef [[PTR:%.*]], i32 noundef [[IDX:%.*]]) local_unnamed_addr #[[ATTR0]] { @@ -2062,21 +2062,21 @@ struct annotated_with_array { // SANITIZE-WITH-ATTR-SAME: ptr noundef [[PTR:%.*]], i32 noundef [[IDX1:%.*]], i32 noundef [[IDX2:%.*]]) local_unnamed_addr #[[ATTR0]] { // SANITIZE-WITH-ATTR-NEXT: [[ENTRY:.*:]] // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = icmp ult i32 [[IDX2]], 43 -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT1:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT1:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i32 [[IDX2]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB46:[0-9]+]], i64 [[TMP1]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i32 [[IDX2]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB46:[0-9]+]], i64 [[TMP1]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT1]]: // SANITIZE-WITH-ATTR-NEXT: [[COUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[PTR]], i64 336 // SANITIZE-WITH-ATTR-NEXT: [[COUNTED_BY_LOAD:%.*]] = load i32, ptr [[COUNTED_BY_GEP]], align 4 // SANITIZE-WITH-ATTR-NEXT: [[IDXPROM4:%.*]] = sext i32 [[IDX1]] to i64 -// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = zext i32 [[COUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP3:%.*]] = icmp ult i64 [[IDXPROM4]], [[TMP2]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP3]], label %[[CONT9:.*]], label %[[HANDLER_OUT_OF_BOUNDS5:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = zext i32 [[COUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP3:%.*]] = icmp ult i64 [[IDXPROM4]], [[TMP2]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP3]], label %[[CONT9:.*]], label %[[HANDLER_OUT_OF_BOUNDS5:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS5]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB48:[0-9]+]], i64 [[IDXPROM4]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB48:[0-9]+]], i64 [[IDXPROM4]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT9]]: // SANITIZE-WITH-ATTR-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw i8, ptr [[PTR]], i64 344 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds nuw i64, ptr [[ARRAY]], i64 [[IDXPROM4]] @@ -2143,11 +2143,11 @@ void test32(struct annotated_with_array *ptr, int idx1, int idx2) { // SANITIZE-WITH-ATTR-SAME: ptr noundef [[PTR:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // SANITIZE-WITH-ATTR-NEXT: [[ENTRY:.*:]] // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = icmp ult i32 [[INDEX]], 43 -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT1:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT1:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i32 [[INDEX]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB49:[0-9]+]], i64 [[TMP1]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i32 [[INDEX]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB49:[0-9]+]], i64 [[TMP1]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT1]]: // SANITIZE-WITH-ATTR-NEXT: [[COUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[PTR]], i64 336 // SANITIZE-WITH-ATTR-NEXT: [[COUNTED_BY_LOAD:%.*]] = load i32, ptr [[COUNTED_BY_GEP]], align 4 @@ -2280,18 +2280,18 @@ struct multi_subscripts { // SANITIZE-WITH-ATTR-SAME: ptr noundef [[PTR:%.*]], i32 noundef [[IDX1:%.*]], i32 noundef [[IDX2:%.*]]) local_unnamed_addr #[[ATTR0]] { // SANITIZE-WITH-ATTR-NEXT: [[ENTRY:.*:]] // SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = icmp ult i32 [[IDX1]], 42 -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT1:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP0]], label %[[CONT1:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i32 [[IDX1]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB51:[0-9]+]], i64 [[TMP1]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = zext i32 [[IDX1]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB51:[0-9]+]], i64 [[TMP1]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT1]]: // SANITIZE-WITH-ATTR-NEXT: [[TMP2:%.*]] = icmp ult i32 [[IDX2]], 43 -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP2]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS2:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP2]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS2:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS2]]: // SANITIZE-WITH-ATTR-NEXT: [[TMP3:%.*]] = zext i32 [[IDX2]] to i64 -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB52:[0-9]+]], i64 [[TMP3]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB52:[0-9]+]], i64 [[TMP3]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT3]]: // SANITIZE-WITH-ATTR-NEXT: ret i64 -1 // @@ -2333,16 +2333,16 @@ size_t test34(struct multi_subscripts *ptr, int idx1, int idx2) { // SANITIZE-WITH-ATTR-NEXT: [[ENTRY:.*:]] // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 8 // SANITIZE-WITH-ATTR-NEXT: [[DOTCOUNTED_BY_LOAD:%.*]] = load i32, ptr [[DOTCOUNTED_BY_GEP]], align 4 -// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[INDEX]], [[TMP0]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF3]], !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP0:%.*]] = zext i32 [[DOTCOUNTED_BY_LOAD]] to i64, !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: [[TMP1:%.*]] = icmp ult i64 [[INDEX]], [[TMP0]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: br i1 [[TMP1]], label %[[CONT3:.*]], label %[[HANDLER_OUT_OF_BOUNDS:.*]], !prof [[PROF7]], !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[HANDLER_OUT_OF_BOUNDS]]: -// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB53:[0-9]+]], i64 [[INDEX]]) #[[ATTR8]], !nosanitize [[META2]] -// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META2]] +// SANITIZE-WITH-ATTR-NEXT: tail call void @__ubsan_handle_out_of_bounds_abort(ptr nonnull @[[GLOB53:[0-9]+]], i64 [[INDEX]]) #[[ATTR8]], !nosanitize [[META6]] +// SANITIZE-WITH-ATTR-NEXT: unreachable, !nosanitize [[META6]] // SANITIZE-WITH-ATTR: [[CONT3]]: // SANITIZE-WITH-ATTR-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 12 // SANITIZE-WITH-ATTR-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAY]], i64 [[INDEX]] -// SANITIZE-WITH-ATTR-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA4]] +// SANITIZE-WITH-ATTR-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA2]] // SANITIZE-WITH-ATTR-NEXT: ret void // // NO-SANITIZE-WITH-ATTR-LABEL: define dso_local void @test35( @@ -2470,19 +2470,19 @@ size_t test37(struct annotated *ptr) { return __builtin_dynamic_object_size((1, 2, (4, 5, (7, 8, 9, (10, ptr->array)))), 1); } //. -// SANITIZE-WITH-ATTR: [[META2]] = !{} -// SANITIZE-WITH-ATTR: [[PROF3]] = !{!"branch_weights", i32 1048575, i32 1} -// SANITIZE-WITH-ATTR: [[INT_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} -// SANITIZE-WITH-ATTR: [[META5]] = !{!"int", [[META6:![0-9]+]], i64 0} -// SANITIZE-WITH-ATTR: [[META6]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} -// SANITIZE-WITH-ATTR: [[META7]] = !{!"Simple C/C++ TBAA"} +// SANITIZE-WITH-ATTR: [[INT_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// SANITIZE-WITH-ATTR: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// SANITIZE-WITH-ATTR: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// SANITIZE-WITH-ATTR: [[META5]] = !{!"Simple C/C++ TBAA"} +// SANITIZE-WITH-ATTR: [[META6]] = !{} +// SANITIZE-WITH-ATTR: [[PROF7]] = !{!"branch_weights", i32 1048575, i32 1} // SANITIZE-WITH-ATTR: [[PROF8]] = !{!"branch_weights", i32 1, i32 1048575} -// SANITIZE-WITH-ATTR: [[CHAR_TBAA9]] = !{[[META6]], [[META6]], i64 0} +// SANITIZE-WITH-ATTR: [[CHAR_TBAA9]] = !{[[META4]], [[META4]], i64 0} // SANITIZE-WITH-ATTR: [[TBAA_STRUCT10]] = !{i64 0, i64 24, [[CHAR_TBAA9]]} // SANITIZE-WITH-ATTR: [[_ZTS10TEST13_BARPTR_TBAA11]] = !{[[META12:![0-9]+]], [[META13:![0-9]+]], i64 0} // SANITIZE-WITH-ATTR: [[META12]] = !{!"test13_foo", [[META13]], i64 0} // SANITIZE-WITH-ATTR: [[META13]] = !{!"p1 _ZTS10test13_bar", [[META14:![0-9]+]], i64 0} -// SANITIZE-WITH-ATTR: [[META14]] = !{!"any pointer", [[META6]], i64 0} +// SANITIZE-WITH-ATTR: [[META14]] = !{!"any pointer", [[META4]], i64 0} // SANITIZE-WITH-ATTR: [[_ZTS10TEST13_FOOPTR_TBAA15]] = !{[[META16:![0-9]+]], [[META16]], i64 0} // SANITIZE-WITH-ATTR: [[META16]] = !{!"p1 _ZTS10test13_foo", [[META14]], i64 0} // SANITIZE-WITH-ATTR: [[_ZTS9TESTS_FOOPTR_TBAA17]] = !{[[META18:![0-9]+]], [[META18]], i64 0} @@ -2494,7 +2494,7 @@ size_t test37(struct annotated *ptr) { // SANITIZE-WITH-ATTR: [[_ZTS9ANNOTATEDPTR_TBAA23]] = !{[[META24:![0-9]+]], [[META24]], i64 0} // SANITIZE-WITH-ATTR: [[META24]] = !{!"p1 _ZTS9annotated", [[META14]], i64 0} // SANITIZE-WITH-ATTR: [[LONG_TBAA25]] = !{[[META26:![0-9]+]], [[META26]], i64 0} -// SANITIZE-WITH-ATTR: [[META26]] = !{!"long", [[META6]], i64 0} +// SANITIZE-WITH-ATTR: [[META26]] = !{!"long", [[META4]], i64 0} //. // NO-SANITIZE-WITH-ATTR: [[INT_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} // NO-SANITIZE-WITH-ATTR: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} diff --git a/clang/test/CodeGen/builtin-maximumnum-minimumnum.c b/clang/test/CodeGen/builtin-maximumnum-minimumnum.c index ea9d2e7..aa18d9c 100644 --- a/clang/test/CodeGen/builtin-maximumnum-minimumnum.c +++ b/clang/test/CodeGen/builtin-maximumnum-minimumnum.c @@ -12,10 +12,10 @@ typedef long double ldouble2 __attribute__((ext_vector_type(2))); // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <8 x half>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <8 x half>, align 16 -// CHECK-NEXT: store <8 x half> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA2:![0-9]+]] -// CHECK-NEXT: store <8 x half> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <8 x half>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <8 x half>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] +// CHECK-NEXT: store <8 x half> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA6:![0-9]+]] +// CHECK-NEXT: store <8 x half> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <8 x half>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <8 x half>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] // CHECK-NEXT: [[ELT_MINIMUMNUM:%.*]] = call <8 x half> @llvm.minimumnum.v8f16(<8 x half> [[TMP0]], <8 x half> [[TMP1]]) // CHECK-NEXT: ret <8 x half> [[ELT_MINIMUMNUM]] // @@ -27,10 +27,10 @@ half8 pfmin16(half8 a, half8 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <8 x bfloat>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <8 x bfloat>, align 16 -// CHECK-NEXT: store <8 x bfloat> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: store <8 x bfloat> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] +// CHECK-NEXT: store <8 x bfloat> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: store <8 x bfloat> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] // CHECK-NEXT: [[ELT_MINIMUMNUM:%.*]] = call <8 x bfloat> @llvm.minimumnum.v8bf16(<8 x bfloat> [[TMP0]], <8 x bfloat> [[TMP1]]) // CHECK-NEXT: ret <8 x bfloat> [[ELT_MINIMUMNUM]] // @@ -42,10 +42,10 @@ bf16x8 pfmin16b(bf16x8 a, bf16x8 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <4 x float>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <4 x float>, align 16 -// CHECK-NEXT: store <4 x float> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: store <4 x float> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <4 x float>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <4 x float>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] +// CHECK-NEXT: store <4 x float> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: store <4 x float> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <4 x float>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <4 x float>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] // CHECK-NEXT: [[ELT_MINIMUMNUM:%.*]] = call <4 x float> @llvm.minimumnum.v4f32(<4 x float> [[TMP0]], <4 x float> [[TMP1]]) // CHECK-NEXT: ret <4 x float> [[ELT_MINIMUMNUM]] // @@ -57,10 +57,10 @@ float4 pfmin32(float4 a, float4 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <2 x double>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <2 x double>, align 16 -// CHECK-NEXT: store <2 x double> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: store <2 x double> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <2 x double>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] +// CHECK-NEXT: store <2 x double> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: store <2 x double> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <2 x double>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] // CHECK-NEXT: [[ELT_MINIMUMNUM:%.*]] = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> [[TMP0]], <2 x double> [[TMP1]]) // CHECK-NEXT: ret <2 x double> [[ELT_MINIMUMNUM]] // @@ -72,12 +72,12 @@ double2 pfmin64(double2 a, double2 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <2 x x86_fp80>, align 32 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <2 x x86_fp80>, align 32 -// CHECK-NEXT: [[A:%.*]] = load <2 x x86_fp80>, ptr [[TMP0]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[B:%.*]] = load <2 x x86_fp80>, ptr [[TMP1]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: store <2 x x86_fp80> [[A]], ptr [[A_ADDR]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: store <2 x x86_fp80> [[B]], ptr [[B_ADDR]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = load <2 x x86_fp80>, ptr [[A_ADDR]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP3:%.*]] = load <2 x x86_fp80>, ptr [[B_ADDR]], align 32, !tbaa [[TBAA2]] +// CHECK-NEXT: [[A:%.*]] = load <2 x x86_fp80>, ptr [[TMP0]], align 32, !tbaa [[TBAA6]] +// CHECK-NEXT: [[B:%.*]] = load <2 x x86_fp80>, ptr [[TMP1]], align 32, !tbaa [[TBAA6]] +// CHECK-NEXT: store <2 x x86_fp80> [[A]], ptr [[A_ADDR]], align 32, !tbaa [[TBAA6]] +// CHECK-NEXT: store <2 x x86_fp80> [[B]], ptr [[B_ADDR]], align 32, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = load <2 x x86_fp80>, ptr [[A_ADDR]], align 32, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP3:%.*]] = load <2 x x86_fp80>, ptr [[B_ADDR]], align 32, !tbaa [[TBAA6]] // CHECK-NEXT: [[ELT_MINIMUMNUM:%.*]] = call <2 x x86_fp80> @llvm.minimumnum.v2f80(<2 x x86_fp80> [[TMP2]], <2 x x86_fp80> [[TMP3]]) // CHECK-NEXT: ret <2 x x86_fp80> [[ELT_MINIMUMNUM]] // @@ -90,10 +90,10 @@ ldouble2 pfmin80(ldouble2 a, ldouble2 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <8 x half>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <8 x half>, align 16 -// CHECK-NEXT: store <8 x half> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: store <8 x half> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <8 x half>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <8 x half>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] +// CHECK-NEXT: store <8 x half> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: store <8 x half> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <8 x half>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <8 x half>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] // CHECK-NEXT: [[ELT_MAXIMUMNUM:%.*]] = call <8 x half> @llvm.maximumnum.v8f16(<8 x half> [[TMP0]], <8 x half> [[TMP1]]) // CHECK-NEXT: ret <8 x half> [[ELT_MAXIMUMNUM]] // @@ -105,10 +105,10 @@ half8 pfmax16(half8 a, half8 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <8 x bfloat>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <8 x bfloat>, align 16 -// CHECK-NEXT: store <8 x bfloat> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: store <8 x bfloat> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] +// CHECK-NEXT: store <8 x bfloat> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: store <8 x bfloat> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] // CHECK-NEXT: [[ELT_MAXIMUMNUM:%.*]] = call <8 x bfloat> @llvm.maximumnum.v8bf16(<8 x bfloat> [[TMP0]], <8 x bfloat> [[TMP1]]) // CHECK-NEXT: ret <8 x bfloat> [[ELT_MAXIMUMNUM]] // @@ -120,10 +120,10 @@ bf16x8 pfmax16b(bf16x8 a, bf16x8 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <4 x float>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <4 x float>, align 16 -// CHECK-NEXT: store <4 x float> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: store <4 x float> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <4 x float>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <4 x float>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] +// CHECK-NEXT: store <4 x float> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: store <4 x float> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <4 x float>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <4 x float>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] // CHECK-NEXT: [[ELT_MAXIMUMNUM:%.*]] = call <4 x float> @llvm.maximumnum.v4f32(<4 x float> [[TMP0]], <4 x float> [[TMP1]]) // CHECK-NEXT: ret <4 x float> [[ELT_MAXIMUMNUM]] // @@ -135,10 +135,10 @@ float4 pfmax32(float4 a, float4 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <2 x double>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <2 x double>, align 16 -// CHECK-NEXT: store <2 x double> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: store <2 x double> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <2 x double>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA2]] +// CHECK-NEXT: store <2 x double> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: store <2 x double> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[A_ADDR]], align 16, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <2 x double>, ptr [[B_ADDR]], align 16, !tbaa [[TBAA6]] // CHECK-NEXT: [[ELT_MAXIMUMNUM:%.*]] = call <2 x double> @llvm.maximumnum.v2f64(<2 x double> [[TMP0]], <2 x double> [[TMP1]]) // CHECK-NEXT: ret <2 x double> [[ELT_MAXIMUMNUM]] // @@ -151,12 +151,12 @@ double2 pfmax64(double2 a, double2 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <2 x x86_fp80>, align 32 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <2 x x86_fp80>, align 32 -// CHECK-NEXT: [[A:%.*]] = load <2 x x86_fp80>, ptr [[TMP0]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[B:%.*]] = load <2 x x86_fp80>, ptr [[TMP1]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: store <2 x x86_fp80> [[A]], ptr [[A_ADDR]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: store <2 x x86_fp80> [[B]], ptr [[B_ADDR]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = load <2 x x86_fp80>, ptr [[A_ADDR]], align 32, !tbaa [[TBAA2]] -// CHECK-NEXT: [[TMP3:%.*]] = load <2 x x86_fp80>, ptr [[B_ADDR]], align 32, !tbaa [[TBAA2]] +// CHECK-NEXT: [[A:%.*]] = load <2 x x86_fp80>, ptr [[TMP0]], align 32, !tbaa [[TBAA6]] +// CHECK-NEXT: [[B:%.*]] = load <2 x x86_fp80>, ptr [[TMP1]], align 32, !tbaa [[TBAA6]] +// CHECK-NEXT: store <2 x x86_fp80> [[A]], ptr [[A_ADDR]], align 32, !tbaa [[TBAA6]] +// CHECK-NEXT: store <2 x x86_fp80> [[B]], ptr [[B_ADDR]], align 32, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = load <2 x x86_fp80>, ptr [[A_ADDR]], align 32, !tbaa [[TBAA6]] +// CHECK-NEXT: [[TMP3:%.*]] = load <2 x x86_fp80>, ptr [[B_ADDR]], align 32, !tbaa [[TBAA6]] // CHECK-NEXT: [[ELT_MINIMUMNUM:%.*]] = call <2 x x86_fp80> @llvm.minimumnum.v2f80(<2 x x86_fp80> [[TMP2]], <2 x x86_fp80> [[TMP3]]) // CHECK-NEXT: ret <2 x x86_fp80> [[ELT_MINIMUMNUM]] // @@ -165,7 +165,7 @@ ldouble2 pfmax80(ldouble2 a, ldouble2 b) { } //. -// CHECK: [[TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"Simple C++ TBAA"} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C++ TBAA"} +// CHECK: [[TBAA6]] = !{[[META4]], [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/builtin-maxnum-minnum.c b/clang/test/CodeGen/builtin-maxnum-minnum.c index 2455f3b6..d05d43c 100644 --- a/clang/test/CodeGen/builtin-maxnum-minnum.c +++ b/clang/test/CodeGen/builtin-maxnum-minnum.c @@ -12,10 +12,10 @@ typedef long double ldouble2 __attribute__((ext_vector_type(2))); // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <8 x half>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <8 x half>, align 16 -// CHECK-NEXT: store <8 x half> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2:![0-9]+]] -// CHECK-NEXT: store <8 x half> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <8 x half>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <8 x half>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x half> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6:![0-9]+]] +// CHECK-NEXT: store <8 x half> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <8 x half>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <8 x half>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[ELT_MINNUM:%.*]] = call <8 x half> @llvm.minnum.v8f16(<8 x half> [[TMP0]], <8 x half> [[TMP1]]) // CHECK-NEXT: ret <8 x half> [[ELT_MINNUM]] // @@ -27,10 +27,10 @@ half8 pfmin16(half8 a, half8 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <8 x bfloat>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <8 x bfloat>, align 16 -// CHECK-NEXT: store <8 x bfloat> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: store <8 x bfloat> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x bfloat> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: store <8 x bfloat> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[ELT_MINNUM:%.*]] = call <8 x bfloat> @llvm.minnum.v8bf16(<8 x bfloat> [[TMP0]], <8 x bfloat> [[TMP1]]) // CHECK-NEXT: ret <8 x bfloat> [[ELT_MINNUM]] // @@ -42,10 +42,10 @@ bf16x8 pfmin16b(bf16x8 a, bf16x8 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <4 x float>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <4 x float>, align 16 -// CHECK-NEXT: store <4 x float> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: store <4 x float> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <4 x float>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <4 x float>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x float> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: store <4 x float> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <4 x float>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <4 x float>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[ELT_MINNUM:%.*]] = call <4 x float> @llvm.minnum.v4f32(<4 x float> [[TMP0]], <4 x float> [[TMP1]]) // CHECK-NEXT: ret <4 x float> [[ELT_MINNUM]] // @@ -57,10 +57,10 @@ float4 pfmin32(float4 a, float4 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <2 x double>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <2 x double>, align 16 -// CHECK-NEXT: store <2 x double> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: store <2 x double> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <2 x double>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <2 x double> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: store <2 x double> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <2 x double>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[ELT_MINNUM:%.*]] = call <2 x double> @llvm.minnum.v2f64(<2 x double> [[TMP0]], <2 x double> [[TMP1]]) // CHECK-NEXT: ret <2 x double> [[ELT_MINNUM]] // @@ -72,12 +72,12 @@ double2 pfmin64(double2 a, double2 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <2 x x86_fp80>, align 32 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <2 x x86_fp80>, align 32 -// CHECK-NEXT: [[A:%.*]] = load <2 x x86_fp80>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[B:%.*]] = load <2 x x86_fp80>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: store <2 x x86_fp80> [[A]], ptr [[A_ADDR]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: store <2 x x86_fp80> [[B]], ptr [[B_ADDR]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = load <2 x x86_fp80>, ptr [[A_ADDR]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP3:%.*]] = load <2 x x86_fp80>, ptr [[B_ADDR]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[A:%.*]] = load <2 x x86_fp80>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[B:%.*]] = load <2 x x86_fp80>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: store <2 x x86_fp80> [[A]], ptr [[A_ADDR]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: store <2 x x86_fp80> [[B]], ptr [[B_ADDR]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = load <2 x x86_fp80>, ptr [[A_ADDR]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP3:%.*]] = load <2 x x86_fp80>, ptr [[B_ADDR]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[ELT_MINNUM:%.*]] = call <2 x x86_fp80> @llvm.minnum.v2f80(<2 x x86_fp80> [[TMP2]], <2 x x86_fp80> [[TMP3]]) // CHECK-NEXT: ret <2 x x86_fp80> [[ELT_MINNUM]] // @@ -90,10 +90,10 @@ ldouble2 pfmin80(ldouble2 a, ldouble2 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <8 x half>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <8 x half>, align 16 -// CHECK-NEXT: store <8 x half> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: store <8 x half> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <8 x half>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <8 x half>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x half> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: store <8 x half> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <8 x half>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <8 x half>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[ELT_MAXNUM:%.*]] = call <8 x half> @llvm.maxnum.v8f16(<8 x half> [[TMP0]], <8 x half> [[TMP1]]) // CHECK-NEXT: ret <8 x half> [[ELT_MAXNUM]] // @@ -105,10 +105,10 @@ half8 pfmax16(half8 a, half8 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <8 x bfloat>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <8 x bfloat>, align 16 -// CHECK-NEXT: store <8 x bfloat> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: store <8 x bfloat> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <8 x bfloat> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: store <8 x bfloat> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[ELT_MAXNUM:%.*]] = call <8 x bfloat> @llvm.maxnum.v8bf16(<8 x bfloat> [[TMP0]], <8 x bfloat> [[TMP1]]) // CHECK-NEXT: ret <8 x bfloat> [[ELT_MAXNUM]] // @@ -120,10 +120,10 @@ bf16x8 pfmax16b(bf16x8 a, bf16x8 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <4 x float>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <4 x float>, align 16 -// CHECK-NEXT: store <4 x float> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: store <4 x float> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <4 x float>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <4 x float>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x float> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: store <4 x float> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <4 x float>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <4 x float>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[ELT_MAXNUM:%.*]] = call <4 x float> @llvm.maxnum.v4f32(<4 x float> [[TMP0]], <4 x float> [[TMP1]]) // CHECK-NEXT: ret <4 x float> [[ELT_MAXNUM]] // @@ -135,10 +135,10 @@ float4 pfmax32(float4 a, float4 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <2 x double>, align 16 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <2 x double>, align 16 -// CHECK-NEXT: store <2 x double> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: store <2 x double> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP1:%.*]] = load <2 x double>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <2 x double> [[A]], ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: store <2 x double> [[B]], ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[A_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP1:%.*]] = load <2 x double>, ptr [[B_ADDR]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[ELT_MAXNUM:%.*]] = call <2 x double> @llvm.maxnum.v2f64(<2 x double> [[TMP0]], <2 x double> [[TMP1]]) // CHECK-NEXT: ret <2 x double> [[ELT_MAXNUM]] // @@ -151,12 +151,12 @@ double2 pfmax64(double2 a, double2 b) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A_ADDR:%.*]] = alloca <2 x x86_fp80>, align 32 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca <2 x x86_fp80>, align 32 -// CHECK-NEXT: [[A:%.*]] = load <2 x x86_fp80>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[B:%.*]] = load <2 x x86_fp80>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: store <2 x x86_fp80> [[A]], ptr [[A_ADDR]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: store <2 x x86_fp80> [[B]], ptr [[B_ADDR]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP2:%.*]] = load <2 x x86_fp80>, ptr [[A_ADDR]], align 32, !tbaa [[CHAR_TBAA2]] -// CHECK-NEXT: [[TMP3:%.*]] = load <2 x x86_fp80>, ptr [[B_ADDR]], align 32, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[A:%.*]] = load <2 x x86_fp80>, ptr [[TMP0]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[B:%.*]] = load <2 x x86_fp80>, ptr [[TMP1]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: store <2 x x86_fp80> [[A]], ptr [[A_ADDR]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: store <2 x x86_fp80> [[B]], ptr [[B_ADDR]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP2:%.*]] = load <2 x x86_fp80>, ptr [[A_ADDR]], align 32, !tbaa [[CHAR_TBAA6]] +// CHECK-NEXT: [[TMP3:%.*]] = load <2 x x86_fp80>, ptr [[B_ADDR]], align 32, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[ELT_MINNUM:%.*]] = call <2 x x86_fp80> @llvm.minnum.v2f80(<2 x x86_fp80> [[TMP2]], <2 x x86_fp80> [[TMP3]]) // CHECK-NEXT: ret <2 x x86_fp80> [[ELT_MINNUM]] // @@ -165,7 +165,7 @@ ldouble2 pfmax80(ldouble2 a, ldouble2 b) { } //. -// CHECK: [[CHAR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"Simple C++ TBAA"} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C++ TBAA"} +// CHECK: [[CHAR_TBAA6]] = !{[[META4]], [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/call-graph-section-callback.cpp b/clang/test/CodeGen/call-graph-section-callback.cpp new file mode 100644 index 0000000..e9b0a18 --- /dev/null +++ b/clang/test/CodeGen/call-graph-section-callback.cpp @@ -0,0 +1,30 @@ +// Tests that callback function whose address is taken is attached Type ID metadata +// as it is a potential indirect call target. + +// RUN: %clang_cc1 -triple x86_64-unknown-linux -fexperimental-call-graph-section \ +// RUN: -emit-llvm -o %t %s +// RUN: FileCheck %s < %t + +//////////////////////////////////////////////////////////////////////////////// +typedef void (*CallbackFn)(int); + +// Callback function with "internal" linkage. +// CHECK-LABEL: define internal void @_ZL10myCallbacki( +// CHECK-SAME: {{.*}} !type [[F_CALLBACK:![0-9]+]] +static void myCallback(int value) +{ + volatile int sink = value; + (void)sink; +} + +int takeCallbackAddress() { + // Take the address of the callback explicitly (address-taken function) + CallbackFn cb = &myCallback; + // Store the address in a volatile pointer to keep it observable + volatile void* addr = (void*)cb; + (void)addr; + + return 0; +} + +// CHECK: [[F_CALLBACK]] = !{i64 0, !"_ZTSFviE.generalized"} diff --git a/clang/test/CodeGen/call-graph-section-templates.cpp b/clang/test/CodeGen/call-graph-section-templates.cpp new file mode 100644 index 0000000..39030d2 --- /dev/null +++ b/clang/test/CodeGen/call-graph-section-templates.cpp @@ -0,0 +1,117 @@ +// Tests that we assign appropriate identifiers to indirect calls and targets +// specifically for C++ templates. + +// RUN: %clang_cc1 -triple x86_64-unknown-linux -fexperimental-call-graph-section \ +// RUN: -emit-llvm -o %t %s +// RUN: FileCheck --check-prefix=FT %s < %t +// RUN: FileCheck --check-prefix=CST %s < %t + +//////////////////////////////////////////////////////////////////////////////// +// Class definitions and template classes (check for indirect target metadata) + +class Cls1 {}; + +// Cls2 is instantiated with T=Cls1 in foo(). Following checks are for this +// instantiation. +template <class T> +class Cls2 { +public: + // FT-LABEL: define {{.*}} void @_ZN4Cls2I4Cls1E2f1Ev( + // FT-SAME: {{.*}} !type [[F_TCLS2F1:![0-9]+]] + void f1() {} + + // FT-LABEL: define {{.*}} void @_ZN4Cls2I4Cls1E2f2ES0_( + // FT-SAME: {{.*}} !type [[F_TCLS2F2:![0-9]+]] + void f2(T a) {} + + // FT-LABEL: define {{.*}} void @_ZN4Cls2I4Cls1E2f3EPS0_( + // FT-SAME: {{.*}} !type [[F_TCLS2F3:![0-9]+]] + void f3(T *a) {} + + // FT-LABEL: define {{.*}} void @_ZN4Cls2I4Cls1E2f4EPKS0_( + // FT-SAME: {{.*}} !type [[F_TCLS2F4:![0-9]+]] + void f4(const T *a) {} + + // FT-LABEL: define {{.*}} void @_ZN4Cls2I4Cls1E2f5ERS0_( + // FT-SAME: {{.*}} !type [[F_TCLS2F5:![0-9]+]] + void f5(T &a) {} + + // FT-LABEL: define {{.*}} void @_ZN4Cls2I4Cls1E2f6ERKS0_( + // FT-SAME: {{.*}} !type [[F_TCLS2F6:![0-9]+]] + void f6(const T &a) {} + + // Mixed type function pointer member + T *(*fp)(T a, T *b, const T *c, T &d, const T &e); +}; + +// FT: [[F_TCLS2F1]] = !{i64 0, !"_ZTSFvvE.generalized"} +// FT: [[F_TCLS2F2]] = !{i64 0, !"_ZTSFv4Cls1E.generalized"} +// FT: [[F_TCLS2F3]] = !{i64 0, !"_ZTSFvP4Cls1E.generalized"} +// FT: [[F_TCLS2F4]] = !{i64 0, !"_ZTSFvPK4Cls1E.generalized"} +// FT: [[F_TCLS2F5]] = !{i64 0, !"_ZTSFvR4Cls1E.generalized"} +// FT: [[F_TCLS2F6]] = !{i64 0, !"_ZTSFvRK4Cls1E.generalized"} + +//////////////////////////////////////////////////////////////////////////////// +// Callsites (check for indirect callsite operand bundles) + +template <class T> +T *T_func(T a, T *b, const T *c, T &d, const T &e) { return b; } + +// CST-LABEL: define {{.*}} @_Z3foov +// CST-SAME: {{.*}} !type [[F_TCLS2F1:![0-9]+]] +void foo() { + // Methods for Cls2<Cls1> is checked above within the template description. + Cls2<Cls1> Obj; + + Obj.fp = T_func<Cls1>; + Cls1 Cls1Obj; + + // CST: call noundef ptr %{{.*}}, !callee_type [[F_TFUNC_CLS1_CT:![0-9]+]] + Obj.fp(Cls1Obj, &Cls1Obj, &Cls1Obj, Cls1Obj, Cls1Obj); + + // Make indirect calls to Cls2's member methods + auto fp_f1 = &Cls2<Cls1>::f1; + auto fp_f2 = &Cls2<Cls1>::f2; + auto fp_f3 = &Cls2<Cls1>::f3; + auto fp_f4 = &Cls2<Cls1>::f4; + auto fp_f5 = &Cls2<Cls1>::f5; + auto fp_f6 = &Cls2<Cls1>::f6; + + auto *Obj2Ptr = &Obj; + + // CST: call void %{{.*}}, !callee_type [[F_TCLS2F1_CT:![0-9]+]] + (Obj2Ptr->*fp_f1)(); + + // CST: call void %{{.*}}, !callee_type [[F_TCLS2F2_CT:![0-9]+]] + (Obj2Ptr->*fp_f2)(Cls1Obj); + + // CST: call void %{{.*}}, !callee_type [[F_TCLS2F3_CT:![0-9]+]] + (Obj2Ptr->*fp_f3)(&Cls1Obj); + + // CST: call void %{{.*}}, !callee_type [[F_TCLS2F4_CT:![0-9]+]] + (Obj2Ptr->*fp_f4)(&Cls1Obj); + + // CST: call void %{{.*}}, !callee_type [[F_TCLS2F5_CT:![0-9]+]] + (Obj2Ptr->*fp_f5)(Cls1Obj); + + // CST: call void %{{.*}}, !callee_type [[F_TCLS2F6_CT:![0-9]+]] + (Obj2Ptr->*fp_f6)(Cls1Obj); +} + +// CST-LABEL: define {{.*}} @_Z6T_funcI4Cls1EPT_S1_S2_PKS1_RS1_RS3_( +// CST-SAME: {{.*}} !type [[F_TFUNC_CLS1:![0-9]+]] + +// CST: [[F_TCLS2F1]] = !{i64 0, !"_ZTSFvvE.generalized"} +// CST: [[F_TFUNC_CLS1_CT]] = !{[[F_TFUNC_CLS1:![0-9]+]]} +// CST: [[F_TFUNC_CLS1]] = !{i64 0, !"_ZTSFP4Cls1S_S0_PKS_RS_RS1_E.generalized"} +// CST: [[F_TCLS2F1_CT]] = !{[[F_TCLS2F1:![0-9]+]]} +// CST: [[F_TCLS2F2_CT]] = !{[[F_TCLS2F2:![0-9]+]]} +// CST: [[F_TCLS2F2]] = !{i64 0, !"_ZTSFv4Cls1E.generalized"} +// CST: [[F_TCLS2F3_CT]] = !{[[F_TCLS2F3:![0-9]+]]} +// CST: [[F_TCLS2F3]] = !{i64 0, !"_ZTSFvP4Cls1E.generalized"} +// CST: [[F_TCLS2F4_CT]] = !{[[F_TCLS2F4:![0-9]+]]} +// CST: [[F_TCLS2F4]] = !{i64 0, !"_ZTSFvPK4Cls1E.generalized"} +// CST: [[F_TCLS2F5_CT]] = !{[[F_TCLS2F5:![0-9]+]]} +// CST: [[F_TCLS2F5]] = !{i64 0, !"_ZTSFvR4Cls1E.generalized"} +// CST: [[F_TCLS2F6_CT]] = !{[[F_TCLS2F6:![0-9]+]]} +// CST: [[F_TCLS2F6]] = !{i64 0, !"_ZTSFvRK4Cls1E.generalized"} diff --git a/clang/test/CodeGen/call-graph-section-virtual-methods.cpp b/clang/test/CodeGen/call-graph-section-virtual-methods.cpp new file mode 100644 index 0000000..afeeae1 --- /dev/null +++ b/clang/test/CodeGen/call-graph-section-virtual-methods.cpp @@ -0,0 +1,56 @@ +// Tests that we assign appropriate identifiers to indirect calls and targets +// specifically for virtual methods. + +// RUN: %clang_cc1 -triple x86_64-unknown-linux -fexperimental-call-graph-section \ +// RUN: -emit-llvm -o %t %s +// RUN: FileCheck --check-prefix=FT %s < %t +// RUN: FileCheck --check-prefix=CST %s < %t + +//////////////////////////////////////////////////////////////////////////////// +// Class definitions (check for indirect target metadata) + +class Base { + public: + // FT-LABEL: define {{.*}} @_ZN4Base2vfEPc( + // FT-SAME: {{.*}} !type [[F_TVF:![0-9]+]] + virtual int vf(char *a) { return 0; }; + }; + + class Derived : public Base { + public: + // FT: define {{.*}} @_ZN7Derived2vfEPc({{.*}} !type [[F_TVF]] + int vf(char *a) override { return 1; }; + }; + + // FT: [[F_TVF]] = !{i64 0, !"_ZTSFiPcE.generalized"} + + //////////////////////////////////////////////////////////////////////////////// + // Callsites (check for indirect callsite operand bundles) + + // CST-LABEL: define {{.*}} @_Z3foov + void foo() { + auto B = Base(); + auto D = Derived(); + + Base *Bptr = &B; + Base *BptrToD = &D; + Derived *Dptr = &D; + + auto FpBaseVf = &Base::vf; + auto FpDerivedVf = &Derived::vf; + + // CST: call noundef i32 %{{.*}}, !callee_type [[F_TVF_CT:![0-9]+]] + (Bptr->*FpBaseVf)(0); + + // CST: call noundef i32 %{{.*}}, !callee_type [[F_TVF_CT:![0-9]+]] + (BptrToD->*FpBaseVf)(0); + + // CST: call noundef i32 %{{.*}}, !callee_type [[F_TVF_CT:![0-9]+]] + (Dptr->*FpBaseVf)(0); + + // CST: call noundef i32 %{{.*}}, !callee_type [[F_TVF_CT:![0-9]+]] + (Dptr->*FpDerivedVf)(0); + } + + // CST: [[F_TVF_CT]] = !{[[F_TVF:![0-9]+]]} + // CST: [[F_TVF]] = !{i64 0, !"_ZTSFiPcE.generalized"} diff --git a/clang/test/CodeGen/call-graph-section.c b/clang/test/CodeGen/call-graph-section.c new file mode 100644 index 0000000..69cdd59 --- /dev/null +++ b/clang/test/CodeGen/call-graph-section.c @@ -0,0 +1,93 @@ +// Tests that we assign appropriate identifiers to indirect calls and targets. + +// RUN: %clang_cc1 -triple x86_64-unknown-linux -fexperimental-call-graph-section \ +// RUN: -emit-llvm -o - %s | FileCheck --check-prefixes=CHECK,ITANIUM %s + +// RUN: %clang_cc1 -triple x86_64-pc-windows-msvc -fexperimental-call-graph-section \ +// RUN: -emit-llvm -o - %s | FileCheck --check-prefixes=CHECK,MS %s + +// CHECK-LABEL: define {{(dso_local)?}} void @foo( +// CHECK-SAME: {{.*}} !type [[F_TVOID:![0-9]+]] +void foo() { +} + +// CHECK-LABEL: define {{(dso_local)?}} void @bar( +// CHECK-SAME: {{.*}} !type [[F_TVOID]] +void bar() { + void (*fp)() = foo; + // ITANIUM: call {{.*}}, !callee_type [[F_TVOID_CT:![0-9]+]] + // MS: call {{.*}}, !callee_type [[F_TVOID_CT:![0-9]+]] + fp(); +} + +// CHECK-LABEL: define {{(dso_local)?}} i32 @baz( +// CHECK-SAME: {{.*}} !type [[F_TPRIMITIVE:![0-9]+]] +int baz(char a, float b, double c) { + return 1; +} + +// CHECK-LABEL: define {{(dso_local)?}} ptr @qux( +// CHECK-SAME: {{.*}} !type [[F_TPTR:![0-9]+]] +int *qux(char *a, float *b, double *c) { + return 0; +} + +// CHECK-LABEL: define {{(dso_local)?}} void @corge( +// CHECK-SAME: {{.*}} !type [[F_TVOID]] +void corge() { + int (*fp_baz)(char, float, double) = baz; + // CHECK: call i32 {{.*}}, !callee_type [[F_TPRIMITIVE_CT:![0-9]+]] + fp_baz('a', .0f, .0); + + int *(*fp_qux)(char *, float *, double *) = qux; + // CHECK: call ptr {{.*}}, !callee_type [[F_TPTR_CT:![0-9]+]] + fp_qux(0, 0, 0); +} + +struct st1 { + int *(*fp)(char *, float *, double *); +}; + +struct st2 { + struct st1 m; +}; + +// CHECK-LABEL: define {{(dso_local)?}} void @stparam( +// CHECK-SAME: {{.*}} !type [[F_TSTRUCT:![0-9]+]] +void stparam(struct st2 a, struct st2 *b) {} + +// CHECK-LABEL: define {{(dso_local)?}} void @stf( +// CHECK-SAME: {{.*}} !type [[F_TVOID]] +void stf() { + struct st1 St1; + St1.fp = qux; + // CHECK: call ptr {{.*}}, !callee_type [[F_TPTR_CT:![0-9]+]] + St1.fp(0, 0, 0); + + struct st2 St2; + St2.m.fp = qux; + // CHECK: call ptr {{.*}}, !callee_type [[F_TPTR_CT:![0-9]+]] + St2.m.fp(0, 0, 0); + + // CHECK: call void {{.*}}, !callee_type [[F_TSTRUCT_CT:![0-9]+]] + void (*fp_stparam)(struct st2, struct st2 *) = stparam; + fp_stparam(St2, &St2); +} + +// ITANIUM: [[F_TVOID]] = !{i64 0, !"_ZTSFvE.generalized"} +// ITANIUM: [[F_TVOID_CT]] = !{[[F_TVOID:![0-9]+]]} +// ITANIUM: [[F_TPRIMITIVE]] = !{i64 0, !"_ZTSFicfdE.generalized"} +// ITANIUM: [[F_TPTR]] = !{i64 0, !"_ZTSFPiPcPfPdE.generalized"} +// ITANIUM: [[F_TPRIMITIVE_CT]] = !{[[F_TPRIMITIVE:![0-9]+]]} +// ITANIUM: [[F_TPTR_CT]] = !{[[F_TPTR:![0-9]+]]} +// ITANIUM: [[F_TSTRUCT]] = !{i64 0, !"_ZTSFv3st2PS_E.generalized"} +// ITANIUM: [[F_TSTRUCT_CT]] = !{[[F_TSTRUCT:![0-9]+]]} + +// MS: [[F_TVOID]] = !{i64 0, !"?6AX@Z.generalized"} +// MS: [[F_TVOID_CT]] = !{[[F_TVOID:![0-9]+]]} +// MS: [[F_TPRIMITIVE]] = !{i64 0, !"?6AHDMN@Z.generalized"} +// MS: [[F_TPTR]] = !{i64 0, !"?6APEAHPEADPEAMPEAN@Z.generalized"} +// MS: [[F_TPRIMITIVE_CT]] = !{[[F_TPRIMITIVE:![0-9]+]]} +// MS: [[F_TPTR_CT]] = !{[[F_TPTR:![0-9]+]]} +// MS: [[F_TSTRUCT]] = !{i64 0, !"?6AXUst2@@PEAU0@@Z.generalized"} +// MS: [[F_TSTRUCT_CT]] = !{[[F_TSTRUCT:![0-9]+]]} diff --git a/clang/test/CodeGen/call-graph-section.cpp b/clang/test/CodeGen/call-graph-section.cpp new file mode 100644 index 0000000..86ed3ee --- /dev/null +++ b/clang/test/CodeGen/call-graph-section.cpp @@ -0,0 +1,147 @@ +// Tests that we assign appropriate identifiers to indirect calls and targets +// specifically for C++ class and instance methods. + +// RUN: %clang_cc1 -triple x86_64-unknown-linux -fexperimental-call-graph-section \ +// RUN: -emit-llvm -o %t %s +// RUN: FileCheck --check-prefix=FT %s < %t +// RUN: FileCheck --check-prefix=CST %s < %t + +//////////////////////////////////////////////////////////////////////////////// +// Class definitions (check for indirect target metadata) + +class Cls1 { +public: + // FT-LABEL: define {{.*}} ptr @_ZN4Cls18receiverEPcPf( + // FT-SAME: {{.*}} !type [[F_TCLS1RECEIVER:![0-9]+]] + static int *receiver(char *a, float *b) { return 0; } +}; + +class Cls2 { +public: + int *(*fp)(char *, float *); + + // FT-LABEL: define {{.*}} i32 @_ZN4Cls22f1Ecfd( + // FT-SAME: {{.*}} !type [[F_TCLS2F1:![0-9]+]] + int f1(char a, float b, double c) { return 0; } + + // FT-LABEL: define {{.*}} ptr @_ZN4Cls22f2EPcPfPd( + // FT-SAME: {{.*}} !type [[F_TCLS2F2:![0-9]+]] + int *f2(char *a, float *b, double *c) { return 0; } + + // FT-LABEL: define {{.*}} void @_ZN4Cls22f3E4Cls1( + // FT-SAME: {{.*}} !type [[F_TCLS2F3F4:![0-9]+]] + void f3(Cls1 a) {} + + // FT-LABEL: define {{.*}} void @_ZN4Cls22f4E4Cls1( + // FT-SAME: {{.*}} !type [[F_TCLS2F3F4]] + void f4(const Cls1 a) {} + + // FT-LABEL: define {{.*}} void @_ZN4Cls22f5EP4Cls1( + // FT-SAME: {{.*}} !type [[F_TCLS2F5:![0-9]+]] + void f5(Cls1 *a) {} + + // FT-LABEL: define {{.*}} void @_ZN4Cls22f6EPK4Cls1( + // FT-SAME: {{.*}} !type [[F_TCLS2F6:![0-9]+]] + void f6(const Cls1 *a) {} + + // FT-LABEL: define {{.*}} void @_ZN4Cls22f7ER4Cls1( + // FT-SAME: {{.*}} !type [[F_TCLS2F7:![0-9]+]] + void f7(Cls1 &a) {} + + // FT-LABEL: define {{.*}} void @_ZN4Cls22f8ERK4Cls1( + // FT-SAME: {{.*}} !type [[F_TCLS2F8:![0-9]+]] + void f8(const Cls1 &a) {} + + // FT-LABEL: define {{.*}} void @_ZNK4Cls22f9Ev( + // FT-SAME: {{.*}} !type [[F_TCLS2F9:![0-9]+]] + void f9() const {} +}; + +// FT: [[F_TCLS1RECEIVER]] = !{i64 0, !"_ZTSFPiPcPfE.generalized"} +// FT: [[F_TCLS2F1]] = !{i64 0, !"_ZTSFicfdE.generalized"} +// FT: [[F_TCLS2F2]] = !{i64 0, !"_ZTSFPiPcPfPdE.generalized"} +// FT: [[F_TCLS2F3F4]] = !{i64 0, !"_ZTSFv4Cls1E.generalized"} +// FT: [[F_TCLS2F5]] = !{i64 0, !"_ZTSFvP4Cls1E.generalized"} +// FT: [[F_TCLS2F6]] = !{i64 0, !"_ZTSFvPK4Cls1E.generalized"} +// FT: [[F_TCLS2F7]] = !{i64 0, !"_ZTSFvR4Cls1E.generalized"} +// FT: [[F_TCLS2F8]] = !{i64 0, !"_ZTSFvRK4Cls1E.generalized"} +// FT: [[F_TCLS2F9]] = !{i64 0, !"_ZTSKFvvE.generalized"} + +//////////////////////////////////////////////////////////////////////////////// +// Callsites (check for indirect callsites' callee_type metadata ) + +// CST-LABEL: define {{.*}} @_Z3foov +void foo() { + Cls2 ObjCls2; + ObjCls2.fp = &Cls1::receiver; + + // CST: call noundef ptr %{{.*}}, !callee_type [[F_TCLS1RECEIVER_CT:![0-9]+]] + ObjCls2.fp(0, 0); + + auto fp_f1 = &Cls2::f1; + auto fp_f2 = &Cls2::f2; + auto fp_f3 = &Cls2::f3; + auto fp_f4 = &Cls2::f4; + auto fp_f5 = &Cls2::f5; + auto fp_f6 = &Cls2::f6; + auto fp_f7 = &Cls2::f7; + auto fp_f8 = &Cls2::f8; + auto fp_f9 = &Cls2::f9; + + Cls2 *ObjCls2Ptr = &ObjCls2; + Cls1 Cls1Param; + + // CST: call noundef i32 %{{.*}}, !callee_type [[F_TCLS2F1_CT:![0-9]+]] + (ObjCls2Ptr->*fp_f1)(0, 0, 0); + + // CST: call noundef ptr %{{.*}}, !callee_type [[F_TCLS2F2_CT:![0-9]+]] + (ObjCls2Ptr->*fp_f2)(0, 0, 0); + + // CST: call void %{{.*}}, !callee_type [[F_TCLS2F3F4_CT:![0-9]+]] + (ObjCls2Ptr->*fp_f3)(Cls1Param); + + // CST: call void %{{.*}}, !callee_type [[F_TCLS2F3F4_CT:![0-9]+]] + (ObjCls2Ptr->*fp_f4)(Cls1Param); + + // CST: call void %{{.*}}, !callee_type [[F_TCLS2F5_CT:![0-9]+]] + (ObjCls2Ptr->*fp_f5)(&Cls1Param); + + // CST: call void %{{.*}}, !callee_type [[F_TCLS2F6_CT:![0-9]+]] + (ObjCls2Ptr->*fp_f6)(&Cls1Param); + + // CST: call void %{{.*}}, !callee_type [[F_TCLS2F7_CT:![0-9]+]] + (ObjCls2Ptr->*fp_f7)(Cls1Param); + + // CST: call void %{{.*}}, !callee_type [[F_TCLS2F8_CT:![0-9]+]] + (ObjCls2Ptr->*fp_f8)(Cls1Param); + + // CST: call void %{{.*}}, !callee_type [[F_TCLS2F9_CT:![0-9]+]] + (ObjCls2Ptr->*fp_f9)(); +} + +// CST: [[F_TCLS1RECEIVER_CT]] = !{[[F_TCLS1RECEIVER:![0-9]+]]} +// CST: [[F_TCLS1RECEIVER]] = !{i64 0, !"_ZTSFPiPcPfE.generalized"} + +// CST: [[F_TCLS2F1_CT]] = !{[[F_TCLS2F1:![0-9]+]]} +// CST: [[F_TCLS2F1]] = !{i64 0, !"_ZTSFicfdE.generalized"} + +// CST: [[F_TCLS2F2_CT]] = !{[[F_TCLS2F2:![0-9]+]]} +// CST: [[F_TCLS2F2]] = !{i64 0, !"_ZTSFPiPcPfPdE.generalized"} + +// CST: [[F_TCLS2F3F4_CT]] = !{[[F_TCLS2F3F4:![0-9]+]]} +// CST: [[F_TCLS2F3F4]] = !{i64 0, !"_ZTSFv4Cls1E.generalized"} + +// CST: [[F_TCLS2F5_CT]] = !{[[F_TCLS2F5:![0-9]+]]} +// CST: [[F_TCLS2F5]] = !{i64 0, !"_ZTSFvP4Cls1E.generalized"} + +// CST: [[F_TCLS2F6_CT]] = !{[[F_TCLS2F6:![0-9]+]]} +// CST: [[F_TCLS2F6]] = !{i64 0, !"_ZTSFvPK4Cls1E.generalized"} + +// CST: [[F_TCLS2F7_CT]] = !{[[F_TCLS2F7:![0-9]+]]} +// CST: [[F_TCLS2F7]] = !{i64 0, !"_ZTSFvR4Cls1E.generalized"} + +// CST: [[F_TCLS2F8_CT]] = !{[[F_TCLS2F8:![0-9]+]]} +// CST: [[F_TCLS2F8]] = !{i64 0, !"_ZTSFvRK4Cls1E.generalized"} + +// CST: [[F_TCLS2F9_CT]] = !{[[F_TCLS2F9:![0-9]+]]} +// CST: [[F_TCLS2F9]] = !{i64 0, !"_ZTSKFvvE.generalized"} diff --git a/clang/test/CodeGen/errno-tbaa.c b/clang/test/CodeGen/errno-tbaa.c new file mode 100644 index 0000000..4ca62a3 --- /dev/null +++ b/clang/test/CodeGen/errno-tbaa.c @@ -0,0 +1,12 @@ +// RUN: %clang_cc1 -triple x86_64-unknown-unknown -emit-llvm -O0 -o - %s | FileCheck %s --check-prefix=NOTBAA +// RUN: %clang_cc1 -triple x86_64-unknown-unknown -emit-llvm -O1 -o - %s | FileCheck %s --check-prefix=ERRNO-TBAA +// RUN: %clang_cc1 -triple x86_64-unknown-unknown -x c++ -emit-llvm -O1 -o - %s | FileCheck %s --check-prefix=ERRNO-TBAA +// RUN: %clang_cc1 -triple x86_64-unknown-unknown -emit-llvm -O1 -relaxed-aliasing -o - %s | FileCheck %s --check-prefix=NOSTRICT + +// Ensure !llvm.errno.tbaa metadata is emitted upon integer accesses, if TBAA is available. + +int int_access(int *ptr) { return ptr ? *ptr : 0; } + +// NOTBAA-NOT: !llvm.errno.tbaa +// ERRNO-TBAA: !llvm.errno.tbaa +// NOSTRICT-NOT: !llvm.errno.tbaa diff --git a/clang/test/CodeGen/isfpclass.c b/clang/test/CodeGen/isfpclass.c index 8a631c4..07e760e 100644 --- a/clang/test/CodeGen/isfpclass.c +++ b/clang/test/CodeGen/isfpclass.c @@ -162,17 +162,17 @@ int4 check_isfpclass_nan_strict_v4f32(float4 x) { // CHECK-LABEL: define dso_local void @check_isfpclass_nan_v4f64( // CHECK-SAME: ptr dead_on_unwind noalias writable writeonly sret(<4 x i64>) align 16 captures(none) initializes((0, 32)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR3:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[X:%.*]] = load <4 x double>, ptr [[TMP0]], align 16, !tbaa [[CHAR_TBAA2:![0-9]+]] +// CHECK-NEXT: [[X:%.*]] = load <4 x double>, ptr [[TMP0]], align 16, !tbaa [[CHAR_TBAA6:![0-9]+]] // CHECK-NEXT: [[TMP1:%.*]] = fcmp uno <4 x double> [[X]], zeroinitializer // CHECK-NEXT: [[TMP2:%.*]] = zext <4 x i1> [[TMP1]] to <4 x i64> -// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 16, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i64> [[TMP2]], ptr [[AGG_RESULT]], align 16, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // long4 check_isfpclass_nan_v4f64(double4 x) { return __builtin_isfpclass(x, 3 /*NaN*/); } //. -// CHECK: [[CHAR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[CHAR_TBAA6]] = !{[[META4]], [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/math-libcalls-tbaa-indirect-args.c b/clang/test/CodeGen/math-libcalls-tbaa-indirect-args.c index 20a3100..e5ab6e00 100644 --- a/clang/test/CodeGen/math-libcalls-tbaa-indirect-args.c +++ b/clang/test/CodeGen/math-libcalls-tbaa-indirect-args.c @@ -54,13 +54,13 @@ long double powl(long double a, long double b); // // CHECK-MINGW32-LABEL: define dso_local void @test_powl( // CHECK-MINGW32-SAME: ptr dead_on_unwind noalias writable writeonly sret(x86_fp80) align 16 captures(none) initializes((0, 10)) [[AGG_RESULT:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]], ptr dead_on_return noundef readonly captures(none) [[TMP1:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { -// CHECK-MINGW32: [[A:%.*]] = load x86_fp80, ptr [[TMP0]], align 16, !tbaa [[LONG_DOUBLE_TBAA6:![0-9]+]] -// CHECK-MINGW32: [[B:%.*]] = load x86_fp80, ptr [[TMP1]], align 16, !tbaa [[LONG_DOUBLE_TBAA6]] -// CHECK-MINGW32: store x86_fp80 [[A]], ptr [[BYVAL_TEMP:%.*]], align 16, !tbaa [[LONG_DOUBLE_TBAA6]] -// CHECK-MINGW32: store x86_fp80 [[B]], ptr [[BYVAL_TEMP1:%.*]], align 16, !tbaa [[LONG_DOUBLE_TBAA6]] +// CHECK-MINGW32: [[A:%.*]] = load x86_fp80, ptr [[TMP0]], align 16, !tbaa [[LONG_DOUBLE_TBAA10:![0-9]+]] +// CHECK-MINGW32: [[B:%.*]] = load x86_fp80, ptr [[TMP1]], align 16, !tbaa [[LONG_DOUBLE_TBAA10]] +// CHECK-MINGW32: store x86_fp80 [[A]], ptr [[BYVAL_TEMP:%.*]], align 16, !tbaa [[LONG_DOUBLE_TBAA10]] +// CHECK-MINGW32: store x86_fp80 [[B]], ptr [[BYVAL_TEMP1:%.*]], align 16, !tbaa [[LONG_DOUBLE_TBAA10]] // CHECK-MINGW32: call void @powl(ptr dead_on_unwind nonnull writable sret(x86_fp80) align 16 [[TMP:%.*]], ptr dead_on_return noundef nonnull [[BYVAL_TEMP]], ptr dead_on_return noundef nonnull [[BYVAL_TEMP1]]) #[[ATTR3:[0-9]+]] -// CHECK-MINGW32: [[TMP2:%.*]] = load x86_fp80, ptr [[TMP]], align 16, !tbaa [[LONG_DOUBLE_TBAA6]] -// CHECK-MINGW32: store x86_fp80 [[TMP2]], ptr [[AGG_RESULT]], align 16, !tbaa [[LONG_DOUBLE_TBAA6]] +// CHECK-MINGW32: [[TMP2:%.*]] = load x86_fp80, ptr [[TMP]], align 16, !tbaa [[LONG_DOUBLE_TBAA10]] +// CHECK-MINGW32: store x86_fp80 [[TMP2]], ptr [[AGG_RESULT]], align 16, !tbaa [[LONG_DOUBLE_TBAA10]] // long double test_powl(long double a, long double b) { return powl(a, b); @@ -137,7 +137,7 @@ long double test_powl(long double a, long double b) { // CHECK-MINGW32: store x86_fp80 [[CLD_REAL]], ptr [[BYVAL_TEMP:%.*]], align 16 // CHECK-MINGW32: store x86_fp80 [[CLD_IMAG]], ptr [[BYVAL_TEMP_IMAGP:%.*]], align 16 // CHECK-MINGW32: call void @cargl(ptr dead_on_unwind nonnull writable sret(x86_fp80) align 16 [[TMP:%.*]], ptr dead_on_return noundef nonnull [[BYVAL_TEMP]]) #[[ATTR3]] -// CHECK-MINGW32: [[TMP0:%.*]] = load x86_fp80, ptr [[TMP]], align 16, !tbaa [[LONG_DOUBLE_TBAA6]] +// CHECK-MINGW32: [[TMP0:%.*]] = load x86_fp80, ptr [[TMP]], align 16, !tbaa [[LONG_DOUBLE_TBAA10]] // CHECK-MINGW32: [[CLD_REAL3:%.*]] = load x86_fp80, ptr [[CLD]], align 16 // CHECK-MINGW32: [[CLD_IMAG5:%.*]] = load x86_fp80, ptr [[CLD_IMAGP]], align 16 // CHECK-MINGW32: store x86_fp80 [[MUL_RL:%.*]], ptr [[AGG_RESULT]], align 16 @@ -190,8 +190,8 @@ int ilogbl(long double a); // // CHECK-MINGW32-LABEL: define dso_local i32 @test_ilogb( // CHECK-MINGW32-SAME: ptr dead_on_return noundef readonly captures(none) [[TMP0:%.*]]) local_unnamed_addr #[[ATTR0]] { -// CHECK-MINGW32: [[A:%.*]] = load x86_fp80, ptr [[TMP0]], align 16, !tbaa [[LONG_DOUBLE_TBAA6]] -// CHECK-MINGW32: store x86_fp80 [[A]], ptr [[BYVAL_TEMP:%.*]], align 16, !tbaa [[LONG_DOUBLE_TBAA6]] +// CHECK-MINGW32: [[A:%.*]] = load x86_fp80, ptr [[TMP0]], align 16, !tbaa [[LONG_DOUBLE_TBAA10]] +// CHECK-MINGW32: store x86_fp80 [[A]], ptr [[BYVAL_TEMP:%.*]], align 16, !tbaa [[LONG_DOUBLE_TBAA10]] // CHECK-MINGW32: [[CALL:%.*]] = call i32 @ilogbl(ptr dead_on_return noundef nonnull [[BYVAL_TEMP]]) #[[ATTR3]] // int test_ilogb(long double a) { @@ -243,8 +243,8 @@ int test_ilogb(long double a) { // CHECK-SPIR: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK-SPIR: [[META5]] = !{!"Simple C/C++ TBAA"} //. -// CHECK-MINGW32: [[LONG_DOUBLE_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK-MINGW32: [[META7]] = !{!"long double", [[META8:![0-9]+]], i64 0} -// CHECK-MINGW32: [[META8]] = !{!"omnipotent char", [[META9:![0-9]+]], i64 0} +// CHECK-MINGW32: [[META8:![0-9]+]] = !{!"omnipotent char", [[META9:![0-9]+]], i64 0} // CHECK-MINGW32: [[META9]] = !{!"Simple C/C++ TBAA"} +// CHECK-MINGW32: [[LONG_DOUBLE_TBAA10]] = !{[[META11:![0-9]+]], [[META11]], i64 0} +// CHECK-MINGW32: [[META11]] = !{!"long double", [[META8]], i64 0} //. diff --git a/clang/test/CodeGen/math-libcalls-tbaa.c b/clang/test/CodeGen/math-libcalls-tbaa.c index 53ca796..ec234bc 100644 --- a/clang/test/CodeGen/math-libcalls-tbaa.c +++ b/clang/test/CodeGen/math-libcalls-tbaa.c @@ -17,8 +17,8 @@ float crealf(float _Complex); // NONEWSTRUCTPATHTBAA-SAME: ptr noundef readonly captures(none) [[NUM:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // NONEWSTRUCTPATHTBAA-NEXT: [[ENTRY:.*:]] // NONEWSTRUCTPATHTBAA-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[NUM]], i64 40 -// NONEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[FLOAT_TBAA2:![0-9]+]] -// NONEWSTRUCTPATHTBAA-NEXT: [[CALL:%.*]] = tail call float @expf(float noundef [[TMP0]]) #[[ATTR9:[0-9]+]], !tbaa [[INT_TBAA6:![0-9]+]] +// NONEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[FLOAT_TBAA6:![0-9]+]] +// NONEWSTRUCTPATHTBAA-NEXT: [[CALL:%.*]] = tail call float @expf(float noundef [[TMP0]]) #[[ATTR9:[0-9]+]], !tbaa [[INT_TBAA2:![0-9]+]] // NONEWSTRUCTPATHTBAA-NEXT: [[MUL:%.*]] = fmul float [[TMP0]], [[CALL]] // NONEWSTRUCTPATHTBAA-NEXT: ret float [[MUL]] // @@ -26,8 +26,8 @@ float crealf(float _Complex); // NEWSTRUCTPATHTBAA-SAME: ptr noundef readonly captures(none) [[NUM:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // NEWSTRUCTPATHTBAA-NEXT: [[ENTRY:.*:]] // NEWSTRUCTPATHTBAA-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[NUM]], i64 40 -// NEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[TBAA2:![0-9]+]] -// NEWSTRUCTPATHTBAA-NEXT: [[CALL:%.*]] = tail call float @expf(float noundef [[TMP0]]) #[[ATTR9:[0-9]+]], !tbaa [[TBAA6:![0-9]+]] +// NEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[TBAA6:![0-9]+]] +// NEWSTRUCTPATHTBAA-NEXT: [[CALL:%.*]] = tail call float @expf(float noundef [[TMP0]]) #[[ATTR9:[0-9]+]], !tbaa [[TBAA2:![0-9]+]] // NEWSTRUCTPATHTBAA-NEXT: [[MUL:%.*]] = fmul float [[TMP0]], [[CALL]] // NEWSTRUCTPATHTBAA-NEXT: ret float [[MUL]] // @@ -41,8 +41,8 @@ float test_expf (float num[]) { // NONEWSTRUCTPATHTBAA-SAME: ptr noundef readonly captures(none) [[NUM:%.*]]) local_unnamed_addr #[[ATTR0]] { // NONEWSTRUCTPATHTBAA-NEXT: [[ENTRY:.*:]] // NONEWSTRUCTPATHTBAA-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[NUM]], i64 40 -// NONEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[FLOAT_TBAA2]] -// NONEWSTRUCTPATHTBAA-NEXT: [[CALL:%.*]] = tail call float @expf(float noundef [[TMP0]]) #[[ATTR9]], !tbaa [[INT_TBAA6]] +// NONEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[FLOAT_TBAA6]] +// NONEWSTRUCTPATHTBAA-NEXT: [[CALL:%.*]] = tail call float @expf(float noundef [[TMP0]]) #[[ATTR9]], !tbaa [[INT_TBAA2]] // NONEWSTRUCTPATHTBAA-NEXT: [[MUL:%.*]] = fmul float [[TMP0]], [[CALL]] // NONEWSTRUCTPATHTBAA-NEXT: ret float [[MUL]] // @@ -50,8 +50,8 @@ float test_expf (float num[]) { // NEWSTRUCTPATHTBAA-SAME: ptr noundef readonly captures(none) [[NUM:%.*]]) local_unnamed_addr #[[ATTR0]] { // NEWSTRUCTPATHTBAA-NEXT: [[ENTRY:.*:]] // NEWSTRUCTPATHTBAA-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[NUM]], i64 40 -// NEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[TBAA2]] -// NEWSTRUCTPATHTBAA-NEXT: [[CALL:%.*]] = tail call float @expf(float noundef [[TMP0]]) #[[ATTR9]], !tbaa [[TBAA6]] +// NEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[TBAA6]] +// NEWSTRUCTPATHTBAA-NEXT: [[CALL:%.*]] = tail call float @expf(float noundef [[TMP0]]) #[[ATTR9]], !tbaa [[TBAA2]] // NEWSTRUCTPATHTBAA-NEXT: [[MUL:%.*]] = fmul float [[TMP0]], [[CALL]] // NEWSTRUCTPATHTBAA-NEXT: ret float [[MUL]] // @@ -92,7 +92,7 @@ double test_fabs (double num[]) { // NONEWSTRUCTPATHTBAA-NEXT: [[ENTRY:.*:]] // NONEWSTRUCTPATHTBAA-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[NUM]], i64 80 // NONEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load double, ptr [[ARRAYIDX]], align 8, !tbaa [[DOUBLE_TBAA8]] -// NONEWSTRUCTPATHTBAA-NEXT: [[CALL:%.*]] = tail call double @remainder(double noundef [[TMP0]], double noundef [[A]]) #[[ATTR9]], !tbaa [[INT_TBAA6]] +// NONEWSTRUCTPATHTBAA-NEXT: [[CALL:%.*]] = tail call double @remainder(double noundef [[TMP0]], double noundef [[A]]) #[[ATTR9]], !tbaa [[INT_TBAA2]] // NONEWSTRUCTPATHTBAA-NEXT: [[MUL:%.*]] = fmul double [[TMP0]], [[CALL]] // NONEWSTRUCTPATHTBAA-NEXT: ret double [[MUL]] // @@ -101,7 +101,7 @@ double test_fabs (double num[]) { // NEWSTRUCTPATHTBAA-NEXT: [[ENTRY:.*:]] // NEWSTRUCTPATHTBAA-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[NUM]], i64 80 // NEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load double, ptr [[ARRAYIDX]], align 8, !tbaa [[TBAA8]] -// NEWSTRUCTPATHTBAA-NEXT: [[CALL:%.*]] = tail call double @remainder(double noundef [[TMP0]], double noundef [[A]]) #[[ATTR9]], !tbaa [[TBAA6]] +// NEWSTRUCTPATHTBAA-NEXT: [[CALL:%.*]] = tail call double @remainder(double noundef [[TMP0]], double noundef [[A]]) #[[ATTR9]], !tbaa [[TBAA2]] // NEWSTRUCTPATHTBAA-NEXT: [[MUL:%.*]] = fmul double [[TMP0]], [[CALL]] // NEWSTRUCTPATHTBAA-NEXT: ret double [[MUL]] // @@ -156,12 +156,12 @@ double test_frexp (double num[]) { // NONEWSTRUCTPATHTBAA-NEXT: call void @llvm.lifetime.start.p0(ptr nonnull [[SIN]]) #[[ATTR9]] // NONEWSTRUCTPATHTBAA-NEXT: call void @llvm.lifetime.start.p0(ptr nonnull [[COS]]) #[[ATTR9]] // NONEWSTRUCTPATHTBAA-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[NUM]], i64 8 -// NONEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[FLOAT_TBAA2]] +// NONEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[FLOAT_TBAA6]] // NONEWSTRUCTPATHTBAA-NEXT: call void @sincos(float noundef [[TMP0]], ptr noundef nonnull [[SIN]], ptr noundef nonnull [[COS]]) #[[ATTR9]] -// NONEWSTRUCTPATHTBAA-NEXT: [[TMP1:%.*]] = load float, ptr [[SIN]], align 4, !tbaa [[FLOAT_TBAA2]] -// NONEWSTRUCTPATHTBAA-NEXT: [[TMP2:%.*]] = load float, ptr [[COS]], align 4, !tbaa [[FLOAT_TBAA2]] +// NONEWSTRUCTPATHTBAA-NEXT: [[TMP1:%.*]] = load float, ptr [[SIN]], align 4, !tbaa [[FLOAT_TBAA6]] +// NONEWSTRUCTPATHTBAA-NEXT: [[TMP2:%.*]] = load float, ptr [[COS]], align 4, !tbaa [[FLOAT_TBAA6]] // NONEWSTRUCTPATHTBAA-NEXT: [[MUL:%.*]] = fmul float [[TMP1]], [[TMP2]] -// NONEWSTRUCTPATHTBAA-NEXT: [[TMP3:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[FLOAT_TBAA2]] +// NONEWSTRUCTPATHTBAA-NEXT: [[TMP3:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[FLOAT_TBAA6]] // NONEWSTRUCTPATHTBAA-NEXT: [[ADD:%.*]] = fadd float [[MUL]], [[TMP3]] // NONEWSTRUCTPATHTBAA-NEXT: call void @llvm.lifetime.end.p0(ptr nonnull [[COS]]) #[[ATTR9]] // NONEWSTRUCTPATHTBAA-NEXT: call void @llvm.lifetime.end.p0(ptr nonnull [[SIN]]) #[[ATTR9]] @@ -175,12 +175,12 @@ double test_frexp (double num[]) { // NEWSTRUCTPATHTBAA-NEXT: call void @llvm.lifetime.start.p0(ptr nonnull [[SIN]]) #[[ATTR9]] // NEWSTRUCTPATHTBAA-NEXT: call void @llvm.lifetime.start.p0(ptr nonnull [[COS]]) #[[ATTR9]] // NEWSTRUCTPATHTBAA-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[NUM]], i64 8 -// NEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[TBAA2]] +// NEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[TBAA6]] // NEWSTRUCTPATHTBAA-NEXT: call void @sincos(float noundef [[TMP0]], ptr noundef nonnull [[SIN]], ptr noundef nonnull [[COS]]) #[[ATTR9]] -// NEWSTRUCTPATHTBAA-NEXT: [[TMP1:%.*]] = load float, ptr [[SIN]], align 4, !tbaa [[TBAA2]] -// NEWSTRUCTPATHTBAA-NEXT: [[TMP2:%.*]] = load float, ptr [[COS]], align 4, !tbaa [[TBAA2]] +// NEWSTRUCTPATHTBAA-NEXT: [[TMP1:%.*]] = load float, ptr [[SIN]], align 4, !tbaa [[TBAA6]] +// NEWSTRUCTPATHTBAA-NEXT: [[TMP2:%.*]] = load float, ptr [[COS]], align 4, !tbaa [[TBAA6]] // NEWSTRUCTPATHTBAA-NEXT: [[MUL:%.*]] = fmul float [[TMP1]], [[TMP2]] -// NEWSTRUCTPATHTBAA-NEXT: [[TMP3:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[TBAA2]] +// NEWSTRUCTPATHTBAA-NEXT: [[TMP3:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[TBAA6]] // NEWSTRUCTPATHTBAA-NEXT: [[ADD:%.*]] = fadd float [[MUL]], [[TMP3]] // NEWSTRUCTPATHTBAA-NEXT: call void @llvm.lifetime.end.p0(ptr nonnull [[COS]]) #[[ATTR9]] // NEWSTRUCTPATHTBAA-NEXT: call void @llvm.lifetime.end.p0(ptr nonnull [[SIN]]) #[[ATTR9]] @@ -198,12 +198,12 @@ float test_sincos (float num[]) { // NONEWSTRUCTPATHTBAA-SAME: ptr noundef readonly captures(none) [[NUM:%.*]]) local_unnamed_addr #[[ATTR7]] { // NONEWSTRUCTPATHTBAA-NEXT: [[ENTRY:.*:]] // NONEWSTRUCTPATHTBAA-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[NUM]], i64 8 -// NONEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[FLOAT_TBAA2]] +// NONEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[FLOAT_TBAA6]] // NONEWSTRUCTPATHTBAA-NEXT: [[DOTFCA_0_INSERT:%.*]] = insertvalue [2 x float] poison, float [[TMP0]], 0 // NONEWSTRUCTPATHTBAA-NEXT: [[DOTFCA_1_INSERT:%.*]] = insertvalue [2 x float] [[DOTFCA_0_INSERT]], float 0.000000e+00, 1 // NONEWSTRUCTPATHTBAA-NEXT: [[CALL:%.*]] = tail call { float, float } @cacoshf([2 x float] noundef alignstack(8) [[DOTFCA_1_INSERT]]) #[[ATTR9]] // NONEWSTRUCTPATHTBAA-NEXT: [[TMP1:%.*]] = extractvalue { float, float } [[CALL]], 0 -// NONEWSTRUCTPATHTBAA-NEXT: [[TMP2:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[FLOAT_TBAA2]] +// NONEWSTRUCTPATHTBAA-NEXT: [[TMP2:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[FLOAT_TBAA6]] // NONEWSTRUCTPATHTBAA-NEXT: [[ADD:%.*]] = fadd float [[TMP1]], [[TMP2]] // NONEWSTRUCTPATHTBAA-NEXT: ret float [[ADD]] // @@ -211,12 +211,12 @@ float test_sincos (float num[]) { // NEWSTRUCTPATHTBAA-SAME: ptr noundef readonly captures(none) [[NUM:%.*]]) local_unnamed_addr #[[ATTR7]] { // NEWSTRUCTPATHTBAA-NEXT: [[ENTRY:.*:]] // NEWSTRUCTPATHTBAA-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[NUM]], i64 8 -// NEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[TBAA2]] +// NEWSTRUCTPATHTBAA-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[TBAA6]] // NEWSTRUCTPATHTBAA-NEXT: [[DOTFCA_0_INSERT:%.*]] = insertvalue [2 x float] poison, float [[TMP0]], 0 // NEWSTRUCTPATHTBAA-NEXT: [[DOTFCA_1_INSERT:%.*]] = insertvalue [2 x float] [[DOTFCA_0_INSERT]], float 0.000000e+00, 1 // NEWSTRUCTPATHTBAA-NEXT: [[CALL:%.*]] = tail call { float, float } @cacoshf([2 x float] noundef alignstack(8) [[DOTFCA_1_INSERT]]) #[[ATTR9]] // NEWSTRUCTPATHTBAA-NEXT: [[TMP1:%.*]] = extractvalue { float, float } [[CALL]], 0 -// NEWSTRUCTPATHTBAA-NEXT: [[TMP2:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[TBAA2]] +// NEWSTRUCTPATHTBAA-NEXT: [[TMP2:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !tbaa [[TBAA6]] // NEWSTRUCTPATHTBAA-NEXT: [[ADD:%.*]] = fadd float [[TMP1]], [[TMP2]] // NEWSTRUCTPATHTBAA-NEXT: ret float [[ADD]] // @@ -227,21 +227,21 @@ float test_cacoshf (float num[]) { } //. -// NONEWSTRUCTPATHTBAA: [[FLOAT_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// NONEWSTRUCTPATHTBAA: [[META3]] = !{!"float", [[META4:![0-9]+]], i64 0} +// NONEWSTRUCTPATHTBAA: [[INT_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// NONEWSTRUCTPATHTBAA: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} // NONEWSTRUCTPATHTBAA: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // NONEWSTRUCTPATHTBAA: [[META5]] = !{!"Simple C/C++ TBAA"} -// NONEWSTRUCTPATHTBAA: [[INT_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// NONEWSTRUCTPATHTBAA: [[META7]] = !{!"int", [[META4]], i64 0} +// NONEWSTRUCTPATHTBAA: [[FLOAT_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// NONEWSTRUCTPATHTBAA: [[META7]] = !{!"float", [[META4]], i64 0} // NONEWSTRUCTPATHTBAA: [[DOUBLE_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} // NONEWSTRUCTPATHTBAA: [[META9]] = !{!"double", [[META4]], i64 0} //. // NEWSTRUCTPATHTBAA: [[TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0, i64 4} -// NEWSTRUCTPATHTBAA: [[META3]] = !{[[META4:![0-9]+]], i64 4, !"float"} +// NEWSTRUCTPATHTBAA: [[META3]] = !{[[META4:![0-9]+]], i64 4, !"int"} // NEWSTRUCTPATHTBAA: [[META4]] = !{[[META5:![0-9]+]], i64 1, !"omnipotent char"} // NEWSTRUCTPATHTBAA: [[META5]] = !{!"Simple C/C++ TBAA"} // NEWSTRUCTPATHTBAA: [[TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0, i64 4} -// NEWSTRUCTPATHTBAA: [[META7]] = !{[[META4]], i64 4, !"int"} +// NEWSTRUCTPATHTBAA: [[META7]] = !{[[META4]], i64 4, !"float"} // NEWSTRUCTPATHTBAA: [[TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0, i64 8} // NEWSTRUCTPATHTBAA: [[META9]] = !{[[META4]], i64 8, !"double"} //. diff --git a/clang/test/CodeGen/pointer-arithmetic-align.c b/clang/test/CodeGen/pointer-arithmetic-align.c index 745ab84..73b1c1f 100644 --- a/clang/test/CodeGen/pointer-arithmetic-align.c +++ b/clang/test/CodeGen/pointer-arithmetic-align.c @@ -13,7 +13,7 @@ struct a { // CHECK-SAME: ptr noundef writeonly captures(none) initializes((8, 9)) [[CTX:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[BLOCK:%.*]] = getelementptr inbounds nuw i8, ptr [[CTX]], i64 8 -// CHECK-NEXT: store i8 0, ptr [[BLOCK]], align 8, !tbaa [[TBAA2:![0-9]+]] +// CHECK-NEXT: store i8 0, ptr [[BLOCK]], align 8, !tbaa [[TBAA6:![0-9]+]] // CHECK-NEXT: ret void // void ptradd_0(struct a *ctx) { @@ -24,7 +24,7 @@ void ptradd_0(struct a *ctx) { // CHECK-SAME: ptr noundef writeonly captures(none) initializes((12, 13)) [[CTX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds nuw i8, ptr [[CTX]], i64 12 -// CHECK-NEXT: store i8 0, ptr [[ADD_PTR]], align 4, !tbaa [[TBAA2]] +// CHECK-NEXT: store i8 0, ptr [[ADD_PTR]], align 4, !tbaa [[TBAA6]] // CHECK-NEXT: ret void // void ptradd_4(struct a *ctx) { @@ -35,7 +35,7 @@ void ptradd_4(struct a *ctx) { // CHECK-SAME: ptr noundef writeonly captures(none) initializes((16, 17)) [[CTX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds nuw i8, ptr [[CTX]], i64 16 -// CHECK-NEXT: store i8 0, ptr [[ADD_PTR]], align 8, !tbaa [[TBAA2]] +// CHECK-NEXT: store i8 0, ptr [[ADD_PTR]], align 8, !tbaa [[TBAA6]] // CHECK-NEXT: ret void // void ptradd_8(struct a *ctx) { @@ -46,7 +46,7 @@ void ptradd_8(struct a *ctx) { // CHECK-SAME: ptr noundef writeonly captures(none) initializes((16, 17)) [[CTX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds nuw i8, ptr [[CTX]], i64 16 -// CHECK-NEXT: store i8 0, ptr [[ADD_PTR]], align 8, !tbaa [[TBAA2]] +// CHECK-NEXT: store i8 0, ptr [[ADD_PTR]], align 8, !tbaa [[TBAA6]] // CHECK-NEXT: ret void // void ptradd_8_commuted(struct a *ctx) { @@ -57,7 +57,7 @@ void ptradd_8_commuted(struct a *ctx) { // CHECK-SAME: ptr noundef writeonly captures(none) initializes((8, 9)) [[CTX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds nuw i8, ptr [[CTX]], i64 8 -// CHECK-NEXT: store i8 0, ptr [[ADD_PTR]], align 4, !tbaa [[TBAA2]] +// CHECK-NEXT: store i8 0, ptr [[ADD_PTR]], align 4, !tbaa [[TBAA6]] // CHECK-NEXT: ret void // void ptrsub_4(struct a *ctx) { @@ -70,14 +70,14 @@ void ptrsub_4(struct a *ctx) { // CHECK-NEXT: [[BLOCK:%.*]] = getelementptr inbounds nuw i8, ptr [[CTX]], i64 8 // CHECK-NEXT: [[IDX_EXT:%.*]] = zext i8 [[IDX]] to i64 // CHECK-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds nuw i8, ptr [[BLOCK]], i64 [[IDX_EXT]] -// CHECK-NEXT: store i8 0, ptr [[ADD_PTR]], align 1, !tbaa [[TBAA2]] +// CHECK-NEXT: store i8 0, ptr [[ADD_PTR]], align 1, !tbaa [[TBAA6]] // CHECK-NEXT: ret void // void neg_ptradd_var_index(struct a *ctx, uint8_t idx) { *(ctx->block + idx) = 0; } //. -// CHECK: [[TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[TBAA6]] = !{[[META4]], [[META4]], i64 0} //. diff --git a/clang/test/CodeGen/sanitize-metadata-ignorelist.c b/clang/test/CodeGen/sanitize-metadata-ignorelist.c index 4dc8c0c..dac69e9 100644 --- a/clang/test/CodeGen/sanitize-metadata-ignorelist.c +++ b/clang/test/CodeGen/sanitize-metadata-ignorelist.c @@ -7,9 +7,9 @@ int y; // ALLOW-LABEL: define {{[^@]+}}@foo -// ALLOW-SAME: () local_unnamed_addr #[[ATTR0:[0-9]+]] !pcsections !2 { +// ALLOW-SAME: () local_unnamed_addr #[[ATTR0:[0-9]+]] !pcsections !6 { // ALLOW-NEXT: entry: -// ALLOW-NEXT: [[TMP0:%.*]] = atomicrmw add ptr @y, i32 1 monotonic, align 4, !pcsections !4 +// ALLOW-NEXT: [[TMP0:%.*]] = atomicrmw add ptr @y, i32 1 monotonic, align 4, !pcsections !8 // ALLOW-NEXT: ret void // // FUN-LABEL: define {{[^@]+}}@foo @@ -29,15 +29,15 @@ void foo() { } // ALLOW-LABEL: define {{[^@]+}}@bar -// ALLOW-SAME: () local_unnamed_addr #[[ATTR0]] !pcsections !2 { +// ALLOW-SAME: () local_unnamed_addr #[[ATTR0]] !pcsections !6 { // ALLOW-NEXT: entry: -// ALLOW-NEXT: [[TMP0:%.*]] = atomicrmw add ptr @y, i32 2 monotonic, align 4, !pcsections !4 +// ALLOW-NEXT: [[TMP0:%.*]] = atomicrmw add ptr @y, i32 2 monotonic, align 4, !pcsections !8 // ALLOW-NEXT: ret void // // FUN-LABEL: define {{[^@]+}}@bar -// FUN-SAME: () local_unnamed_addr #[[ATTR0]] !pcsections !2 { +// FUN-SAME: () local_unnamed_addr #[[ATTR0]] !pcsections !6 { // FUN-NEXT: entry: -// FUN-NEXT: [[TMP0:%.*]] = atomicrmw add ptr @y, i32 2 monotonic, align 4, !pcsections !4 +// FUN-NEXT: [[TMP0:%.*]] = atomicrmw add ptr @y, i32 2 monotonic, align 4, !pcsections !8 // FUN-NEXT: ret void // // SRC-LABEL: define {{[^@]+}}@bar diff --git a/clang/test/CodeGen/sanitize-metadata-nosanitize.c b/clang/test/CodeGen/sanitize-metadata-nosanitize.c index 22ed25b..f2672d7 100644 --- a/clang/test/CodeGen/sanitize-metadata-nosanitize.c +++ b/clang/test/CodeGen/sanitize-metadata-nosanitize.c @@ -12,7 +12,7 @@ //. // CHECK: Function Attrs: mustprogress nofree noinline norecurse nosync nounwind willreturn memory(write, argmem: none, inaccessiblemem: none) // CHECK-LABEL: define dso_local void @escape( -// CHECK-SAME: ptr noundef [[P:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] !pcsections [[META2:![0-9]+]] { +// CHECK-SAME: ptr noundef [[P:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] !pcsections [[META6:![0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: ret void // @@ -23,13 +23,13 @@ __attribute__((noinline, not_tail_called)) void escape(const volatile void *p) { // CHECK: Function Attrs: mustprogress nofree norecurse nounwind willreturn memory(write, argmem: readwrite, inaccessiblemem: none) // CHECK-LABEL: define dso_local i32 @normal_function( -// CHECK-SAME: ptr noundef [[X:%.*]], ptr noundef readonly captures(none) [[Y:%.*]]) local_unnamed_addr #[[ATTR1:[0-9]+]] !pcsections [[META4:![0-9]+]] { +// CHECK-SAME: ptr noundef [[X:%.*]], ptr noundef readonly captures(none) [[Y:%.*]]) local_unnamed_addr #[[ATTR1:[0-9]+]] !pcsections [[META8:![0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8 -// CHECK-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8, !tbaa [[INTPTR_TBAA6:![0-9]+]] -// CHECK-NEXT: store atomic i32 1, ptr [[X]] monotonic, align 4, !pcsections [[META11:![0-9]+]] +// CHECK-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8, !tbaa [[INTPTR_TBAA10:![0-9]+]] +// CHECK-NEXT: store atomic i32 1, ptr [[X]] monotonic, align 4, !pcsections [[META13:![0-9]+]] // CHECK-NEXT: notail call void @escape(ptr noundef nonnull [[X_ADDR]]) -// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[Y]], align 4, !tbaa [[INT_TBAA12:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[Y]], align 4, !tbaa [[INT_TBAA2:![0-9]+]] // CHECK-NEXT: ret i32 [[TMP0]] // int normal_function(int *x, int *y) { @@ -43,10 +43,10 @@ int normal_function(int *x, int *y) { // CHECK-SAME: ptr noundef [[X:%.*]], ptr noundef readonly captures(none) [[Y:%.*]]) local_unnamed_addr #[[ATTR2:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8 -// CHECK-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8, !tbaa [[INTPTR_TBAA6]] +// CHECK-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8, !tbaa [[INTPTR_TBAA10]] // CHECK-NEXT: store atomic i32 1, ptr [[X]] monotonic, align 4 // CHECK-NEXT: notail call void @escape(ptr noundef nonnull [[X_ADDR]]) -// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[Y]], align 4, !tbaa [[INT_TBAA12]] +// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[Y]], align 4, !tbaa [[INT_TBAA2]] // CHECK-NEXT: ret i32 [[TMP0]] // __attribute__((disable_sanitizer_instrumentation)) int test_disable_sanitize_instrumentation(int *x, int *y) { @@ -60,10 +60,10 @@ __attribute__((disable_sanitizer_instrumentation)) int test_disable_sanitize_ins // CHECK-SAME: ptr noundef [[X:%.*]], ptr noundef readonly captures(none) [[Y:%.*]]) local_unnamed_addr #[[ATTR3:[0-9]+]] !pcsections [[META14:![0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8 -// CHECK-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8, !tbaa [[INTPTR_TBAA6]] -// CHECK-NEXT: store atomic i32 1, ptr [[X]] monotonic, align 4, !pcsections [[META11]] +// CHECK-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8, !tbaa [[INTPTR_TBAA10]] +// CHECK-NEXT: store atomic i32 1, ptr [[X]] monotonic, align 4, !pcsections [[META13]] // CHECK-NEXT: notail call void @escape(ptr noundef nonnull [[X_ADDR]]) -// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[Y]], align 4, !tbaa [[INT_TBAA12]] +// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[Y]], align 4, !tbaa [[INT_TBAA2]] // CHECK-NEXT: ret i32 [[TMP0]] // __attribute__((no_sanitize("thread"))) int test_no_sanitize_thread(int *x, int *y) { @@ -77,10 +77,10 @@ __attribute__((no_sanitize("thread"))) int test_no_sanitize_thread(int *x, int * // CHECK-SAME: ptr noundef [[X:%.*]], ptr noundef readonly captures(none) [[Y:%.*]]) local_unnamed_addr #[[ATTR3]] !pcsections [[META14]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8 -// CHECK-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8, !tbaa [[INTPTR_TBAA6]] -// CHECK-NEXT: store atomic i32 1, ptr [[X]] monotonic, align 4, !pcsections [[META11]] +// CHECK-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8, !tbaa [[INTPTR_TBAA10]] +// CHECK-NEXT: store atomic i32 1, ptr [[X]] monotonic, align 4, !pcsections [[META13]] // CHECK-NEXT: notail call void @escape(ptr noundef nonnull [[X_ADDR]]) -// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[Y]], align 4, !tbaa [[INT_TBAA12]] +// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[Y]], align 4, !tbaa [[INT_TBAA2]] // CHECK-NEXT: ret i32 [[TMP0]] // __attribute__((no_sanitize("all"))) int test_no_sanitize_all(int *x, int *y) { @@ -97,18 +97,18 @@ __attribute__((no_sanitize("all"))) int test_no_sanitize_all(int *x, int *y) { //. // CHECK: [[META0:![0-9]+]] = !{i32 1, !"wchar_size", i32 4} // CHECK: [[META1:![0-9]+]] = !{!"{{.*}}clang version {{.*}}"} -// CHECK: [[META2]] = !{!"sanmd_covered2!C", [[META3:![0-9]+]]} -// CHECK: [[META3]] = !{i64 0} -// CHECK: [[META4]] = !{!"sanmd_covered2!C", [[META5:![0-9]+]]} -// CHECK: [[META5]] = !{i64 3} -// CHECK: [[INTPTR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK: [[META7]] = !{!"p1 int", [[META8:![0-9]+]], i64 0} -// CHECK: [[META8]] = !{!"any pointer", [[META9:![0-9]+]], i64 0} -// CHECK: [[META9]] = !{!"omnipotent char", [[META10:![0-9]+]], i64 0} -// CHECK: [[META10]] = !{!"Simple C/C++ TBAA"} -// CHECK: [[META11]] = !{!"sanmd_atomics2!C"} -// CHECK: [[INT_TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} -// CHECK: [[META13]] = !{!"int", [[META9]], i64 0} +// CHECK: [[INT_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// CHECK: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META6]] = !{!"sanmd_covered2!C", [[META7:![0-9]+]]} +// CHECK: [[META7]] = !{i64 0} +// CHECK: [[META8]] = !{!"sanmd_covered2!C", [[META9:![0-9]+]]} +// CHECK: [[META9]] = !{i64 3} +// CHECK: [[INTPTR_TBAA10]] = !{[[META11:![0-9]+]], [[META11]], i64 0} +// CHECK: [[META11]] = !{!"p1 int", [[META12:![0-9]+]], i64 0} +// CHECK: [[META12]] = !{!"any pointer", [[META4]], i64 0} +// CHECK: [[META13]] = !{!"sanmd_atomics2!C"} // CHECK: [[META14]] = !{!"sanmd_covered2!C", [[META15:![0-9]+]]} // CHECK: [[META15]] = !{i64 2} //. diff --git a/clang/test/CodeGen/scoped-atomic-ops.c b/clang/test/CodeGen/scoped-atomic-ops.c index 545a6c9..c390481 100644 --- a/clang/test/CodeGen/scoped-atomic-ops.c +++ b/clang/test/CodeGen/scoped-atomic-ops.c @@ -1,113 +1,772 @@ // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 // RUN: %clang_cc1 %s -emit-llvm -o - -triple=amdgcn-amd-amdhsa -ffreestanding \ -// RUN: -fvisibility=hidden | FileCheck --check-prefix=AMDGCN %s +// RUN: -fvisibility=hidden | FileCheck --check-prefixes=AMDGCN,AMDGCN_CL_DEF %s // RUN: %clang_cc1 %s -emit-llvm -o - -triple=amdgcn-amd-amdhsa -ffreestanding \ -// RUN: -cl-std=CL2.0 -fvisibility=hidden | FileCheck --check-prefix=AMDGCN %s +// RUN: -cl-std=CL2.0 -fvisibility=hidden | FileCheck --check-prefixes=AMDGCN,AMDGCN_CL_20 %s // RUN: %clang_cc1 %s -emit-llvm -o - -triple=spirv64-unknown-unknown -ffreestanding \ // RUN: -fvisibility=hidden | FileCheck --check-prefix=SPIRV %s -// AMDGCN-LABEL: define hidden i32 @fi1a( -// AMDGCN: [[TMP0:%.*]] = load atomic i32, ptr [[PTR0:.+]] monotonic, align 4 -// AMDGCN: [[TMP1:%.*]] = load atomic i32, ptr [[PTR1:.+]] syncscope("agent") monotonic, align 4 -// AMDGCN: [[TMP2:%.*]] = load atomic i32, ptr [[PTR2:.+]] syncscope("workgroup") monotonic, align 4 -// AMDGCN: [[TMP3:%.*]] = load atomic i32, ptr [[PTR3:.+]] syncscope("wavefront") monotonic, align 4 -// AMDGCN: [[TMP4:%.*]] = load atomic i32, ptr [[PTR4:.+]] syncscope("singlethread") monotonic, align 4 -// SPIRV: define hidden spir_func i32 @fi1a( -// SPIRV: [[TMP0:%.*]] = load atomic i32, ptr [[PTR0:.+]] monotonic, align 4 -// SPIRV: [[TMP1:%.*]] = load atomic i32, ptr [[PTR1:.+]] syncscope("device") monotonic, align 4 -// SPIRV: [[TMP2:%.*]] = load atomic i32, ptr [[PTR2:.+]] syncscope("workgroup") monotonic, align 4 -// SPIRV: [[TMP3:%.*]] = load atomic i32, ptr [[PTR3:.+]] syncscope("subgroup") monotonic, align 4 -// SPIRV: [[TMP4:%.*]] = load atomic i32, ptr [[PTR4:.+]] syncscope("singlethread") monotonic, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden i32 @fi1a( +// AMDGCN_CL_DEF-SAME: ptr noundef [[I:%.*]]) #[[ATTR0:[0-9]+]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[V:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[V_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load atomic i32, ptr [[TMP0]] monotonic, align 4 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP1]], ptr [[V_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = load atomic i32, ptr [[TMP2]] syncscope("agent") monotonic, align 4 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP3]], ptr [[V_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = load atomic i32, ptr [[TMP4]] syncscope("workgroup") monotonic, align 4 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP5]], ptr [[V_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP7:%.*]] = load atomic i32, ptr [[TMP6]] syncscope("cluster") monotonic, align 4 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP7]], ptr [[V_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP8:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP9:%.*]] = load atomic i32, ptr [[TMP8]] syncscope("wavefront") monotonic, align 4 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP9]], ptr [[V_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP10:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP11:%.*]] = load atomic i32, ptr [[TMP10]] syncscope("singlethread") monotonic, align 4 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP11]], ptr [[V_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP12:%.*]] = load i32, ptr [[V_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: ret i32 [[TMP12]] +// +// AMDGCN_CL_20-LABEL: define hidden i32 @fi1a( +// AMDGCN_CL_20-SAME: ptr noundef [[I:%.*]]) #[[ATTR0:[0-9]+]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[V:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load atomic i32, ptr [[TMP0]] monotonic, align 4 +// AMDGCN_CL_20-NEXT: store i32 [[TMP1]], ptr addrspace(5) [[V]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load atomic i32, ptr [[TMP2]] syncscope("agent") monotonic, align 4 +// AMDGCN_CL_20-NEXT: store i32 [[TMP3]], ptr addrspace(5) [[V]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = load atomic i32, ptr [[TMP4]] syncscope("workgroup") monotonic, align 4 +// AMDGCN_CL_20-NEXT: store i32 [[TMP5]], ptr addrspace(5) [[V]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP7:%.*]] = load atomic i32, ptr [[TMP6]] syncscope("cluster") monotonic, align 4 +// AMDGCN_CL_20-NEXT: store i32 [[TMP7]], ptr addrspace(5) [[V]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP8:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP9:%.*]] = load atomic i32, ptr [[TMP8]] syncscope("wavefront") monotonic, align 4 +// AMDGCN_CL_20-NEXT: store i32 [[TMP9]], ptr addrspace(5) [[V]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP10:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP11:%.*]] = load atomic i32, ptr [[TMP10]] syncscope("singlethread") monotonic, align 4 +// AMDGCN_CL_20-NEXT: store i32 [[TMP11]], ptr addrspace(5) [[V]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP12:%.*]] = load i32, ptr addrspace(5) [[V]], align 4 +// AMDGCN_CL_20-NEXT: ret i32 [[TMP12]] +// +// SPIRV-LABEL: define hidden spir_func i32 @fi1a( +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0:[0-9]+]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[V:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load atomic i32, ptr [[TMP0]] monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP1]], ptr [[V]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP3:%.*]] = load atomic i32, ptr [[TMP2]] syncscope("device") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP3]], ptr [[V]], align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP5:%.*]] = load atomic i32, ptr [[TMP4]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP5]], ptr [[V]], align 4 +// SPIRV-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP7:%.*]] = load atomic i32, ptr [[TMP6]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP7]], ptr [[V]], align 4 +// SPIRV-NEXT: [[TMP8:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP9:%.*]] = load atomic i32, ptr [[TMP8]] syncscope("subgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP9]], ptr [[V]], align 4 +// SPIRV-NEXT: [[TMP10:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP11:%.*]] = load atomic i32, ptr [[TMP10]] syncscope("singlethread") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP11]], ptr [[V]], align 4 +// SPIRV-NEXT: [[TMP12:%.*]] = load i32, ptr [[V]], align 4 +// SPIRV-NEXT: ret i32 [[TMP12]] +// int fi1a(int *i) { int v; __scoped_atomic_load(i, &v, __ATOMIC_RELAXED, __MEMORY_SCOPE_SYSTEM); __scoped_atomic_load(i, &v, __ATOMIC_RELAXED, __MEMORY_SCOPE_DEVICE); __scoped_atomic_load(i, &v, __ATOMIC_RELAXED, __MEMORY_SCOPE_WRKGRP); + __scoped_atomic_load(i, &v, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR); __scoped_atomic_load(i, &v, __ATOMIC_RELAXED, __MEMORY_SCOPE_WVFRNT); __scoped_atomic_load(i, &v, __ATOMIC_RELAXED, __MEMORY_SCOPE_SINGLE); return v; } // AMDGCN-LABEL: define hidden i32 @fi1b( -// AMDGCN: [[TMP0:%.*]] = load atomic i32, ptr [[PTR0:%.+]] monotonic, align 4 -// AMDGCN: [[TMP1:%.*]] = load atomic i32, ptr [[PTR1:%.+]] syncscope("agent") monotonic, align 4 -// AMDGCN: [[TMP2:%.*]] = load atomic i32, ptr [[PTR2:%.+]] syncscope("workgroup") monotonic, align 4 -// AMDGCN: [[TMP3:%.*]] = load atomic i32, ptr [[PTR3:%.+]] syncscope("wavefront") monotonic, align 4 -// AMDGCN: [[TMP4:%.*]] = load atomic i32, ptr [[PTR4:%.+]] syncscope("singlethread") monotonic, align 4 +// AMDGCN-SAME: ptr noundef [[I:%.*]]) #[[ATTR0:[0-9]+]] { +// AMDGCN-NEXT: [[ENTRY:.*:]] +// AMDGCN-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN-NEXT: [[ATOMIC_TEMP1:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN-NEXT: [[ATOMIC_TEMP3:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN-NEXT: [[ATOMIC_TEMP5:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN-NEXT: [[ATOMIC_TEMP1_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP1]] to ptr +// AMDGCN-NEXT: [[ATOMIC_TEMP2_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP2]] to ptr +// AMDGCN-NEXT: [[ATOMIC_TEMP3_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP3]] to ptr +// AMDGCN-NEXT: [[ATOMIC_TEMP4_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP4]] to ptr +// AMDGCN-NEXT: [[ATOMIC_TEMP5_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP5]] to ptr +// AMDGCN-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: [[TMP1:%.*]] = load atomic i32, ptr [[TMP0]] monotonic, align 4 +// AMDGCN-NEXT: store i32 [[TMP1]], ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP2:%.*]] = load i32, ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP3:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: store i32 [[TMP2]], ptr [[TMP3]], align 4 +// AMDGCN-NEXT: [[TMP4:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: [[TMP5:%.*]] = load atomic i32, ptr [[TMP4]] syncscope("agent") monotonic, align 4 +// AMDGCN-NEXT: store i32 [[TMP5]], ptr [[ATOMIC_TEMP1_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP6:%.*]] = load i32, ptr [[ATOMIC_TEMP1_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP7:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: store i32 [[TMP6]], ptr [[TMP7]], align 4 +// AMDGCN-NEXT: [[TMP8:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: [[TMP9:%.*]] = load atomic i32, ptr [[TMP8]] syncscope("workgroup") monotonic, align 4 +// AMDGCN-NEXT: store i32 [[TMP9]], ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP10:%.*]] = load i32, ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP11:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: store i32 [[TMP10]], ptr [[TMP11]], align 4 +// AMDGCN-NEXT: [[TMP12:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: [[TMP13:%.*]] = load atomic i32, ptr [[TMP12]] syncscope("cluster") monotonic, align 4 +// AMDGCN-NEXT: store i32 [[TMP13]], ptr [[ATOMIC_TEMP3_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP14:%.*]] = load i32, ptr [[ATOMIC_TEMP3_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP15:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: store i32 [[TMP14]], ptr [[TMP15]], align 4 +// AMDGCN-NEXT: [[TMP16:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: [[TMP17:%.*]] = load atomic i32, ptr [[TMP16]] syncscope("wavefront") monotonic, align 4 +// AMDGCN-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP19:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// AMDGCN-NEXT: [[TMP20:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: [[TMP21:%.*]] = load atomic i32, ptr [[TMP20]] syncscope("singlethread") monotonic, align 4 +// AMDGCN-NEXT: store i32 [[TMP21]], ptr [[ATOMIC_TEMP5_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP22:%.*]] = load i32, ptr [[ATOMIC_TEMP5_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP23:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: store i32 [[TMP22]], ptr [[TMP23]], align 4 +// AMDGCN-NEXT: [[TMP24:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: [[TMP25:%.*]] = load i32, ptr [[TMP24]], align 4 +// AMDGCN-NEXT: ret i32 [[TMP25]] +// // SPIRV-LABEL: define hidden spir_func i32 @fi1b( -// SPIRV: [[TMP0:%.*]] = load atomic i32, ptr [[PTR0:%.+]] monotonic, align 4 -// SPIRV: [[TMP1:%.*]] = load atomic i32, ptr [[PTR1:%.+]] syncscope("device") monotonic, align 4 -// SPIRV: [[TMP2:%.*]] = load atomic i32, ptr [[PTR2:%.+]] syncscope("workgroup") monotonic, align 4 -// SPIRV: [[TMP3:%.*]] = load atomic i32, ptr [[PTR3:%.+]] syncscope("subgroup") monotonic, align 4 -// SPIRV: [[TMP4:%.*]] = load atomic i32, ptr [[PTR4:%.+]] syncscope("singlethread") monotonic, align 4 +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP1:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP3:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP5:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load atomic i32, ptr [[TMP0]] monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP1]], ptr [[ATOMIC_TEMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[ATOMIC_TEMP]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP2]], ptr [[TMP3]], align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP5:%.*]] = load atomic i32, ptr [[TMP4]] syncscope("device") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP5]], ptr [[ATOMIC_TEMP1]], align 4 +// SPIRV-NEXT: [[TMP6:%.*]] = load i32, ptr [[ATOMIC_TEMP1]], align 4 +// SPIRV-NEXT: [[TMP7:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP6]], ptr [[TMP7]], align 4 +// SPIRV-NEXT: [[TMP8:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP9:%.*]] = load atomic i32, ptr [[TMP8]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP9]], ptr [[ATOMIC_TEMP2]], align 4 +// SPIRV-NEXT: [[TMP10:%.*]] = load i32, ptr [[ATOMIC_TEMP2]], align 4 +// SPIRV-NEXT: [[TMP11:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP10]], ptr [[TMP11]], align 4 +// SPIRV-NEXT: [[TMP12:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP13:%.*]] = load atomic i32, ptr [[TMP12]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP13]], ptr [[ATOMIC_TEMP3]], align 4 +// SPIRV-NEXT: [[TMP14:%.*]] = load i32, ptr [[ATOMIC_TEMP3]], align 4 +// SPIRV-NEXT: [[TMP15:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP14]], ptr [[TMP15]], align 4 +// SPIRV-NEXT: [[TMP16:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP17:%.*]] = load atomic i32, ptr [[TMP16]] syncscope("subgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP4]], align 4 +// SPIRV-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP4]], align 4 +// SPIRV-NEXT: [[TMP19:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// SPIRV-NEXT: [[TMP20:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP21:%.*]] = load atomic i32, ptr [[TMP20]] syncscope("singlethread") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP21]], ptr [[ATOMIC_TEMP5]], align 4 +// SPIRV-NEXT: [[TMP22:%.*]] = load i32, ptr [[ATOMIC_TEMP5]], align 4 +// SPIRV-NEXT: [[TMP23:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP22]], ptr [[TMP23]], align 4 +// SPIRV-NEXT: [[TMP24:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP25:%.*]] = load i32, ptr [[TMP24]], align 4 +// SPIRV-NEXT: ret i32 [[TMP25]] +// int fi1b(int *i) { *i = __scoped_atomic_load_n(i, __ATOMIC_RELAXED, __MEMORY_SCOPE_SYSTEM); *i = __scoped_atomic_load_n(i, __ATOMIC_RELAXED, __MEMORY_SCOPE_DEVICE); *i = __scoped_atomic_load_n(i, __ATOMIC_RELAXED, __MEMORY_SCOPE_WRKGRP); + *i = __scoped_atomic_load_n(i, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR); *i = __scoped_atomic_load_n(i, __ATOMIC_RELAXED, __MEMORY_SCOPE_WVFRNT); *i = __scoped_atomic_load_n(i, __ATOMIC_RELAXED, __MEMORY_SCOPE_SINGLE); return *i; } -// AMDGCN-LABEL: define hidden void @fi2a( -// AMDGCN: store atomic i32 [[TMP0:%.+]], ptr [[PTR0:%.+]] monotonic, align 4 -// AMDGCN: store atomic i32 [[TMP1:%.+]], ptr [[PTR1:%.+]] syncscope("agent") monotonic, align 4 -// AMDGCN: store atomic i32 [[TMP2:%.+]], ptr [[PTR2:%.+]] syncscope("workgroup") monotonic, align 4 -// AMDGCN: store atomic i32 [[TMP3:%.+]], ptr [[PTR3:%.+]] syncscope("wavefront") monotonic, align 4 -// AMDGCN: store atomic i32 [[TMP4:%.+]], ptr [[PTR4:%.+]] syncscope("singlethread") monotonic, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden void @fi2a( +// AMDGCN_CL_DEF-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[V:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[V_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[V_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[V_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: store atomic i32 [[TMP1]], ptr [[TMP0]] monotonic, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = load i32, ptr [[V_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: store atomic i32 [[TMP3]], ptr [[TMP2]] syncscope("agent") monotonic, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = load i32, ptr [[V_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: store atomic i32 [[TMP5]], ptr [[TMP4]] syncscope("workgroup") monotonic, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP7:%.*]] = load i32, ptr [[V_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: store atomic i32 [[TMP7]], ptr [[TMP6]] syncscope("cluster") monotonic, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP8:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP9:%.*]] = load i32, ptr [[V_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: store atomic i32 [[TMP9]], ptr [[TMP8]] syncscope("wavefront") monotonic, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP10:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP11:%.*]] = load i32, ptr [[V_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: store atomic i32 [[TMP11]], ptr [[TMP10]] syncscope("singlethread") monotonic, align 4 +// AMDGCN_CL_DEF-NEXT: ret void +// +// AMDGCN_CL_20-LABEL: define hidden void @fi2a( +// AMDGCN_CL_20-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[V:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr addrspace(5) [[V]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = addrspacecast ptr addrspace(5) [[V]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_20-NEXT: store atomic i32 [[TMP2]], ptr [[TMP0]] monotonic, align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = addrspacecast ptr addrspace(5) [[V]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4 +// AMDGCN_CL_20-NEXT: store atomic i32 [[TMP5]], ptr [[TMP3]] syncscope("agent") monotonic, align 4 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP7:%.*]] = addrspacecast ptr addrspace(5) [[V]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 +// AMDGCN_CL_20-NEXT: store atomic i32 [[TMP8]], ptr [[TMP6]] syncscope("workgroup") monotonic, align 4 +// AMDGCN_CL_20-NEXT: [[TMP9:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP10:%.*]] = addrspacecast ptr addrspace(5) [[V]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4 +// AMDGCN_CL_20-NEXT: store atomic i32 [[TMP11]], ptr [[TMP9]] syncscope("cluster") monotonic, align 4 +// AMDGCN_CL_20-NEXT: [[TMP12:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP13:%.*]] = addrspacecast ptr addrspace(5) [[V]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4 +// AMDGCN_CL_20-NEXT: store atomic i32 [[TMP14]], ptr [[TMP12]] syncscope("wavefront") monotonic, align 4 +// AMDGCN_CL_20-NEXT: [[TMP15:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP16:%.*]] = addrspacecast ptr addrspace(5) [[V]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4 +// AMDGCN_CL_20-NEXT: store atomic i32 [[TMP17]], ptr [[TMP15]] syncscope("singlethread") monotonic, align 4 +// AMDGCN_CL_20-NEXT: ret void +// // SPIRV-LABEL: define hidden spir_func void @fi2a( -// SPIRV: store atomic i32 [[TMP0:%.+]], ptr [[PTR0:%.+]] monotonic, align 4 -// SPIRV: store atomic i32 [[TMP1:%.+]], ptr [[PTR1:%.+]] syncscope("device") monotonic, align 4 -// SPIRV: store atomic i32 [[TMP2:%.+]], ptr [[PTR2:%.+]] syncscope("workgroup") monotonic, align 4 -// SPIRV: store atomic i32 [[TMP3:%.+]], ptr [[PTR3:%.+]] syncscope("subgroup") monotonic, align 4 -// SPIRV: store atomic i32 [[TMP4:%.+]], ptr [[PTR4:%.+]] syncscope("singlethread") monotonic, align 4 +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[V:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[V]], align 4 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[V]], align 4 +// SPIRV-NEXT: store atomic i32 [[TMP1]], ptr [[TMP0]] monotonic, align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP3:%.*]] = load i32, ptr [[V]], align 4 +// SPIRV-NEXT: store atomic i32 [[TMP3]], ptr [[TMP2]] syncscope("device") monotonic, align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP5:%.*]] = load i32, ptr [[V]], align 4 +// SPIRV-NEXT: store atomic i32 [[TMP5]], ptr [[TMP4]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP7:%.*]] = load i32, ptr [[V]], align 4 +// SPIRV-NEXT: store atomic i32 [[TMP7]], ptr [[TMP6]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: [[TMP8:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP9:%.*]] = load i32, ptr [[V]], align 4 +// SPIRV-NEXT: store atomic i32 [[TMP9]], ptr [[TMP8]] syncscope("subgroup") monotonic, align 4 +// SPIRV-NEXT: [[TMP10:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP11:%.*]] = load i32, ptr [[V]], align 4 +// SPIRV-NEXT: store atomic i32 [[TMP11]], ptr [[TMP10]] syncscope("singlethread") monotonic, align 4 +// SPIRV-NEXT: ret void +// void fi2a(int *i) { int v = 1; __scoped_atomic_store(i, &v, __ATOMIC_RELAXED, __MEMORY_SCOPE_SYSTEM); __scoped_atomic_store(i, &v, __ATOMIC_RELAXED, __MEMORY_SCOPE_DEVICE); __scoped_atomic_store(i, &v, __ATOMIC_RELAXED, __MEMORY_SCOPE_WRKGRP); + __scoped_atomic_store(i, &v, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR); __scoped_atomic_store(i, &v, __ATOMIC_RELAXED, __MEMORY_SCOPE_WVFRNT); __scoped_atomic_store(i, &v, __ATOMIC_RELAXED, __MEMORY_SCOPE_SINGLE); } // AMDGCN-LABEL: define hidden void @fi2b( -// AMDGCN: store atomic i32 [[TMP0:%.+]], ptr [[PTR0:%.+]] monotonic, align 4 -// AMDGCN: store atomic i32 [[TMP1:%.+]], ptr [[PTR1:%.+]] syncscope("agent") monotonic, align 4 -// AMDGCN: store atomic i32 [[TMP2:%.+]], ptr [[PTR2:%.+]] syncscope("workgroup") monotonic, align 4 -// AMDGCN: store atomic i32 [[TMP3:%.+]], ptr [[PTR3:%.+]] syncscope("wavefront") monotonic, align 4 -// AMDGCN: store atomic i32 [[TMP4:%.+]], ptr [[PTR4:%.+]] syncscope("singlethread") monotonic, align 4 +// AMDGCN-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN-NEXT: [[ENTRY:.*:]] +// AMDGCN-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN-NEXT: [[DOTATOMICTMP2:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN-NEXT: [[DOTATOMICTMP4:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN-NEXT: [[DOTATOMICTMP1_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP1]] to ptr +// AMDGCN-NEXT: [[DOTATOMICTMP2_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP2]] to ptr +// AMDGCN-NEXT: [[DOTATOMICTMP3_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP3]] to ptr +// AMDGCN-NEXT: [[DOTATOMICTMP4_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP4]] to ptr +// AMDGCN-NEXT: [[DOTATOMICTMP5_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP5]] to ptr +// AMDGCN-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN-NEXT: store atomic i32 [[TMP1]], ptr [[TMP0]] monotonic, align 4 +// AMDGCN-NEXT: [[TMP2:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: store i32 1, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN-NEXT: store atomic i32 [[TMP3]], ptr [[TMP2]] syncscope("agent") monotonic, align 4 +// AMDGCN-NEXT: [[TMP4:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: store i32 1, ptr [[DOTATOMICTMP2_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTATOMICTMP2_ASCAST]], align 4 +// AMDGCN-NEXT: store atomic i32 [[TMP5]], ptr [[TMP4]] syncscope("workgroup") monotonic, align 4 +// AMDGCN-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: store i32 1, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN-NEXT: store atomic i32 [[TMP7]], ptr [[TMP6]] syncscope("cluster") monotonic, align 4 +// AMDGCN-NEXT: [[TMP8:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: store i32 1, ptr [[DOTATOMICTMP4_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTATOMICTMP4_ASCAST]], align 4 +// AMDGCN-NEXT: store atomic i32 [[TMP9]], ptr [[TMP8]] syncscope("wavefront") monotonic, align 4 +// AMDGCN-NEXT: [[TMP10:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN-NEXT: store i32 1, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN-NEXT: store atomic i32 [[TMP11]], ptr [[TMP10]] syncscope("singlethread") monotonic, align 4 +// AMDGCN-NEXT: ret void +// // SPIRV-LABEL: define hidden spir_func void @fi2b( -// SPIRV: store atomic i32 [[TMP0:%.+]], ptr [[PTR0:%.+]] monotonic, align 4 -// SPIRV: store atomic i32 [[TMP1:%.+]], ptr [[PTR1:%.+]] syncscope("device") monotonic, align 4 -// SPIRV: store atomic i32 [[TMP2:%.+]], ptr [[PTR2:%.+]] syncscope("workgroup") monotonic, align 4 -// SPIRV: store atomic i32 [[TMP3:%.+]], ptr [[PTR3:%.+]] syncscope("subgroup") monotonic, align 4 -// SPIRV: store atomic i32 [[TMP4:%.+]], ptr [[PTR4:%.+]] syncscope("singlethread") monotonic, align 4 +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP2:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP4:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: store atomic i32 [[TMP1]], ptr [[TMP0]] monotonic, align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP1]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTATOMICTMP1]], align 4 +// SPIRV-NEXT: store atomic i32 [[TMP3]], ptr [[TMP2]] syncscope("device") monotonic, align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP2]], align 4 +// SPIRV-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTATOMICTMP2]], align 4 +// SPIRV-NEXT: store atomic i32 [[TMP5]], ptr [[TMP4]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP3]], align 4 +// SPIRV-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTATOMICTMP3]], align 4 +// SPIRV-NEXT: store atomic i32 [[TMP7]], ptr [[TMP6]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: [[TMP8:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP4]], align 4 +// SPIRV-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTATOMICTMP4]], align 4 +// SPIRV-NEXT: store atomic i32 [[TMP9]], ptr [[TMP8]] syncscope("subgroup") monotonic, align 4 +// SPIRV-NEXT: [[TMP10:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP5]], align 4 +// SPIRV-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP5]], align 4 +// SPIRV-NEXT: store atomic i32 [[TMP11]], ptr [[TMP10]] syncscope("singlethread") monotonic, align 4 +// SPIRV-NEXT: ret void +// void fi2b(int *i) { __scoped_atomic_store_n(i, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_SYSTEM); __scoped_atomic_store_n(i, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_DEVICE); __scoped_atomic_store_n(i, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_WRKGRP); + __scoped_atomic_store_n(i, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR); __scoped_atomic_store_n(i, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_WVFRNT); __scoped_atomic_store_n(i, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_SINGLE); } -// AMDGCN-LABEL: define hidden void @fi3a( -// AMDGCN: [[TMP0:%.*]] = atomicrmw add ptr [[PTR0:%.+]], i32 [[VAL0:.+]] monotonic, align 4 -// AMDGCN: [[TMP1:%.*]] = atomicrmw sub ptr [[PTR1:%.+]], i32 [[VAL1:.+]] monotonic, align 4 -// AMDGCN: [[TMP2:%.*]] = atomicrmw and ptr [[PTR2:%.+]], i32 [[VAL2:.+]] monotonic, align 4 -// AMDGCN: [[TMP3:%.*]] = atomicrmw or ptr [[PTR3:%.+]], i32 [[VAL3:.+]] monotonic, align 4 -// AMDGCN: [[TMP4:%.*]] = atomicrmw xor ptr [[PTR4:%.+]], i32 [[VAL4:.+]] monotonic, align 4 -// AMDGCN: [[TMP5:%.*]] = atomicrmw nand ptr [[PTR5:%.+]], i32 [[VAL5:.+]] monotonic, align 4 -// AMDGCN: [[TMP6:%.*]] = atomicrmw min ptr [[PTR6:%.+]], i32 [[VAL6:.+]] monotonic, align 4 -// AMDGCN: [[TMP7:%.*]] = atomicrmw max ptr [[PTR7:%.+]], i32 [[VAL7:.+]] monotonic, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden void @fi3a( +// AMDGCN_CL_DEF-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[A_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[A_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[B_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[B_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[E_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[E_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[F_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[G_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[G_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[H_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[H_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP1_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP1]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP2_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP2]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP3_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP3]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP4_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP4]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP5_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP5]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP6_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP6]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP7_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP7]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP8_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP8]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP9_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP9]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP10_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP10]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP11_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP11]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP12_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP12]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP13_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP13]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP14_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP14]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[A]], ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[B]], ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[E]], ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[F]], ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[G]], ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[H]], ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3:![0-9]+]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// AMDGCN_CL_DEF-NEXT: ret void +// +// AMDGCN_CL_20-LABEL: define hidden void @fi3a( +// AMDGCN_CL_20-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[A_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[A_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[B_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[B_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[E_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[E_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[F_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[G_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[G_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[H_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[H_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP1_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP1]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP2_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP2]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP3_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP3]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP4_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP4]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP5_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP5]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP6_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP6]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP7_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP7]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP8_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP8]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP9_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP9]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP10_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP10]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP11_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP11]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP12_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP12]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP13_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP13]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP14_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP14]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[A]], ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[B]], ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[E]], ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[F]], ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[G]], ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[H]], ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4:![0-9]+]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// AMDGCN_CL_20-NEXT: ret void +// // SPIRV-LABEL: define hidden spir_func void @fi3a( -// SPIRV: [[TMP0:%.*]] = atomicrmw add ptr [[PTR0:%.+]], i32 [[VAL0:.+]] monotonic, align 4 -// SPIRV: [[TMP1:%.*]] = atomicrmw sub ptr [[PTR1:%.+]], i32 [[VAL1:.+]] monotonic, align 4 -// SPIRV: [[TMP2:%.*]] = atomicrmw and ptr [[PTR2:%.+]], i32 [[VAL2:.+]] monotonic, align 4 -// SPIRV: [[TMP3:%.*]] = atomicrmw or ptr [[PTR3:%.+]], i32 [[VAL3:.+]] monotonic, align 4 -// SPIRV: [[TMP4:%.*]] = atomicrmw xor ptr [[PTR4:%.+]], i32 [[VAL4:.+]] monotonic, align 4 -// SPIRV: [[TMP5:%.*]] = atomicrmw nand ptr [[PTR5:%.+]], i32 [[VAL5:.+]] monotonic, align 4 -// SPIRV: [[TMP6:%.*]] = atomicrmw min ptr [[PTR6:%.+]], i32 [[VAL6:.+]] monotonic, align 4 -// SPIRV: [[TMP7:%.*]] = atomicrmw max ptr [[PTR7:%.+]], i32 [[VAL7:.+]] monotonic, align 4 +// SPIRV-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[E]], ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[F]], ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[G]], ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[H]], ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP]], align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// SPIRV-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP1]], align 4 +// SPIRV-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1]], align 4 +// SPIRV-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2]], align 4 +// SPIRV-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2]], align 4 +// SPIRV-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// SPIRV-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP3]], align 4 +// SPIRV-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3]], align 4 +// SPIRV-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4]], align 4 +// SPIRV-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4]], align 4 +// SPIRV-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// SPIRV-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP5]], align 4 +// SPIRV-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5]], align 4 +// SPIRV-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6]], align 4 +// SPIRV-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6]], align 4 +// SPIRV-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// SPIRV-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP7]], align 4 +// SPIRV-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7]], align 4 +// SPIRV-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8]], align 4 +// SPIRV-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8]], align 4 +// SPIRV-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// SPIRV-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP9]], align 4 +// SPIRV-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9]], align 4 +// SPIRV-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10]], align 4 +// SPIRV-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10]], align 4 +// SPIRV-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// SPIRV-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP11]], align 4 +// SPIRV-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11]], align 4 +// SPIRV-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12]], align 4 +// SPIRV-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12]], align 4 +// SPIRV-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// SPIRV-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP13]], align 4 +// SPIRV-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13]], align 4 +// SPIRV-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14]], align 4 +// SPIRV-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14]], align 4 +// SPIRV-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// SPIRV-NEXT: ret void +// void fi3a(int *a, int *b, int *c, int *d, int *e, int *f, int *g, int *h) { *a = __scoped_atomic_fetch_add(a, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_SYSTEM); *b = __scoped_atomic_fetch_sub(b, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_SYSTEM); @@ -119,24 +778,357 @@ void fi3a(int *a, int *b, int *c, int *d, int *e, int *f, int *g, int *h) { *h = __scoped_atomic_fetch_max(h, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_SYSTEM); } -// AMDGCN-LABEL: define hidden void @fi3b( -// AMDGCN: [[TMP0:%.*]] = atomicrmw add ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("agent") monotonic, align 4 -// AMDGCN: [[TMP1:%.*]] = atomicrmw sub ptr [[PTR1:%.+]], i32 [[VAL1:.+]] syncscope("agent") monotonic, align 4 -// AMDGCN: [[TMP2:%.*]] = atomicrmw and ptr [[PTR2:%.+]], i32 [[VAL2:.+]] syncscope("agent") monotonic, align 4 -// AMDGCN: [[TMP3:%.*]] = atomicrmw or ptr [[PTR3:%.+]], i32 [[VAL3:.+]] syncscope("agent") monotonic, align 4 -// AMDGCN: [[TMP4:%.*]] = atomicrmw xor ptr [[PTR4:%.+]], i32 [[VAL4:.+]] syncscope("agent") monotonic, align 4 -// AMDGCN: [[TMP5:%.*]] = atomicrmw nand ptr [[PTR5:%.+]], i32 [[VAL5:.+]] syncscope("agent") monotonic, align 4 -// AMDGCN: [[TMP6:%.*]] = atomicrmw min ptr [[PTR6:%.+]], i32 [[VAL6:.+]] syncscope("agent") monotonic, align 4 -// AMDGCN: [[TMP7:%.*]] = atomicrmw max ptr [[PTR7:%.+]], i32 [[VAL7:.+]] syncscope("agent") monotonic, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden void @fi3b( +// AMDGCN_CL_DEF-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[A_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[A_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[B_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[B_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[E_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[E_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[F_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[G_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[G_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[H_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[H_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP1_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP1]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP2_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP2]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP3_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP3]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP4_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP4]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP5_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP5]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP6_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP6]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP7_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP7]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP8_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP8]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP9_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP9]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP10_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP10]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP11_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP11]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP12_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP12]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP13_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP13]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP14_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP14]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[A]], ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[B]], ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[E]], ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[F]], ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[G]], ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[H]], ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// AMDGCN_CL_DEF-NEXT: ret void +// +// AMDGCN_CL_20-LABEL: define hidden void @fi3b( +// AMDGCN_CL_20-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[A_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[A_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[B_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[B_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[E_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[E_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[F_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[G_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[G_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[H_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[H_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP1_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP1]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP2_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP2]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP3_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP3]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP4_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP4]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP5_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP5]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP6_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP6]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP7_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP7]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP8_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP8]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP9_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP9]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP10_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP10]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP11_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP11]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP12_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP12]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP13_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP13]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP14_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP14]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[A]], ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[B]], ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[E]], ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[F]], ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[G]], ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[H]], ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// AMDGCN_CL_20-NEXT: ret void +// // SPIRV-LABEL: define hidden spir_func void @fi3b( -// SPIRV: [[TMP0:%.*]] = atomicrmw add ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("device") monotonic, align 4 -// SPIRV: [[TMP1:%.*]] = atomicrmw sub ptr [[PTR1:%.+]], i32 [[VAL1:.+]] syncscope("device") monotonic, align 4 -// SPIRV: [[TMP2:%.*]] = atomicrmw and ptr [[PTR2:%.+]], i32 [[VAL2:.+]] syncscope("device") monotonic, align 4 -// SPIRV: [[TMP3:%.*]] = atomicrmw or ptr [[PTR3:%.+]], i32 [[VAL3:.+]] syncscope("device") monotonic, align 4 -// SPIRV: [[TMP4:%.*]] = atomicrmw xor ptr [[PTR4:%.+]], i32 [[VAL4:.+]] syncscope("device") monotonic, align 4 -// SPIRV: [[TMP5:%.*]] = atomicrmw nand ptr [[PTR5:%.+]], i32 [[VAL5:.+]] syncscope("device") monotonic, align 4 -// SPIRV: [[TMP6:%.*]] = atomicrmw min ptr [[PTR6:%.+]], i32 [[VAL6:.+]] syncscope("device") monotonic, align 4 -// SPIRV: [[TMP7:%.*]] = atomicrmw max ptr [[PTR7:%.+]], i32 [[VAL7:.+]] syncscope("device") monotonic, align 4 +// SPIRV-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[E]], ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[F]], ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[G]], ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[H]], ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("device") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP]], align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// SPIRV-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP1]], align 4 +// SPIRV-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1]], align 4 +// SPIRV-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("device") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2]], align 4 +// SPIRV-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2]], align 4 +// SPIRV-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// SPIRV-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP3]], align 4 +// SPIRV-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3]], align 4 +// SPIRV-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("device") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4]], align 4 +// SPIRV-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4]], align 4 +// SPIRV-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// SPIRV-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP5]], align 4 +// SPIRV-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5]], align 4 +// SPIRV-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("device") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6]], align 4 +// SPIRV-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6]], align 4 +// SPIRV-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// SPIRV-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP7]], align 4 +// SPIRV-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7]], align 4 +// SPIRV-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("device") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8]], align 4 +// SPIRV-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8]], align 4 +// SPIRV-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// SPIRV-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP9]], align 4 +// SPIRV-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9]], align 4 +// SPIRV-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("device") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10]], align 4 +// SPIRV-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10]], align 4 +// SPIRV-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// SPIRV-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP11]], align 4 +// SPIRV-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11]], align 4 +// SPIRV-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("device") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12]], align 4 +// SPIRV-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12]], align 4 +// SPIRV-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// SPIRV-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP13]], align 4 +// SPIRV-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13]], align 4 +// SPIRV-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("device") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14]], align 4 +// SPIRV-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14]], align 4 +// SPIRV-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// SPIRV-NEXT: ret void +// void fi3b(int *a, int *b, int *c, int *d, int *e, int *f, int *g, int *h) { *a = __scoped_atomic_fetch_add(a, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_DEVICE); *b = __scoped_atomic_fetch_sub(b, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_DEVICE); @@ -148,24 +1140,357 @@ void fi3b(int *a, int *b, int *c, int *d, int *e, int *f, int *g, int *h) { *h = __scoped_atomic_fetch_max(h, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_DEVICE); } -// AMDGCN-LABEL: define hidden void @fi3c( -// AMDGCN: [[TMP0:%.*]] = atomicrmw add ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("workgroup") monotonic, align 4 -// AMDGCN: [[TMP1:%.*]] = atomicrmw sub ptr [[PTR1:%.+]], i32 [[VAL1:.+]] syncscope("workgroup") monotonic, align 4 -// AMDGCN: [[TMP2:%.*]] = atomicrmw and ptr [[PTR2:%.+]], i32 [[VAL2:.+]] syncscope("workgroup") monotonic, align 4 -// AMDGCN: [[TMP3:%.*]] = atomicrmw or ptr [[PTR3:%.+]], i32 [[VAL3:.+]] syncscope("workgroup") monotonic, align 4 -// AMDGCN: [[TMP4:%.*]] = atomicrmw xor ptr [[PTR4:%.+]], i32 [[VAL4:.+]] syncscope("workgroup") monotonic, align 4 -// AMDGCN: [[TMP5:%.*]] = atomicrmw nand ptr [[PTR5:%.+]], i32 [[VAL5:.+]] syncscope("workgroup") monotonic, align 4 -// AMDGCN: [[TMP6:%.*]] = atomicrmw min ptr [[PTR6:%.+]], i32 [[VAL6:.+]] syncscope("workgroup") monotonic, align 4 -// AMDGCN: [[TMP7:%.*]] = atomicrmw max ptr [[PTR7:%.+]], i32 [[VAL7:.+]] syncscope("workgroup") monotonic, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden void @fi3c( +// AMDGCN_CL_DEF-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[A_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[A_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[B_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[B_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[E_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[E_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[F_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[G_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[G_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[H_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[H_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP1_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP1]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP2_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP2]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP3_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP3]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP4_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP4]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP5_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP5]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP6_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP6]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP7_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP7]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP8_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP8]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP9_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP9]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP10_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP10]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP11_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP11]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP12_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP12]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP13_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP13]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP14_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP14]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[A]], ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[B]], ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[E]], ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[F]], ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[G]], ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[H]], ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// AMDGCN_CL_DEF-NEXT: ret void +// +// AMDGCN_CL_20-LABEL: define hidden void @fi3c( +// AMDGCN_CL_20-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[A_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[A_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[B_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[B_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[E_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[E_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[F_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[G_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[G_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[H_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[H_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP1_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP1]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP2_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP2]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP3_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP3]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP4_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP4]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP5_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP5]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP6_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP6]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP7_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP7]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP8_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP8]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP9_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP9]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP10_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP10]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP11_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP11]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP12_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP12]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP13_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP13]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP14_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP14]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[A]], ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[B]], ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[E]], ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[F]], ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[G]], ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[H]], ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// AMDGCN_CL_20-NEXT: ret void +// // SPIRV-LABEL: define hidden spir_func void @fi3c( -// SPIRV: [[TMP0:%.*]] = atomicrmw add ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("workgroup") monotonic, align 4 -// SPIRV: [[TMP1:%.*]] = atomicrmw sub ptr [[PTR1:%.+]], i32 [[VAL1:.+]] syncscope("workgroup") monotonic, align 4 -// SPIRV: [[TMP2:%.*]] = atomicrmw and ptr [[PTR2:%.+]], i32 [[VAL2:.+]] syncscope("workgroup") monotonic, align 4 -// SPIRV: [[TMP3:%.*]] = atomicrmw or ptr [[PTR3:%.+]], i32 [[VAL3:.+]] syncscope("workgroup") monotonic, align 4 -// SPIRV: [[TMP4:%.*]] = atomicrmw xor ptr [[PTR4:%.+]], i32 [[VAL4:.+]] syncscope("workgroup") monotonic, align 4 -// SPIRV: [[TMP5:%.*]] = atomicrmw nand ptr [[PTR5:%.+]], i32 [[VAL5:.+]] syncscope("workgroup") monotonic, align 4 -// SPIRV: [[TMP6:%.*]] = atomicrmw min ptr [[PTR6:%.+]], i32 [[VAL6:.+]] syncscope("workgroup") monotonic, align 4 -// SPIRV: [[TMP7:%.*]] = atomicrmw max ptr [[PTR7:%.+]], i32 [[VAL7:.+]] syncscope("workgroup") monotonic, align 4 +// SPIRV-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[E]], ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[F]], ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[G]], ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[H]], ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP]], align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// SPIRV-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP1]], align 4 +// SPIRV-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1]], align 4 +// SPIRV-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2]], align 4 +// SPIRV-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2]], align 4 +// SPIRV-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// SPIRV-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP3]], align 4 +// SPIRV-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3]], align 4 +// SPIRV-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4]], align 4 +// SPIRV-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4]], align 4 +// SPIRV-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// SPIRV-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP5]], align 4 +// SPIRV-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5]], align 4 +// SPIRV-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6]], align 4 +// SPIRV-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6]], align 4 +// SPIRV-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// SPIRV-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP7]], align 4 +// SPIRV-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7]], align 4 +// SPIRV-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8]], align 4 +// SPIRV-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8]], align 4 +// SPIRV-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// SPIRV-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP9]], align 4 +// SPIRV-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9]], align 4 +// SPIRV-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10]], align 4 +// SPIRV-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10]], align 4 +// SPIRV-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// SPIRV-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP11]], align 4 +// SPIRV-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11]], align 4 +// SPIRV-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12]], align 4 +// SPIRV-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12]], align 4 +// SPIRV-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// SPIRV-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP13]], align 4 +// SPIRV-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13]], align 4 +// SPIRV-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14]], align 4 +// SPIRV-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14]], align 4 +// SPIRV-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// SPIRV-NEXT: ret void +// void fi3c(int *a, int *b, int *c, int *d, int *e, int *f, int *g, int *h) { *a = __scoped_atomic_fetch_add(a, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_WRKGRP); *b = __scoped_atomic_fetch_sub(b, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_WRKGRP); @@ -177,24 +1502,719 @@ void fi3c(int *a, int *b, int *c, int *d, int *e, int *f, int *g, int *h) { *h = __scoped_atomic_fetch_max(h, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_WRKGRP); } -// AMDGCN-LABEL: define hidden void @fi3d( -// AMDGCN: [[TMP0:%.*]] = atomicrmw add ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("wavefront") monotonic, align 4 -// AMDGCN: [[TMP1:%.*]] = atomicrmw sub ptr [[PTR1:%.+]], i32 [[VAL1:.+]] syncscope("wavefront") monotonic, align 4 -// AMDGCN: [[TMP2:%.*]] = atomicrmw and ptr [[PTR2:%.+]], i32 [[VAL2:.+]] syncscope("wavefront") monotonic, align 4 -// AMDGCN: [[TMP3:%.*]] = atomicrmw or ptr [[PTR3:%.+]], i32 [[VAL3:.+]] syncscope("wavefront") monotonic, align 4 -// AMDGCN: [[TMP4:%.*]] = atomicrmw xor ptr [[PTR4:%.+]], i32 [[VAL4:.+]] syncscope("wavefront") monotonic, align 4 -// AMDGCN: [[TMP5:%.*]] = atomicrmw nand ptr [[PTR5:%.+]], i32 [[VAL5:.+]] syncscope("wavefront") monotonic, align 4 -// AMDGCN: [[TMP6:%.*]] = atomicrmw min ptr [[PTR6:%.+]], i32 [[VAL6:.+]] syncscope("wavefront") monotonic, align 4 -// AMDGCN: [[TMP7:%.*]] = atomicrmw max ptr [[PTR7:%.+]], i32 [[VAL7:.+]] syncscope("wavefront") monotonic, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden void @fi3_clustr( +// AMDGCN_CL_DEF-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[A_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[A_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[B_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[B_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[E_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[E_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[F_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[G_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[G_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[H_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[H_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP1_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP1]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP2_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP2]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP3_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP3]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP4_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP4]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP5_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP5]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP6_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP6]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP7_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP7]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP8_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP8]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP9_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP9]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP10_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP10]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP11_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP11]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP12_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP12]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP13_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP13]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP14_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP14]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[A]], ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[B]], ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[E]], ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[F]], ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[G]], ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[H]], ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// AMDGCN_CL_DEF-NEXT: ret void +// +// AMDGCN_CL_20-LABEL: define hidden void @fi3_clustr( +// AMDGCN_CL_20-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[A_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[A_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[B_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[B_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[E_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[E_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[F_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[G_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[G_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[H_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[H_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP1_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP1]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP2_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP2]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP3_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP3]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP4_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP4]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP5_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP5]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP6_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP6]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP7_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP7]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP8_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP8]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP9_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP9]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP10_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP10]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP11_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP11]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP12_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP12]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP13_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP13]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP14_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP14]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[A]], ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[B]], ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[E]], ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[F]], ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[G]], ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[H]], ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// AMDGCN_CL_20-NEXT: ret void +// +// SPIRV-LABEL: define hidden spir_func void @fi3_clustr( +// SPIRV-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[E]], ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[F]], ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[G]], ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[H]], ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP]], align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// SPIRV-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP1]], align 4 +// SPIRV-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1]], align 4 +// SPIRV-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2]], align 4 +// SPIRV-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2]], align 4 +// SPIRV-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// SPIRV-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP3]], align 4 +// SPIRV-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3]], align 4 +// SPIRV-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4]], align 4 +// SPIRV-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4]], align 4 +// SPIRV-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// SPIRV-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP5]], align 4 +// SPIRV-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5]], align 4 +// SPIRV-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6]], align 4 +// SPIRV-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6]], align 4 +// SPIRV-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// SPIRV-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP7]], align 4 +// SPIRV-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7]], align 4 +// SPIRV-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8]], align 4 +// SPIRV-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8]], align 4 +// SPIRV-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// SPIRV-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP9]], align 4 +// SPIRV-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9]], align 4 +// SPIRV-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10]], align 4 +// SPIRV-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10]], align 4 +// SPIRV-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// SPIRV-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP11]], align 4 +// SPIRV-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11]], align 4 +// SPIRV-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12]], align 4 +// SPIRV-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12]], align 4 +// SPIRV-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// SPIRV-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP13]], align 4 +// SPIRV-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13]], align 4 +// SPIRV-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14]], align 4 +// SPIRV-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14]], align 4 +// SPIRV-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// SPIRV-NEXT: ret void +// +void fi3_clustr(int *a, int *b, int *c, int *d, int *e, int *f, int *g, int *h) { + *a = __scoped_atomic_fetch_add(a, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR); + *b = __scoped_atomic_fetch_sub(b, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR); + *c = __scoped_atomic_fetch_and(c, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR); + *d = __scoped_atomic_fetch_or(d, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR); + *e = __scoped_atomic_fetch_xor(e, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR); + *f = __scoped_atomic_fetch_nand(f, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR); + *g = __scoped_atomic_fetch_min(g, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR); + *h = __scoped_atomic_fetch_max(h, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR); +} + +// AMDGCN_CL_DEF-LABEL: define hidden void @fi3d( +// AMDGCN_CL_DEF-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[A_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[A_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[B_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[B_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[E_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[E_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[F_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[G_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[G_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[H_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[H_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP1_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP1]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP2_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP2]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP3_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP3]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP4_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP4]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP5_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP5]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP6_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP6]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP7_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP7]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP8_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP8]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP9_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP9]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP10_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP10]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP11_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP11]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP12_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP12]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP13_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP13]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP14_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP14]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[A]], ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[B]], ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[E]], ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[F]], ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[G]], ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[H]], ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// AMDGCN_CL_DEF-NEXT: ret void +// +// AMDGCN_CL_20-LABEL: define hidden void @fi3d( +// AMDGCN_CL_20-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[A_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[A_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[B_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[B_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[E_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[E_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[F_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[G_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[G_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[H_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[H_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP1_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP1]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP2_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP2]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP3_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP3]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP4_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP4]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP5_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP5]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP6_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP6]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP7_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP7]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP8_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP8]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP9_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP9]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP10_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP10]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP11_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP11]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP12_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP12]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP13_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP13]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP14_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP14]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[A]], ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[B]], ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[E]], ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[F]], ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[G]], ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[H]], ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// AMDGCN_CL_20-NEXT: ret void +// // SPIRV-LABEL: define hidden spir_func void @fi3d( -// SPIRV: [[TMP0:%.*]] = atomicrmw add ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("subgroup") monotonic, align 4 -// SPIRV: [[TMP1:%.*]] = atomicrmw sub ptr [[PTR1:%.+]], i32 [[VAL1:.+]] syncscope("subgroup") monotonic, align 4 -// SPIRV: [[TMP2:%.*]] = atomicrmw and ptr [[PTR2:%.+]], i32 [[VAL2:.+]] syncscope("subgroup") monotonic, align 4 -// SPIRV: [[TMP3:%.*]] = atomicrmw or ptr [[PTR3:%.+]], i32 [[VAL3:.+]] syncscope("subgroup") monotonic, align 4 -// SPIRV: [[TMP4:%.*]] = atomicrmw xor ptr [[PTR4:%.+]], i32 [[VAL4:.+]] syncscope("subgroup") monotonic, align 4 -// SPIRV: [[TMP5:%.*]] = atomicrmw nand ptr [[PTR5:%.+]], i32 [[VAL5:.+]] syncscope("subgroup") monotonic, align 4 -// SPIRV: [[TMP6:%.*]] = atomicrmw min ptr [[PTR6:%.+]], i32 [[VAL6:.+]] syncscope("subgroup") monotonic, align 4 -// SPIRV: [[TMP7:%.*]] = atomicrmw max ptr [[PTR7:%.+]], i32 [[VAL7:.+]] syncscope("subgroup") monotonic, align 4 +// SPIRV-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[E]], ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[F]], ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[G]], ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[H]], ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("subgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP]], align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// SPIRV-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP1]], align 4 +// SPIRV-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1]], align 4 +// SPIRV-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("subgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2]], align 4 +// SPIRV-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2]], align 4 +// SPIRV-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// SPIRV-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP3]], align 4 +// SPIRV-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3]], align 4 +// SPIRV-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("subgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4]], align 4 +// SPIRV-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4]], align 4 +// SPIRV-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// SPIRV-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP5]], align 4 +// SPIRV-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5]], align 4 +// SPIRV-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("subgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6]], align 4 +// SPIRV-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6]], align 4 +// SPIRV-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// SPIRV-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP7]], align 4 +// SPIRV-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7]], align 4 +// SPIRV-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("subgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8]], align 4 +// SPIRV-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8]], align 4 +// SPIRV-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// SPIRV-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP9]], align 4 +// SPIRV-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9]], align 4 +// SPIRV-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("subgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10]], align 4 +// SPIRV-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10]], align 4 +// SPIRV-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// SPIRV-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP11]], align 4 +// SPIRV-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11]], align 4 +// SPIRV-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("subgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12]], align 4 +// SPIRV-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12]], align 4 +// SPIRV-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// SPIRV-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP13]], align 4 +// SPIRV-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13]], align 4 +// SPIRV-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("subgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14]], align 4 +// SPIRV-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14]], align 4 +// SPIRV-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// SPIRV-NEXT: ret void +// void fi3d(int *a, int *b, int *c, int *d, int *e, int *f, int *g, int *h) { *a = __scoped_atomic_fetch_add(a, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_WVFRNT); *b = __scoped_atomic_fetch_sub(b, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_WVFRNT); @@ -206,24 +2226,357 @@ void fi3d(int *a, int *b, int *c, int *d, int *e, int *f, int *g, int *h) { *h = __scoped_atomic_fetch_max(h, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_WVFRNT); } -// AMDGCN-LABEL: define hidden void @fi3e( -// AMDGCN: [[TMP0:%.*]] = atomicrmw add ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("singlethread") monotonic, align 4 -// AMDGCN: [[TMP1:%.*]] = atomicrmw sub ptr [[PTR1:%.+]], i32 [[VAL1:.+]] syncscope("singlethread") monotonic, align 4 -// AMDGCN: [[TMP2:%.*]] = atomicrmw and ptr [[PTR2:%.+]], i32 [[VAL2:.+]] syncscope("singlethread") monotonic, align 4 -// AMDGCN: [[TMP3:%.*]] = atomicrmw or ptr [[PTR3:%.+]], i32 [[VAL3:.+]] syncscope("singlethread") monotonic, align 4 -// AMDGCN: [[TMP4:%.*]] = atomicrmw xor ptr [[PTR4:%.+]], i32 [[VAL4:.+]] syncscope("singlethread") monotonic, align 4 -// AMDGCN: [[TMP5:%.*]] = atomicrmw nand ptr [[PTR5:%.+]], i32 [[VAL5:.+]] syncscope("singlethread") monotonic, align 4 -// AMDGCN: [[TMP6:%.*]] = atomicrmw min ptr [[PTR6:%.+]], i32 [[VAL6:.+]] syncscope("singlethread") monotonic, align 4 -// AMDGCN: [[TMP7:%.*]] = atomicrmw max ptr [[PTR7:%.+]], i32 [[VAL7:.+]] syncscope("singlethread") monotonic, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden void @fi3e( +// AMDGCN_CL_DEF-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[A_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[A_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[B_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[B_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[E_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[E_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[F_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[G_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[G_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[H_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[H_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP1_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP1]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP2_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP2]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP3_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP3]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP4_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP4]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP5_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP5]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP6_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP6]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP7_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP7]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP8_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP8]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP9_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP9]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP10_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP10]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP11_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP11]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP12_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP12]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP13_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP13]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP14_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP14]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[A]], ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[B]], ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[E]], ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[F]], ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[G]], ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[H]], ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// AMDGCN_CL_DEF-NEXT: ret void +// +// AMDGCN_CL_20-LABEL: define hidden void @fi3e( +// AMDGCN_CL_20-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[A_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[A_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[B_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[B_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[E_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[E_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[F_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[G_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[G_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[H_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[H_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP1_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP1]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP2_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP2]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP3_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP3]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP4_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP4]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP5_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP5]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP6_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP6]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP7_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP7]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP8_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP8]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP9_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP9]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP10_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP10]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP11_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP11]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP12_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP12]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP13_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP13]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP14_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP14]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[A]], ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[B]], ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[E]], ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[F]], ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[G]], ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[H]], ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// AMDGCN_CL_20-NEXT: ret void +// // SPIRV-LABEL: define hidden spir_func void @fi3e( -// SPIRV: [[TMP0:%.*]] = atomicrmw add ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("singlethread") monotonic, align 4 -// SPIRV: [[TMP1:%.*]] = atomicrmw sub ptr [[PTR1:%.+]], i32 [[VAL1:.+]] syncscope("singlethread") monotonic, align 4 -// SPIRV: [[TMP2:%.*]] = atomicrmw and ptr [[PTR2:%.+]], i32 [[VAL2:.+]] syncscope("singlethread") monotonic, align 4 -// SPIRV: [[TMP3:%.*]] = atomicrmw or ptr [[PTR3:%.+]], i32 [[VAL3:.+]] syncscope("singlethread") monotonic, align 4 -// SPIRV: [[TMP4:%.*]] = atomicrmw xor ptr [[PTR4:%.+]], i32 [[VAL4:.+]] syncscope("singlethread") monotonic, align 4 -// SPIRV: [[TMP5:%.*]] = atomicrmw nand ptr [[PTR5:%.+]], i32 [[VAL5:.+]] syncscope("singlethread") monotonic, align 4 -// SPIRV: [[TMP6:%.*]] = atomicrmw min ptr [[PTR6:%.+]], i32 [[VAL6:.+]] syncscope("singlethread") monotonic, align 4 -// SPIRV: [[TMP7:%.*]] = atomicrmw max ptr [[PTR7:%.+]], i32 [[VAL7:.+]] syncscope("singlethread") monotonic, align 4 +// SPIRV-SAME: ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], ptr noundef [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef [[E:%.*]], ptr noundef [[F:%.*]], ptr noundef [[G:%.*]], ptr noundef [[H:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[F_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[G_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[H_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP1:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP2:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP3:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP4:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP5:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP6:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP7:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP8:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP9:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP10:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP11:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP12:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP13:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[ATOMIC_TEMP14:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[E]], ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[F]], ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[G]], ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[H]], ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP1]] syncscope("singlethread") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP2]], ptr [[ATOMIC_TEMP]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP]], align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 +// SPIRV-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP1]], align 4 +// SPIRV-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTATOMICTMP1]], align 4 +// SPIRV-NEXT: [[TMP7:%.*]] = atomicrmw sub ptr [[TMP5]], i32 [[TMP6]] syncscope("singlethread") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP7]], ptr [[ATOMIC_TEMP2]], align 4 +// SPIRV-NEXT: [[TMP8:%.*]] = load i32, ptr [[ATOMIC_TEMP2]], align 4 +// SPIRV-NEXT: [[TMP9:%.*]] = load ptr, ptr [[B_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP8]], ptr [[TMP9]], align 4 +// SPIRV-NEXT: [[TMP10:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP3]], align 4 +// SPIRV-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTATOMICTMP3]], align 4 +// SPIRV-NEXT: [[TMP12:%.*]] = atomicrmw and ptr [[TMP10]], i32 [[TMP11]] syncscope("singlethread") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP12]], ptr [[ATOMIC_TEMP4]], align 4 +// SPIRV-NEXT: [[TMP13:%.*]] = load i32, ptr [[ATOMIC_TEMP4]], align 4 +// SPIRV-NEXT: [[TMP14:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP13]], ptr [[TMP14]], align 4 +// SPIRV-NEXT: [[TMP15:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP5]], align 4 +// SPIRV-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTATOMICTMP5]], align 4 +// SPIRV-NEXT: [[TMP17:%.*]] = atomicrmw or ptr [[TMP15]], i32 [[TMP16]] syncscope("singlethread") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP17]], ptr [[ATOMIC_TEMP6]], align 4 +// SPIRV-NEXT: [[TMP18:%.*]] = load i32, ptr [[ATOMIC_TEMP6]], align 4 +// SPIRV-NEXT: [[TMP19:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP18]], ptr [[TMP19]], align 4 +// SPIRV-NEXT: [[TMP20:%.*]] = load ptr, ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP7]], align 4 +// SPIRV-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTATOMICTMP7]], align 4 +// SPIRV-NEXT: [[TMP22:%.*]] = atomicrmw xor ptr [[TMP20]], i32 [[TMP21]] syncscope("singlethread") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP22]], ptr [[ATOMIC_TEMP8]], align 4 +// SPIRV-NEXT: [[TMP23:%.*]] = load i32, ptr [[ATOMIC_TEMP8]], align 4 +// SPIRV-NEXT: [[TMP24:%.*]] = load ptr, ptr [[E_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP23]], ptr [[TMP24]], align 4 +// SPIRV-NEXT: [[TMP25:%.*]] = load ptr, ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP9]], align 4 +// SPIRV-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTATOMICTMP9]], align 4 +// SPIRV-NEXT: [[TMP27:%.*]] = atomicrmw nand ptr [[TMP25]], i32 [[TMP26]] syncscope("singlethread") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP27]], ptr [[ATOMIC_TEMP10]], align 4 +// SPIRV-NEXT: [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP10]], align 4 +// SPIRV-NEXT: [[TMP29:%.*]] = load ptr, ptr [[F_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP28]], ptr [[TMP29]], align 4 +// SPIRV-NEXT: [[TMP30:%.*]] = load ptr, ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP11]], align 4 +// SPIRV-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTATOMICTMP11]], align 4 +// SPIRV-NEXT: [[TMP32:%.*]] = atomicrmw min ptr [[TMP30]], i32 [[TMP31]] syncscope("singlethread") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP32]], ptr [[ATOMIC_TEMP12]], align 4 +// SPIRV-NEXT: [[TMP33:%.*]] = load i32, ptr [[ATOMIC_TEMP12]], align 4 +// SPIRV-NEXT: [[TMP34:%.*]] = load ptr, ptr [[G_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP33]], ptr [[TMP34]], align 4 +// SPIRV-NEXT: [[TMP35:%.*]] = load ptr, ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP13]], align 4 +// SPIRV-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTATOMICTMP13]], align 4 +// SPIRV-NEXT: [[TMP37:%.*]] = atomicrmw max ptr [[TMP35]], i32 [[TMP36]] syncscope("singlethread") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP37]], ptr [[ATOMIC_TEMP14]], align 4 +// SPIRV-NEXT: [[TMP38:%.*]] = load i32, ptr [[ATOMIC_TEMP14]], align 4 +// SPIRV-NEXT: [[TMP39:%.*]] = load ptr, ptr [[H_ADDR]], align 8 +// SPIRV-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 4 +// SPIRV-NEXT: ret void +// void fi3e(int *a, int *b, int *c, int *d, int *e, int *f, int *g, int *h) { *a = __scoped_atomic_fetch_add(a, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_SINGLE); *b = __scoped_atomic_fetch_sub(b, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_SINGLE); @@ -235,10 +2588,98 @@ void fi3e(int *a, int *b, int *c, int *d, int *e, int *f, int *g, int *h) { *h = __scoped_atomic_fetch_max(h, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_SINGLE); } -// AMDGCN-LABEL: define hidden zeroext i1 @fi4a( -// AMDGCN-DAG: [[TMP0:%.*]] = cmpxchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] acquire acquire, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi4a( +// AMDGCN_CL_DEF-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DESIRED:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DESIRED_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DESIRED]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 0, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DESIRED_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[DESIRED_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] acquire acquire, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_DEF-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_DEF: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP4]], ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_DEF: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_DEF-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_DEF-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi4a( +// AMDGCN_CL_20-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DESIRED:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 0, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: store i32 1, ptr addrspace(5) [[DESIRED]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = addrspacecast ptr addrspace(5) [[DESIRED]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = load i32, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP2]], i32 [[TMP3]] acquire acquire, align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP4]], 0 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = extractvalue { i32, i1 } [[TMP4]], 1 +// AMDGCN_CL_20-NEXT: br i1 [[TMP6]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_20: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_20-NEXT: store i32 [[TMP5]], ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_20: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_20-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP6]] to i8 +// AMDGCN_CL_20-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP7:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP7]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi4a( -// SPIRV-DAG: [[TMP0:%.*]] = cmpxchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] acquire acquire, align 4 +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[CMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DESIRED:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 0, ptr [[CMP]], align 4 +// SPIRV-NEXT: store i32 1, ptr [[DESIRED]], align 4 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[DESIRED]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] acquire acquire, align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// SPIRV-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// SPIRV-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// SPIRV: [[CMPXCHG_STORE_EXPECTED]]: +// SPIRV-NEXT: store i32 [[TMP4]], ptr [[CMP]], align 4 +// SPIRV-NEXT: br label %[[CMPXCHG_CONTINUE]] +// SPIRV: [[CMPXCHG_CONTINUE]]: +// SPIRV-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// SPIRV-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi4a(int *i) { int cmp = 0; int desired = 1; @@ -247,10 +2688,98 @@ _Bool fi4a(int *i) { __MEMORY_SCOPE_SYSTEM); } -// AMDGCN-LABEL: define hidden zeroext i1 @fi4b( -// AMDGCN-DAG: [[TMP0:%.*]] = cmpxchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("agent") acquire acquire, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi4b( +// AMDGCN_CL_DEF-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DESIRED:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DESIRED_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DESIRED]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 0, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DESIRED_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[DESIRED_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("agent") acquire acquire, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_DEF-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_DEF: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP4]], ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_DEF: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_DEF-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_DEF-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi4b( +// AMDGCN_CL_20-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DESIRED:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 0, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: store i32 1, ptr addrspace(5) [[DESIRED]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = addrspacecast ptr addrspace(5) [[DESIRED]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = load i32, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP2]], i32 [[TMP3]] syncscope("agent") acquire acquire, align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP4]], 0 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = extractvalue { i32, i1 } [[TMP4]], 1 +// AMDGCN_CL_20-NEXT: br i1 [[TMP6]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_20: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_20-NEXT: store i32 [[TMP5]], ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_20: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_20-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP6]] to i8 +// AMDGCN_CL_20-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP7:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP7]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi4b( -// SPIRV-DAG: [[TMP0:%.*]] = cmpxchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("device") acquire acquire, align 4 +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[CMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DESIRED:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 0, ptr [[CMP]], align 4 +// SPIRV-NEXT: store i32 1, ptr [[DESIRED]], align 4 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[DESIRED]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("device") acquire acquire, align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// SPIRV-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// SPIRV-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// SPIRV: [[CMPXCHG_STORE_EXPECTED]]: +// SPIRV-NEXT: store i32 [[TMP4]], ptr [[CMP]], align 4 +// SPIRV-NEXT: br label %[[CMPXCHG_CONTINUE]] +// SPIRV: [[CMPXCHG_CONTINUE]]: +// SPIRV-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// SPIRV-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi4b(int *i) { int cmp = 0; int desired = 1; @@ -259,10 +2788,98 @@ _Bool fi4b(int *i) { __MEMORY_SCOPE_DEVICE); } -// AMDGCN-LABEL: define hidden zeroext i1 @fi4c( -// AMDGCN: [[TMP0:%.*]] = cmpxchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("workgroup") acquire acquire, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi4c( +// AMDGCN_CL_DEF-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DESIRED:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DESIRED_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DESIRED]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 0, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DESIRED_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[DESIRED_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("workgroup") acquire acquire, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_DEF-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_DEF: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP4]], ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_DEF: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_DEF-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_DEF-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi4c( +// AMDGCN_CL_20-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DESIRED:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 0, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: store i32 1, ptr addrspace(5) [[DESIRED]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = addrspacecast ptr addrspace(5) [[DESIRED]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = load i32, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP2]], i32 [[TMP3]] syncscope("workgroup") acquire acquire, align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP4]], 0 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = extractvalue { i32, i1 } [[TMP4]], 1 +// AMDGCN_CL_20-NEXT: br i1 [[TMP6]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_20: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_20-NEXT: store i32 [[TMP5]], ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_20: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_20-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP6]] to i8 +// AMDGCN_CL_20-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP7:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP7]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi4c( -// SPIRV: [[TMP0:%.*]] = cmpxchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("workgroup") acquire acquire, align 4 +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[CMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DESIRED:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 0, ptr [[CMP]], align 4 +// SPIRV-NEXT: store i32 1, ptr [[DESIRED]], align 4 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[DESIRED]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("workgroup") acquire acquire, align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// SPIRV-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// SPIRV-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// SPIRV: [[CMPXCHG_STORE_EXPECTED]]: +// SPIRV-NEXT: store i32 [[TMP4]], ptr [[CMP]], align 4 +// SPIRV-NEXT: br label %[[CMPXCHG_CONTINUE]] +// SPIRV: [[CMPXCHG_CONTINUE]]: +// SPIRV-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// SPIRV-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi4c(int *i) { int cmp = 0; int desired = 1; @@ -271,10 +2888,198 @@ _Bool fi4c(int *i) { __MEMORY_SCOPE_WRKGRP); } -// AMDGCN-LABEL: define hidden zeroext i1 @fi4d( -// AMDGCN: [[TMP0:%.*]] = cmpxchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("wavefront") acquire acquire, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi4_clustr( +// AMDGCN_CL_DEF-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DESIRED:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DESIRED_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DESIRED]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 0, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DESIRED_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[DESIRED_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("cluster") acquire acquire, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_DEF-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_DEF: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP4]], ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_DEF: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_DEF-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_DEF-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi4_clustr( +// AMDGCN_CL_20-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DESIRED:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 0, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: store i32 1, ptr addrspace(5) [[DESIRED]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = addrspacecast ptr addrspace(5) [[DESIRED]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = load i32, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP2]], i32 [[TMP3]] syncscope("cluster") acquire acquire, align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP4]], 0 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = extractvalue { i32, i1 } [[TMP4]], 1 +// AMDGCN_CL_20-NEXT: br i1 [[TMP6]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_20: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_20-NEXT: store i32 [[TMP5]], ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_20: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_20-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP6]] to i8 +// AMDGCN_CL_20-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP7:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP7]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// +// SPIRV-LABEL: define hidden spir_func zeroext i1 @fi4_clustr( +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[CMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DESIRED:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 0, ptr [[CMP]], align 4 +// SPIRV-NEXT: store i32 1, ptr [[DESIRED]], align 4 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[DESIRED]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("workgroup") acquire acquire, align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// SPIRV-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// SPIRV-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// SPIRV: [[CMPXCHG_STORE_EXPECTED]]: +// SPIRV-NEXT: store i32 [[TMP4]], ptr [[CMP]], align 4 +// SPIRV-NEXT: br label %[[CMPXCHG_CONTINUE]] +// SPIRV: [[CMPXCHG_CONTINUE]]: +// SPIRV-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// SPIRV-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// +_Bool fi4_clustr(int *i) { + int cmp = 0; + int desired = 1; + return __scoped_atomic_compare_exchange(i, &cmp, &desired, 0, + __ATOMIC_ACQUIRE, __ATOMIC_ACQUIRE, + __MEMORY_SCOPE_CLUSTR); +} + +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi4d( +// AMDGCN_CL_DEF-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DESIRED:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DESIRED_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DESIRED]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 0, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DESIRED_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[DESIRED_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("wavefront") acquire acquire, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_DEF-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_DEF: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP4]], ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_DEF: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_DEF-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_DEF-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi4d( +// AMDGCN_CL_20-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DESIRED:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 0, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: store i32 1, ptr addrspace(5) [[DESIRED]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = addrspacecast ptr addrspace(5) [[DESIRED]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = load i32, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP2]], i32 [[TMP3]] syncscope("wavefront") acquire acquire, align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP4]], 0 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = extractvalue { i32, i1 } [[TMP4]], 1 +// AMDGCN_CL_20-NEXT: br i1 [[TMP6]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_20: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_20-NEXT: store i32 [[TMP5]], ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_20: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_20-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP6]] to i8 +// AMDGCN_CL_20-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP7:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP7]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi4d( -// SPIRV: [[TMP0:%.*]] = cmpxchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("subgroup") acquire acquire, align 4 +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[CMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DESIRED:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 0, ptr [[CMP]], align 4 +// SPIRV-NEXT: store i32 1, ptr [[DESIRED]], align 4 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[DESIRED]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("subgroup") acquire acquire, align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// SPIRV-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// SPIRV-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// SPIRV: [[CMPXCHG_STORE_EXPECTED]]: +// SPIRV-NEXT: store i32 [[TMP4]], ptr [[CMP]], align 4 +// SPIRV-NEXT: br label %[[CMPXCHG_CONTINUE]] +// SPIRV: [[CMPXCHG_CONTINUE]]: +// SPIRV-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// SPIRV-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi4d(int *i) { int cmp = 0; int desired = 1; @@ -283,10 +3088,98 @@ _Bool fi4d(int *i) { __MEMORY_SCOPE_WVFRNT); } -// AMDGCN-LABEL: define hidden zeroext i1 @fi4e( -// AMDGCN: [[TMP0:%.*]] = cmpxchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("singlethread") acquire acquire, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi4e( +// AMDGCN_CL_DEF-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DESIRED:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DESIRED_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DESIRED]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 0, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DESIRED_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[DESIRED_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("singlethread") acquire acquire, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_DEF-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_DEF: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP4]], ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_DEF: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_DEF-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_DEF-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi4e( +// AMDGCN_CL_20-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DESIRED:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 0, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: store i32 1, ptr addrspace(5) [[DESIRED]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = addrspacecast ptr addrspace(5) [[DESIRED]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = load i32, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP2]], i32 [[TMP3]] syncscope("singlethread") acquire acquire, align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP4]], 0 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = extractvalue { i32, i1 } [[TMP4]], 1 +// AMDGCN_CL_20-NEXT: br i1 [[TMP6]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_20: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_20-NEXT: store i32 [[TMP5]], ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_20: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_20-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP6]] to i8 +// AMDGCN_CL_20-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP7:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP7]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi4e( -// SPIRV: [[TMP0:%.*]] = cmpxchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("singlethread") acquire acquire, align 4 +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[CMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DESIRED:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 0, ptr [[CMP]], align 4 +// SPIRV-NEXT: store i32 1, ptr [[DESIRED]], align 4 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[DESIRED]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("singlethread") acquire acquire, align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// SPIRV-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// SPIRV-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// SPIRV: [[CMPXCHG_STORE_EXPECTED]]: +// SPIRV-NEXT: store i32 [[TMP4]], ptr [[CMP]], align 4 +// SPIRV-NEXT: br label %[[CMPXCHG_CONTINUE]] +// SPIRV: [[CMPXCHG_CONTINUE]]: +// SPIRV-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// SPIRV-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi4e(int *i) { int cmp = 0; int desired = 1; @@ -295,10 +3188,98 @@ _Bool fi4e(int *i) { __MEMORY_SCOPE_SINGLE); } -// AMDGCN-LABEL: define hidden zeroext i1 @fi5a( -// AMDGCN: [[TMP0:%.*]] = cmpxchg weak ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] acquire acquire, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi5a( +// AMDGCN_CL_DEF-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 0, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] acquire acquire, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_DEF-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_DEF: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP4]], ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_DEF: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_DEF-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_DEF-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi5a( +// AMDGCN_CL_20-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 0, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i32, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] acquire acquire, align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_20-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_20: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_20-NEXT: store i32 [[TMP4]], ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_20: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_20-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_20-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi5a( -// SPIRV: [[TMP0:%.*]] = cmpxchg weak ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] acquire acquire, align 4 +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[CMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 0, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] acquire acquire, align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// SPIRV-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// SPIRV-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// SPIRV: [[CMPXCHG_STORE_EXPECTED]]: +// SPIRV-NEXT: store i32 [[TMP4]], ptr [[CMP]], align 4 +// SPIRV-NEXT: br label %[[CMPXCHG_CONTINUE]] +// SPIRV: [[CMPXCHG_CONTINUE]]: +// SPIRV-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// SPIRV-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi5a(int *i) { int cmp = 0; return __scoped_atomic_compare_exchange_n(i, &cmp, 1, 1, __ATOMIC_ACQUIRE, @@ -306,10 +3287,98 @@ _Bool fi5a(int *i) { __MEMORY_SCOPE_SYSTEM); } -// AMDGCN-LABEL: define hidden zeroext i1 @fi5b( -// AMDGCN: [[TMP0:%.*]] = cmpxchg weak ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("agent") acquire acquire, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi5b( +// AMDGCN_CL_DEF-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 0, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("agent") acquire acquire, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_DEF-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_DEF: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP4]], ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_DEF: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_DEF-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_DEF-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi5b( +// AMDGCN_CL_20-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 0, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i32, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("agent") acquire acquire, align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_20-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_20: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_20-NEXT: store i32 [[TMP4]], ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_20: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_20-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_20-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi5b( -// SPIRV: [[TMP0:%.*]] = cmpxchg weak ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("device") acquire acquire, align 4 +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[CMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 0, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("device") acquire acquire, align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// SPIRV-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// SPIRV-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// SPIRV: [[CMPXCHG_STORE_EXPECTED]]: +// SPIRV-NEXT: store i32 [[TMP4]], ptr [[CMP]], align 4 +// SPIRV-NEXT: br label %[[CMPXCHG_CONTINUE]] +// SPIRV: [[CMPXCHG_CONTINUE]]: +// SPIRV-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// SPIRV-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi5b(int *i) { int cmp = 0; return __scoped_atomic_compare_exchange_n(i, &cmp, 1, 1, __ATOMIC_ACQUIRE, @@ -317,127 +3386,1161 @@ _Bool fi5b(int *i) { __MEMORY_SCOPE_DEVICE); } -// AMDGCN-LABEL: define hidden zeroext i1 @fi5c( -// AMDGCN: [[TMP0:%.*]] = cmpxchg weak ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("workgroup") acquire acquire, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi5c( +// AMDGCN_CL_DEF-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 0, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("workgroup") acquire acquire, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_DEF-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_DEF: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP4]], ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_DEF: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_DEF-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_DEF-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi5c( +// AMDGCN_CL_20-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 0, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i32, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("workgroup") acquire acquire, align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_20-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_20: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_20-NEXT: store i32 [[TMP4]], ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_20: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_20-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_20-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi5c( -// SPIRV: [[TMP0:%.*]] = cmpxchg weak ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("workgroup") acquire acquire, align 4 +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[CMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 0, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("workgroup") acquire acquire, align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// SPIRV-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// SPIRV-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// SPIRV: [[CMPXCHG_STORE_EXPECTED]]: +// SPIRV-NEXT: store i32 [[TMP4]], ptr [[CMP]], align 4 +// SPIRV-NEXT: br label %[[CMPXCHG_CONTINUE]] +// SPIRV: [[CMPXCHG_CONTINUE]]: +// SPIRV-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// SPIRV-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi5c(int *i) { int cmp = 0; return __scoped_atomic_compare_exchange_n( i, &cmp, 1, 1, __ATOMIC_ACQUIRE, __ATOMIC_ACQUIRE, __MEMORY_SCOPE_WRKGRP); } -// AMDGCN-LABEL: define hidden zeroext i1 @fi5d( -// AMDGCN: [[TMP0:%.*]] = cmpxchg weak ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("wavefront") acquire acquire, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi5_clustr( +// AMDGCN_CL_DEF-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 0, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("cluster") acquire acquire, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_DEF-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_DEF: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP4]], ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_DEF: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_DEF-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_DEF-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi5_clustr( +// AMDGCN_CL_20-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 0, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i32, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("cluster") acquire acquire, align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_20-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_20: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_20-NEXT: store i32 [[TMP4]], ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_20: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_20-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_20-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// +// SPIRV-LABEL: define hidden spir_func zeroext i1 @fi5_clustr( +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[CMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 0, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("workgroup") acquire acquire, align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// SPIRV-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// SPIRV-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// SPIRV: [[CMPXCHG_STORE_EXPECTED]]: +// SPIRV-NEXT: store i32 [[TMP4]], ptr [[CMP]], align 4 +// SPIRV-NEXT: br label %[[CMPXCHG_CONTINUE]] +// SPIRV: [[CMPXCHG_CONTINUE]]: +// SPIRV-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// SPIRV-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// +_Bool fi5_clustr(int *i) { + int cmp = 0; + return __scoped_atomic_compare_exchange_n( + i, &cmp, 1, 1, __ATOMIC_ACQUIRE, __ATOMIC_ACQUIRE, __MEMORY_SCOPE_CLUSTR); +} + +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi5d( +// AMDGCN_CL_DEF-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 0, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("wavefront") acquire acquire, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_DEF-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_DEF: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP4]], ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_DEF: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_DEF-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_DEF-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi5d( +// AMDGCN_CL_20-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 0, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i32, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("wavefront") acquire acquire, align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_20-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_20: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_20-NEXT: store i32 [[TMP4]], ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_20: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_20-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_20-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi5d( -// SPIRV: [[TMP0:%.*]] = cmpxchg weak ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("subgroup") acquire acquire, align 4 +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[CMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 0, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("subgroup") acquire acquire, align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// SPIRV-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// SPIRV-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// SPIRV: [[CMPXCHG_STORE_EXPECTED]]: +// SPIRV-NEXT: store i32 [[TMP4]], ptr [[CMP]], align 4 +// SPIRV-NEXT: br label %[[CMPXCHG_CONTINUE]] +// SPIRV: [[CMPXCHG_CONTINUE]]: +// SPIRV-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// SPIRV-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi5d(int *i) { int cmp = 0; return __scoped_atomic_compare_exchange_n( i, &cmp, 1, 1, __ATOMIC_ACQUIRE, __ATOMIC_ACQUIRE, __MEMORY_SCOPE_WVFRNT); } -// AMDGCN-LABEL: define hidden zeroext i1 @fi5e( -// AMDGCN: [[TMP0:%.*]] = cmpxchg weak ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("singlethread") acquire acquire, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi5e( +// AMDGCN_CL_DEF-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 0, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("singlethread") acquire acquire, align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_DEF-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_DEF-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_DEF: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP4]], ptr [[CMP_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_DEF: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_DEF-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_DEF-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi5e( +// AMDGCN_CL_20-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[I_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[CMPXCHG_BOOL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[CMPXCHG_BOOL]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[I]], ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 0, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i32 1, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i32, ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP_ASCAST]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("singlethread") acquire acquire, align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// AMDGCN_CL_20-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// AMDGCN_CL_20: [[CMPXCHG_STORE_EXPECTED]]: +// AMDGCN_CL_20-NEXT: store i32 [[TMP4]], ptr addrspace(5) [[CMP]], align 4 +// AMDGCN_CL_20-NEXT: br label %[[CMPXCHG_CONTINUE]] +// AMDGCN_CL_20: [[CMPXCHG_CONTINUE]]: +// AMDGCN_CL_20-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// AMDGCN_CL_20-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi5e( -// SPIRV: [[TMP0:%.*]] = cmpxchg weak ptr [[PTR0:%.+]], i32 [[VAL0:.+]], i32 [[VAL1:.+]] syncscope("singlethread") acquire acquire, align 4 +// SPIRV-SAME: ptr noundef [[I:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[CMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: [[CMPXCHG_BOOL:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 0, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 +// SPIRV-NEXT: store i32 1, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP1:%.*]] = load i32, ptr [[CMP]], align 4 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTATOMICTMP]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = cmpxchg weak ptr [[TMP0]], i32 [[TMP1]], i32 [[TMP2]] syncscope("singlethread") acquire acquire, align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = extractvalue { i32, i1 } [[TMP3]], 0 +// SPIRV-NEXT: [[TMP5:%.*]] = extractvalue { i32, i1 } [[TMP3]], 1 +// SPIRV-NEXT: br i1 [[TMP5]], label %[[CMPXCHG_CONTINUE:.*]], label %[[CMPXCHG_STORE_EXPECTED:.*]] +// SPIRV: [[CMPXCHG_STORE_EXPECTED]]: +// SPIRV-NEXT: store i32 [[TMP4]], ptr [[CMP]], align 4 +// SPIRV-NEXT: br label %[[CMPXCHG_CONTINUE]] +// SPIRV: [[CMPXCHG_CONTINUE]]: +// SPIRV-NEXT: [[STOREDV:%.*]] = zext i1 [[TMP5]] to i8 +// SPIRV-NEXT: store i8 [[STOREDV]], ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr [[CMPXCHG_BOOL]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi5e(int *i) { int cmp = 0; return __scoped_atomic_compare_exchange_n( i, &cmp, 1, 1, __ATOMIC_ACQUIRE, __ATOMIC_ACQUIRE, __MEMORY_SCOPE_SINGLE); } -// AMDGCN-LABEL: define hidden i32 @fi6a( -// AMDGCN: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]] monotonic, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden i32 @fi6a( +// AMDGCN_CL_DEF-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RET:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[RET_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RET]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP2]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP3]], ptr [[RET_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = load i32, ptr [[RET_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: ret i32 [[TMP4]] +// +// AMDGCN_CL_20-LABEL: define hidden i32 @fi6a( +// AMDGCN_CL_20-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RET:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = addrspacecast ptr addrspace(5) [[RET]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP3]] monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP4]], ptr [[TMP2]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = load i32, ptr addrspace(5) [[RET]], align 4 +// AMDGCN_CL_20-NEXT: ret i32 [[TMP5]] +// // SPIRV-LABEL: define hidden spir_func i32 @fi6a( -// SPIRV: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]] monotonic, align 4 +// SPIRV-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[RET:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP2]] monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP3]], ptr [[RET]], align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load i32, ptr [[RET]], align 4 +// SPIRV-NEXT: ret i32 [[TMP4]] +// int fi6a(int *c, int *d) { int ret; __scoped_atomic_exchange(c, d, &ret, __ATOMIC_RELAXED, __MEMORY_SCOPE_SYSTEM); return ret; } -// AMDGCN-LABEL: define hidden i32 @fi6b( -// AMDGCN: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("agent") monotonic, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden i32 @fi6b( +// AMDGCN_CL_DEF-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RET:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[RET_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RET]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP2]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP3]], ptr [[RET_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = load i32, ptr [[RET_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: ret i32 [[TMP4]] +// +// AMDGCN_CL_20-LABEL: define hidden i32 @fi6b( +// AMDGCN_CL_20-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RET:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = addrspacecast ptr addrspace(5) [[RET]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP3]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP4]], ptr [[TMP2]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = load i32, ptr addrspace(5) [[RET]], align 4 +// AMDGCN_CL_20-NEXT: ret i32 [[TMP5]] +// // SPIRV-LABEL: define hidden spir_func i32 @fi6b( -// SPIRV: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("device") monotonic, align 4 +// SPIRV-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[RET:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP2]] syncscope("device") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP3]], ptr [[RET]], align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load i32, ptr [[RET]], align 4 +// SPIRV-NEXT: ret i32 [[TMP4]] +// int fi6b(int *c, int *d) { int ret; __scoped_atomic_exchange(c, d, &ret, __ATOMIC_RELAXED, __MEMORY_SCOPE_DEVICE); return ret; } -// AMDGCN-LABEL: define hidden i32 @fi6c( -// AMDGCN: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("workgroup") monotonic, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden i32 @fi6c( +// AMDGCN_CL_DEF-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RET:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[RET_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RET]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP2]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP3]], ptr [[RET_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = load i32, ptr [[RET_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: ret i32 [[TMP4]] +// +// AMDGCN_CL_20-LABEL: define hidden i32 @fi6c( +// AMDGCN_CL_20-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RET:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = addrspacecast ptr addrspace(5) [[RET]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP3]] syncscope("workgroup") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP4]], ptr [[TMP2]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = load i32, ptr addrspace(5) [[RET]], align 4 +// AMDGCN_CL_20-NEXT: ret i32 [[TMP5]] +// // SPIRV-LABEL: define hidden spir_func i32 @fi6c( -// SPIRV: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("workgroup") monotonic, align 4 +// SPIRV-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[RET:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP2]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP3]], ptr [[RET]], align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load i32, ptr [[RET]], align 4 +// SPIRV-NEXT: ret i32 [[TMP4]] +// int fi6c(int *c, int *d) { int ret; __scoped_atomic_exchange(c, d, &ret, __ATOMIC_RELAXED, __MEMORY_SCOPE_WRKGRP); return ret; } -// AMDGCN-LABEL: define hidden i32 @fi6d( -// AMDGCN: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("wavefront") monotonic, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden i32 @fi6_clustr( +// AMDGCN_CL_DEF-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RET:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[RET_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RET]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP2]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP3]], ptr [[RET_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = load i32, ptr [[RET_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: ret i32 [[TMP4]] +// +// AMDGCN_CL_20-LABEL: define hidden i32 @fi6_clustr( +// AMDGCN_CL_20-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RET:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = addrspacecast ptr addrspace(5) [[RET]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP3]] syncscope("cluster") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP4]], ptr [[TMP2]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = load i32, ptr addrspace(5) [[RET]], align 4 +// AMDGCN_CL_20-NEXT: ret i32 [[TMP5]] +// +// SPIRV-LABEL: define hidden spir_func i32 @fi6_clustr( +// SPIRV-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[RET:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP2]] syncscope("workgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP3]], ptr [[RET]], align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load i32, ptr [[RET]], align 4 +// SPIRV-NEXT: ret i32 [[TMP4]] +// +int fi6_clustr(int *c, int *d) { + int ret; + __scoped_atomic_exchange(c, d, &ret, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR); + return ret; +} + +// AMDGCN_CL_DEF-LABEL: define hidden i32 @fi6d( +// AMDGCN_CL_DEF-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RET:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[RET_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RET]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP2]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP3]], ptr [[RET_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = load i32, ptr [[RET_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: ret i32 [[TMP4]] +// +// AMDGCN_CL_20-LABEL: define hidden i32 @fi6d( +// AMDGCN_CL_20-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RET:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = addrspacecast ptr addrspace(5) [[RET]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP3]] syncscope("wavefront") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP4]], ptr [[TMP2]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = load i32, ptr addrspace(5) [[RET]], align 4 +// AMDGCN_CL_20-NEXT: ret i32 [[TMP5]] +// // SPIRV-LABEL: define hidden spir_func i32 @fi6d( -// SPIRV: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("subgroup") monotonic, align 4 +// SPIRV-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[RET:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP2]] syncscope("subgroup") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP3]], ptr [[RET]], align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load i32, ptr [[RET]], align 4 +// SPIRV-NEXT: ret i32 [[TMP4]] +// int fi6d(int *c, int *d) { int ret; __scoped_atomic_exchange(c, d, &ret, __ATOMIC_RELAXED, __MEMORY_SCOPE_WVFRNT); return ret; } -// AMDGCN-LABEL: define hidden i32 @fi6e( -// AMDGCN: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("singlethread") monotonic, align 4 +// AMDGCN_CL_DEF-LABEL: define hidden i32 @fi6e( +// AMDGCN_CL_DEF-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RET:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[RET_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RET]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP2]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i32 [[TMP3]], ptr [[RET_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: [[TMP4:%.*]] = load i32, ptr [[RET_ASCAST]], align 4 +// AMDGCN_CL_DEF-NEXT: ret i32 [[TMP4]] +// +// AMDGCN_CL_20-LABEL: define hidden i32 @fi6e( +// AMDGCN_CL_20-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RET:%.*]] = alloca i32, align 4, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store ptr [[D]], ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = addrspacecast ptr addrspace(5) [[RET]] to ptr +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP4:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP3]] syncscope("singlethread") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i32 [[TMP4]], ptr [[TMP2]], align 4 +// AMDGCN_CL_20-NEXT: [[TMP5:%.*]] = load i32, ptr addrspace(5) [[RET]], align 4 +// AMDGCN_CL_20-NEXT: ret i32 [[TMP5]] +// // SPIRV-LABEL: define hidden spir_func i32 @fi6e( -// SPIRV: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i32 [[VAL0:.+]] syncscope("singlethread") monotonic, align 4 +// SPIRV-SAME: ptr noundef [[C:%.*]], ptr noundef [[D:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[D_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[RET:%.*]] = alloca i32, align 4 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: [[TMP1:%.*]] = load ptr, ptr [[D_ADDR]], align 8 +// SPIRV-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 +// SPIRV-NEXT: [[TMP3:%.*]] = atomicrmw xchg ptr [[TMP0]], i32 [[TMP2]] syncscope("singlethread") monotonic, align 4 +// SPIRV-NEXT: store i32 [[TMP3]], ptr [[RET]], align 4 +// SPIRV-NEXT: [[TMP4:%.*]] = load i32, ptr [[RET]], align 4 +// SPIRV-NEXT: ret i32 [[TMP4]] +// int fi6e(int *c, int *d) { int ret; __scoped_atomic_exchange(c, d, &ret, __ATOMIC_RELAXED, __MEMORY_SCOPE_SINGLE); return ret; } -// AMDGCN-LABEL: define hidden zeroext i1 @fi7a( -// AMDGCN: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i8 [[VAL0:.+]] monotonic, align 1 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi7a( +// AMDGCN_CL_DEF-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i8 1, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] monotonic, align 1, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi7a( +// AMDGCN_CL_20-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i8 1, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] monotonic, align 1, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi7a( -// SPIRV: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i8 [[VAL0:.+]] monotonic, align 1 +// SPIRV-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i8 1, ptr [[DOTATOMICTMP]], align 1 +// SPIRV-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP]], align 1 +// SPIRV-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] monotonic, align 1 +// SPIRV-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP]], align 1 +// SPIRV-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi7a(_Bool *c) { return __scoped_atomic_exchange_n(c, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_SYSTEM); } -// AMDGCN-LABEL: define hidden zeroext i1 @fi7b( -// AMDGCN: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i8 [[VAL0:.+]] syncscope("agent") monotonic, align 1 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi7b( +// AMDGCN_CL_DEF-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i8 1, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("agent") monotonic, align 1, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi7b( +// AMDGCN_CL_20-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i8 1, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("agent") monotonic, align 1, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi7b( -// SPIRV: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i8 [[VAL0:.+]] syncscope("device") monotonic, align 1 +// SPIRV-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i8 1, ptr [[DOTATOMICTMP]], align 1 +// SPIRV-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP]], align 1 +// SPIRV-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("device") monotonic, align 1 +// SPIRV-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP]], align 1 +// SPIRV-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi7b(_Bool *c) { return __scoped_atomic_exchange_n(c, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_DEVICE); } -// AMDGCN-LABEL: define hidden zeroext i1 @fi7c( -// AMDGCN: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i8 [[VAL0:.+]] syncscope("workgroup") monotonic, align 1 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi7c( +// AMDGCN_CL_DEF-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i8 1, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("workgroup") monotonic, align 1, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi7c( +// AMDGCN_CL_20-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i8 1, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("workgroup") monotonic, align 1, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi7c( -// SPIRV: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i8 [[VAL0:.+]] syncscope("workgroup") monotonic, align 1 +// SPIRV-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i8 1, ptr [[DOTATOMICTMP]], align 1 +// SPIRV-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP]], align 1 +// SPIRV-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("workgroup") monotonic, align 1 +// SPIRV-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP]], align 1 +// SPIRV-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi7c(_Bool *c) { return __scoped_atomic_exchange_n(c, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_WRKGRP); } -// AMDGCN-LABEL: define hidden zeroext i1 @fi7d( -// AMDGCN: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i8 [[VAL0:.+]] syncscope("wavefront") monotonic, align 1 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi7_clustr( +// AMDGCN_CL_DEF-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i8 1, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("cluster") monotonic, align 1, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi7_clustr( +// AMDGCN_CL_20-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i8 1, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("cluster") monotonic, align 1, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// +// SPIRV-LABEL: define hidden spir_func zeroext i1 @fi7_clustr( +// SPIRV-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i8 1, ptr [[DOTATOMICTMP]], align 1 +// SPIRV-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP]], align 1 +// SPIRV-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("workgroup") monotonic, align 1 +// SPIRV-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP]], align 1 +// SPIRV-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// +_Bool fi7_clustr(_Bool *c) { + return __scoped_atomic_exchange_n(c, 1, __ATOMIC_RELAXED, + __MEMORY_SCOPE_CLUSTR); +} + +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi7d( +// AMDGCN_CL_DEF-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i8 1, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("wavefront") monotonic, align 1, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi7d( +// AMDGCN_CL_20-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i8 1, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("wavefront") monotonic, align 1, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi7d( -// SPIRV: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i8 [[VAL0:.+]] syncscope("subgroup") monotonic, align 1 +// SPIRV-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i8 1, ptr [[DOTATOMICTMP]], align 1 +// SPIRV-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP]], align 1 +// SPIRV-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("subgroup") monotonic, align 1 +// SPIRV-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP]], align 1 +// SPIRV-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi7d(_Bool *c) { return __scoped_atomic_exchange_n(c, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_WVFRNT); } -// AMDGCN-LABEL: define hidden zeroext i1 @fi7e( -// AMDGCN: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i8 [[VAL0:.+]] syncscope("singlethread") monotonic, align 1 +// AMDGCN_CL_DEF-LABEL: define hidden zeroext i1 @fi7e( +// AMDGCN_CL_DEF-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_DEF-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_DEF-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_DEF-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_DEF-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_DEF-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_DEF-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_DEF-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_DEF-NEXT: store i8 1, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("singlethread") monotonic, align 1, !amdgpu.no.fine.grained.memory [[META3]], !amdgpu.no.remote.memory [[META3]] +// AMDGCN_CL_DEF-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_DEF-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// AMDGCN_CL_DEF-NEXT: ret i1 [[LOADEDV]] +// +// AMDGCN_CL_20-LABEL: define hidden zeroext i1 @fi7e( +// AMDGCN_CL_20-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// AMDGCN_CL_20-NEXT: [[ENTRY:.*:]] +// AMDGCN_CL_20-NEXT: [[RETVAL:%.*]] = alloca i1, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1, addrspace(5) +// AMDGCN_CL_20-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr +// AMDGCN_CL_20-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr +// AMDGCN_CL_20-NEXT: [[DOTATOMICTMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTATOMICTMP]] to ptr +// AMDGCN_CL_20-NEXT: [[ATOMIC_TEMP_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ATOMIC_TEMP]] to ptr +// AMDGCN_CL_20-NEXT: store ptr [[C]], ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR_ASCAST]], align 8 +// AMDGCN_CL_20-NEXT: store i8 1, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("singlethread") monotonic, align 1, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.no.remote.memory [[META4]] +// AMDGCN_CL_20-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP_ASCAST]], align 1 +// AMDGCN_CL_20-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// AMDGCN_CL_20-NEXT: ret i1 [[LOADEDV]] +// // SPIRV-LABEL: define hidden spir_func zeroext i1 @fi7e( -// SPIRV: [[TMP0:%.*]] = atomicrmw xchg ptr [[PTR0:%.+]], i8 [[VAL0:.+]] syncscope("singlethread") monotonic, align 1 +// SPIRV-SAME: ptr noundef [[C:%.*]]) #[[ATTR0]] { +// SPIRV-NEXT: [[ENTRY:.*:]] +// SPIRV-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 +// SPIRV-NEXT: [[DOTATOMICTMP:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1 +// SPIRV-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 +// SPIRV-NEXT: store i8 1, ptr [[DOTATOMICTMP]], align 1 +// SPIRV-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTATOMICTMP]], align 1 +// SPIRV-NEXT: [[TMP2:%.*]] = atomicrmw xchg ptr [[TMP0]], i8 [[TMP1]] syncscope("singlethread") monotonic, align 1 +// SPIRV-NEXT: store i8 [[TMP2]], ptr [[ATOMIC_TEMP]], align 1 +// SPIRV-NEXT: [[TMP3:%.*]] = load i8, ptr [[ATOMIC_TEMP]], align 1 +// SPIRV-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 +// SPIRV-NEXT: ret i1 [[LOADEDV]] +// _Bool fi7e(_Bool *c) { return __scoped_atomic_exchange_n(c, 1, __ATOMIC_RELAXED, __MEMORY_SCOPE_SINGLE); } +//. +// AMDGCN_CL_DEF: [[META3]] = !{} +//. +// AMDGCN_CL_20: [[META4]] = !{} +//. diff --git a/clang/test/CodeGen/scoped-fence-ops.c b/clang/test/CodeGen/scoped-fence-ops.c index d83ae05..259e8d33 100644 --- a/clang/test/CodeGen/scoped-fence-ops.c +++ b/clang/test/CodeGen/scoped-fence-ops.c @@ -1,8 +1,8 @@ // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 // RUN: %clang_cc1 %s -emit-llvm -o - -triple=amdgcn-amd-amdhsa -ffreestanding \ -// RUN: -fvisibility=hidden | FileCheck --check-prefix=AMDGCN %s +// RUN: -fvisibility=hidden | FileCheck --check-prefixes=AMDGCN,AMDGCN_CL_DEF %s // RUN: %clang_cc1 %s -emit-llvm -o - -triple=amdgcn-amd-amdhsa -ffreestanding \ -// RUN: -cl-std=CL2.0 -fvisibility=hidden | FileCheck --check-prefix=AMDGCN %s +// RUN: -cl-std=CL2.0 -fvisibility=hidden | FileCheck --check-prefixes=AMDGCN,AMDGCN_CL_20 %s // RUN: %clang_cc1 %s -emit-llvm -o - -triple=spirv64-unknown-unknown -ffreestanding \ // RUN: -fvisibility=hidden | FileCheck --check-prefix=SPIRV %s // RUN: %clang_cc1 %s -emit-llvm -o - -triple=x86_64-unknown-linux-gnu -ffreestanding \ @@ -127,23 +127,27 @@ void fe1b(int ord) { // AMDGCN-NEXT: store i32 [[SCOPE]], ptr [[SCOPE_ADDR_ASCAST]], align 4 // AMDGCN-NEXT: [[TMP0:%.*]] = load i32, ptr [[SCOPE_ADDR_ASCAST]], align 4 // AMDGCN-NEXT: switch i32 [[TMP0]], label %[[ATOMIC_SCOPE_CONTINUE:.*]] [ -// AMDGCN-NEXT: i32 1, label %[[DEVICE_SCOPE:.*]] // AMDGCN-NEXT: i32 0, label %[[SYSTEM_SCOPE:.*]] +// AMDGCN-NEXT: i32 1, label %[[DEVICE_SCOPE:.*]] // AMDGCN-NEXT: i32 2, label %[[WORKGROUP_SCOPE:.*]] +// AMDGCN-NEXT: i32 5, label %[[CLUSTER_SCOPE:.*]] // AMDGCN-NEXT: i32 3, label %[[WAVEFRONT_SCOPE:.*]] // AMDGCN-NEXT: i32 4, label %[[SINGLE_SCOPE:.*]] // AMDGCN-NEXT: ] // AMDGCN: [[ATOMIC_SCOPE_CONTINUE]]: // AMDGCN-NEXT: ret void -// AMDGCN: [[DEVICE_SCOPE]]: -// AMDGCN-NEXT: fence syncscope("agent") release -// AMDGCN-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] // AMDGCN: [[SYSTEM_SCOPE]]: // AMDGCN-NEXT: fence release // AMDGCN-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] +// AMDGCN: [[DEVICE_SCOPE]]: +// AMDGCN-NEXT: fence syncscope("agent") release +// AMDGCN-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] // AMDGCN: [[WORKGROUP_SCOPE]]: // AMDGCN-NEXT: fence syncscope("workgroup") release // AMDGCN-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] +// AMDGCN: [[CLUSTER_SCOPE]]: +// AMDGCN-NEXT: fence syncscope("cluster") release +// AMDGCN-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] // AMDGCN: [[WAVEFRONT_SCOPE]]: // AMDGCN-NEXT: fence syncscope("wavefront") release // AMDGCN-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] @@ -158,23 +162,27 @@ void fe1b(int ord) { // SPIRV-NEXT: store i32 [[SCOPE]], ptr [[SCOPE_ADDR]], align 4 // SPIRV-NEXT: [[TMP0:%.*]] = load i32, ptr [[SCOPE_ADDR]], align 4 // SPIRV-NEXT: switch i32 [[TMP0]], label %[[ATOMIC_SCOPE_CONTINUE:.*]] [ -// SPIRV-NEXT: i32 1, label %[[DEVICE_SCOPE:.*]] // SPIRV-NEXT: i32 0, label %[[SYSTEM_SCOPE:.*]] +// SPIRV-NEXT: i32 1, label %[[DEVICE_SCOPE:.*]] // SPIRV-NEXT: i32 2, label %[[WORKGROUP_SCOPE:.*]] +// SPIRV-NEXT: i32 5, label %[[CLUSTER_SCOPE:.*]] // SPIRV-NEXT: i32 3, label %[[WAVEFRONT_SCOPE:.*]] // SPIRV-NEXT: i32 4, label %[[SINGLE_SCOPE:.*]] // SPIRV-NEXT: ] // SPIRV: [[ATOMIC_SCOPE_CONTINUE]]: // SPIRV-NEXT: ret void -// SPIRV: [[DEVICE_SCOPE]]: -// SPIRV-NEXT: fence syncscope("device") release -// SPIRV-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] // SPIRV: [[SYSTEM_SCOPE]]: // SPIRV-NEXT: fence release // SPIRV-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] +// SPIRV: [[DEVICE_SCOPE]]: +// SPIRV-NEXT: fence syncscope("device") release +// SPIRV-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] // SPIRV: [[WORKGROUP_SCOPE]]: // SPIRV-NEXT: fence syncscope("workgroup") release // SPIRV-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] +// SPIRV: [[CLUSTER_SCOPE]]: +// SPIRV-NEXT: fence syncscope("workgroup") release +// SPIRV-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] // SPIRV: [[WAVEFRONT_SCOPE]]: // SPIRV-NEXT: fence syncscope("subgroup") release // SPIRV-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] @@ -189,23 +197,27 @@ void fe1b(int ord) { // X86_64-NEXT: store i32 [[SCOPE]], ptr [[SCOPE_ADDR]], align 4 // X86_64-NEXT: [[TMP0:%.*]] = load i32, ptr [[SCOPE_ADDR]], align 4 // X86_64-NEXT: switch i32 [[TMP0]], label %[[ATOMIC_SCOPE_CONTINUE:.*]] [ -// X86_64-NEXT: i32 1, label %[[DEVICE_SCOPE:.*]] // X86_64-NEXT: i32 0, label %[[SYSTEM_SCOPE:.*]] +// X86_64-NEXT: i32 1, label %[[DEVICE_SCOPE:.*]] // X86_64-NEXT: i32 2, label %[[WORKGROUP_SCOPE:.*]] +// X86_64-NEXT: i32 5, label %[[CLUSTER_SCOPE:.*]] // X86_64-NEXT: i32 3, label %[[WAVEFRONT_SCOPE:.*]] // X86_64-NEXT: i32 4, label %[[SINGLE_SCOPE:.*]] // X86_64-NEXT: ] // X86_64: [[ATOMIC_SCOPE_CONTINUE]]: // X86_64-NEXT: ret void -// X86_64: [[DEVICE_SCOPE]]: +// X86_64: [[SYSTEM_SCOPE]]: // X86_64-NEXT: fence release // X86_64-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] -// X86_64: [[SYSTEM_SCOPE]]: +// X86_64: [[DEVICE_SCOPE]]: // X86_64-NEXT: fence release // X86_64-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] // X86_64: [[WORKGROUP_SCOPE]]: // X86_64-NEXT: fence release // X86_64-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] +// X86_64: [[CLUSTER_SCOPE]]: +// X86_64-NEXT: fence release +// X86_64-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] // X86_64: [[WAVEFRONT_SCOPE]]: // X86_64-NEXT: fence release // X86_64-NEXT: br label %[[ATOMIC_SCOPE_CONTINUE]] @@ -257,3 +269,6 @@ void fe2a() { void fe2b() { __scoped_atomic_thread_fence(__ATOMIC_RELEASE, 999); } +//// NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line: +// AMDGCN_CL_20: {{.*}} +// AMDGCN_CL_DEF: {{.*}} diff --git a/clang/test/CodeGen/tbaa-class.cpp b/clang/test/CodeGen/tbaa-class.cpp index 0ac5908..4193e10 100644 --- a/clang/test/CodeGen/tbaa-class.cpp +++ b/clang/test/CodeGen/tbaa-class.cpp @@ -242,16 +242,16 @@ uint32_t g14(StructM2 *M, StructS *S) { return S->f16; } +// CHECK: [[TAG_i32]] = !{[[TYPE_i32:!.*]], [[TYPE_i32]], i64 0} +// CHECK: [[TYPE_i32]] = !{!"int", [[TYPE_char:!.*]], // CHECK: [[TYPE_char:!.*]] = !{!"omnipotent char", [[TAG_cxx_tbaa:!.*]], // CHECK: [[TAG_cxx_tbaa]] = !{!"Simple C++ TBAA"} -// CHECK: [[TAG_i32]] = !{[[TYPE_i32:!.*]], [[TYPE_i32]], i64 0} -// CHECK: [[TYPE_i32]] = !{!"int", [[TYPE_char]], // CHECK: [[TAG_i16]] = !{[[TYPE_i16:!.*]], [[TYPE_i16]], i64 0} // CHECK: [[TYPE_i16]] = !{!"short", [[TYPE_char]], -// OLD-PATH: [[TYPE_CHAR:!.*]] = !{!"omnipotent char", ! // OLD-PATH: [[TAG_i32]] = !{[[TYPE_INT:!.*]], [[TYPE_INT]], i64 0} -// OLD-PATH: [[TYPE_INT]] = !{!"int", [[TYPE_CHAR]] +// OLD-PATH: [[TYPE_INT]] = !{!"int", [[TYPE_CHAR:!.*]] +// OLD-PATH: [[TYPE_CHAR:!.*]] = !{!"omnipotent char", ! // OLD-PATH: [[TAG_A_f32]] = !{[[TYPE_A:!.*]], [[TYPE_INT]], i64 4} // OLD-PATH: [[TYPE_A]] = !{!"_ZTS7StructA", [[TYPE_SHORT:!.*]], i64 0, [[TYPE_INT]], i64 4, [[TYPE_SHORT]], i64 8, [[TYPE_INT]], i64 12} // OLD-PATH: [[TYPE_SHORT:!.*]] = !{!"short", [[TYPE_CHAR]] @@ -277,9 +277,9 @@ uint32_t g14(StructM2 *M, StructS *S) { // OLD-PATH: [[TYPE_M2]] = !{!"_ZTS8StructM2", [[TYPE_DYN:!.*]], i64 0, [[TYPE_S]], i64 12, [[TYPE_SHORT]], i64 20} // OLD_PATH: [[TYPE_DYN]] = !{!"_ZTS9StructDyn", [[TYPE_INT]], i64 8} +// NEW-PATH: [[TAG_i32]] = !{[[TYPE_INT:!.*]], [[TYPE_INT:!.*]], i64 0, i64 4} +// NEW-PATH: [[TYPE_INT]] = !{[[TYPE_CHAR:!.*]], i64 4, !"int"} // NEW-PATH: [[TYPE_CHAR:!.*]] = !{!{{.*}}, i64 1, !"omnipotent char"} -// NEW-PATH: [[TAG_i32]] = !{[[TYPE_INT:!.*]], [[TYPE_INT]], i64 0, i64 4} -// NEW-PATH: [[TYPE_INT]] = !{[[TYPE_CHAR]], i64 4, !"int"} // NEW-PATH: [[TAG_A_f32]] = !{[[TYPE_A:!.*]], [[TYPE_INT]], i64 4, i64 4} // NEW-PATH: [[TYPE_A]] = !{[[TYPE_CHAR]], i64 16, !"_ZTS7StructA", [[TYPE_SHORT:!.*]], i64 0, i64 2, [[TYPE_INT]], i64 4, i64 4, [[TYPE_SHORT]], i64 8, i64 2, [[TYPE_INT]], i64 12, i64 4} // NEW-PATH: [[TYPE_SHORT:!.*]] = !{[[TYPE_CHAR]], i64 2, !"short"} diff --git a/clang/test/CodeGen/tbaa-pointers.c b/clang/test/CodeGen/tbaa-pointers.c index 9cfaa0a..249cf56 100644 --- a/clang/test/CodeGen/tbaa-pointers.c +++ b/clang/test/CodeGen/tbaa-pointers.c @@ -1,18 +1,17 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 // RUN: %clang_cc1 -triple x86_64-apple-darwin -O1 -disable-llvm-passes -no-pointer-tbaa %s -emit-llvm -o - | FileCheck --check-prefixes=COMMON,DISABLE %s // RUN: %clang_cc1 -triple x86_64-apple-darwin -O1 -disable-llvm-passes %s -emit-llvm -o - | FileCheck --check-prefixes=COMMON,DEFAULT %s -// RUN: %clang --target=x86_64-apple-darwin -O1 -fno-pointer-tbaa %s -emit-llvm -S -mllvm -disable-llvm-optzns -o - | FileCheck --check-prefixes=COMMON,DISABLE %s -// RUN: %clang --target=x86_64-apple-darwin -O1 %s -emit-llvm -S -mllvm -disable-llvm-optzns -o - | FileCheck --check-prefixes=COMMON,DEFAULT %s void p2unsigned(unsigned **ptr) { // COMMON-LABEL: define void @p2unsigned( // COMMON-SAME: ptr noundef [[PTR:%.+]]) // COMMON: [[PTR_ADDR:%.+]] = alloca ptr, align 8 - // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[P2INT_0:!.+]] - // DEFAULT-NEXT: [[BASE:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[P2INT_0]] - // DEFAULT-NEXT: store ptr null, ptr [[BASE]], align 8, !tbaa [[P1INT_0:!.+]] - // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR:!.+]] - // DISABLE-NEXT: [[BASE:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: store ptr null, ptr [[BASE]], align 8, !tbaa [[ANYPTR]] + // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6:![0-9]+]] + // DEFAULT-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DEFAULT-NEXT: store ptr null, ptr [[TMP0]], align 8, !tbaa [[TBAA10:![0-9]+]] + // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6:![0-9]+]] + // DISABLE-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: store ptr null, ptr [[TMP0]], align 8, !tbaa [[TBAA6]] // COMMON-NEXT: ret void // *ptr = 0; @@ -22,12 +21,12 @@ void p2unsigned_volatile(unsigned *volatile *ptr) { // COMMON-LABEL: define void @p2unsigned_volatile( // COMMON-SAME: ptr noundef [[PTR:%.+]]) // COMMON: [[PTR_ADDR:%.+]] = alloca ptr, align 8 - // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[P2INT_0]] - // DEFAULT-NEXT: [[BASE:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[P2INT_0]] - // DEFAULT-NEXT: store volatile ptr null, ptr [[BASE]], align 8, !tbaa [[P1INT_0]] - // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: [[BASE:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: store volatile ptr null, ptr [[BASE]], align 8, !tbaa [[ANYPTR]] + // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DEFAULT-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DEFAULT-NEXT: store volatile ptr null, ptr [[TMP0]], align 8, !tbaa [[TBAA10]] + // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: store volatile ptr null, ptr [[TMP0]], align 8, !tbaa [[TBAA6]] // COMMON-NEXT: ret void // *ptr = 0; @@ -37,14 +36,14 @@ void p3int(int ***ptr) { // COMMON-LABEL: define void @p3int( // COMMON-SAME: ptr noundef [[PTR:%.+]]) // COMMON: [[PTR_ADDR:%.+]] = alloca ptr, align 8 - // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[P3INT_0:!.+]] - // DEFAULT-NEXT: [[BASE_0:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[P3INT_0]] - // DEFAULT-NEXT: [[BASE_1:%.+]] = load ptr, ptr [[BASE_0]], align 8, !tbaa [[P2INT_0]] - // DEFAULT-NEXT: store ptr null, ptr [[BASE_1]], align 8, !tbaa [[P1INT_0]] - // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: [[BASE_0:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: [[BASE_1:%.+]] = load ptr, ptr [[BASE_0]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: store ptr null, ptr [[BASE_1]], align 8, !tbaa [[ANYPTR]] + // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA12:![0-9]+]] + // DEFAULT-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA12]] + // DEFAULT-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA6]] + // DEFAULT-NEXT: store ptr null, ptr [[TMP1]], align 8, !tbaa [[TBAA10]] + // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: store ptr null, ptr [[TMP1]], align 8, !tbaa [[TBAA6]] // COMMON-NEXT: ret void // **ptr = 0; @@ -54,16 +53,16 @@ void p4char(char ****ptr) { // COMMON-LABEL: define void @p4char( // COMMON-SAME: ptr noundef [[PTR:%.+]]) // COMMON: [[PTR_ADDR:%.+]] = alloca ptr, align 8 - // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[P4CHAR_0:!.+]] - // DEFAULT-NEXT: [[BASE_0:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[P4CHAR_0]] - // DEFAULT-NEXT: [[BASE_1:%.+]] = load ptr, ptr [[BASE_0]], align 8, !tbaa [[P3CHAR_0:!.+]] - // DEFAULT-NEXT: [[BASE_2:%.+]] = load ptr, ptr [[BASE_1]], align 8, !tbaa [[P2CHAR_0:!.+]] - // DEFAULT-NEXT: store ptr null, ptr [[BASE_2]], align 8, !tbaa [[P1CHAR_0:!.+]] - // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: [[BASE_0:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: [[BASE_1:%.+]] = load ptr, ptr [[BASE_0]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: [[BASE_2:%.+]] = load ptr, ptr [[BASE_1]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: store ptr null, ptr [[BASE_2]], align 8, !tbaa [[ANYPTR]] + // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA15:![0-9]+]] + // DEFAULT-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA15]] + // DEFAULT-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA18:![0-9]+]] + // DEFAULT-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP1]], align 8, !tbaa [[TBAA20:![0-9]+]] + // DEFAULT-NEXT: store ptr null, ptr [[TMP2]], align 8, !tbaa [[TBAA22:![0-9]+]] + // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP1]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: store ptr null, ptr [[TMP2]], align 8, !tbaa [[TBAA6]] // COMMON-NEXT: ret void // ***ptr = 0; @@ -73,16 +72,16 @@ void p4char_const1(const char ****ptr) { // COMMON-LABEL: define void @p4char_const1( // COMMON-SAME: ptr noundef [[PTR:%.+]]) // COMMON: [[PTR_ADDR:%.+]] = alloca ptr, align 8 - // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[P4CHAR_0]] - // DEFAULT-NEXT: [[BASE_0:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[P4CHAR_0]] - // DEFAULT-NEXT: [[BASE_1:%.+]] = load ptr, ptr [[BASE_0]], align 8, !tbaa [[P3CHAR_0]] - // DEFAULT-NEXT: [[BASE_2:%.+]] = load ptr, ptr [[BASE_1]], align 8, !tbaa [[P2CHAR_0]] - // DEFAULT-NEXT: store ptr null, ptr [[BASE_2]], align 8, !tbaa [[P1CHAR_0]] - // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: [[BASE_0:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: [[BASE_1:%.+]] = load ptr, ptr [[BASE_0]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: [[BASE_2:%.+]] = load ptr, ptr [[BASE_1]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: store ptr null, ptr [[BASE_2]], align 8, !tbaa [[ANYPTR]] + // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA15]] + // DEFAULT-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA15]] + // DEFAULT-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA18]] + // DEFAULT-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP1]], align 8, !tbaa [[TBAA20]] + // DEFAULT-NEXT: store ptr null, ptr [[TMP2]], align 8, !tbaa [[TBAA22]] + // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP1]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: store ptr null, ptr [[TMP2]], align 8, !tbaa [[TBAA6]] // COMMON-NEXT: ret void // ***ptr = 0; @@ -92,16 +91,16 @@ void p4char_const2(const char **const **ptr) { // COMMON-LABEL: define void @p4char_const2( // COMMON-SAME: ptr noundef [[PTR:%.+]]) // COMMON: [[PTR_ADDR:%.+]] = alloca ptr, align 8 - // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[P4CHAR_0]] - // DEFAULT-NEXT: [[BASE_0:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[P4CHAR_0]] - // DEFAULT-NEXT: [[BASE_1:%.+]] = load ptr, ptr [[BASE_0]], align 8, !tbaa [[P3CHAR_0]] - // DEFAULT-NEXT: [[BASE_2:%.+]] = load ptr, ptr [[BASE_1]], align 8, !tbaa [[P2CHAR_0]] - // DEFAULT-NEXT: store ptr null, ptr [[BASE_2]], align 8, !tbaa [[P1CHAR_0]] - // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: [[BASE_0:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: [[BASE_1:%.+]] = load ptr, ptr [[BASE_0]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: [[BASE_2:%.+]] = load ptr, ptr [[BASE_1]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: store ptr null, ptr [[BASE_2]], align 8, !tbaa [[ANYPTR]] + // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA15]] + // DEFAULT-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA15]] + // DEFAULT-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA18]] + // DEFAULT-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP1]], align 8, !tbaa [[TBAA20]] + // DEFAULT-NEXT: store ptr null, ptr [[TMP2]], align 8, !tbaa [[TBAA22]] + // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP1]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: store ptr null, ptr [[TMP2]], align 8, !tbaa [[TBAA6]] // COMMON-NEXT: ret void // ***ptr = 0; @@ -116,12 +115,12 @@ void p2struct(struct S1 **ptr) { // COMMON-LABEL: define void @p2struct( // COMMON-SAME: ptr noundef [[PTR:%.+]]) // COMMON: [[PTR_ADDR:%.+]] = alloca ptr, align 8 - // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[P2S1_TAG:!.+]] - // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DEFAULT-NEXT: [[BASE:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[P2S1_TAG]] - // DISABLE-NEXT: [[BASE:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DEFAULT-NEXT: store ptr null, ptr [[BASE]], align 8, !tbaa [[P1S1_TAG:!.+]] - // DISABLE-NEXT: store ptr null, ptr [[BASE]], align 8, !tbaa [[ANYPTR]] + // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA24:![0-9]+]] + // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DEFAULT-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA24]] + // DISABLE-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DEFAULT-NEXT: store ptr null, ptr [[TMP0]], align 8, !tbaa [[TBAA26:![0-9]+]] + // DISABLE-NEXT: store ptr null, ptr [[TMP0]], align 8, !tbaa [[TBAA6]] // COMMON-NEXT: ret void // *ptr = 0; @@ -131,10 +130,12 @@ void p2struct_const(struct S1 const **ptr) { // COMMON-LABEL: define void @p2struct_const( // COMMON-SAME: ptr noundef [[PTR:%.+]]) // COMMON: [[PTR_ADDR:%.+]] = alloca ptr, align 8 - // COMMON-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR:!.+]] - // COMMON-NEXT: [[BASE:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DEFAULT-NEXT: store ptr null, ptr [[BASE]], align 8, !tbaa [[P1S1_TAG]] - // DISABLE-NEXT: store ptr null, ptr [[BASE]], align 8, !tbaa [[ANYPTR]] + // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA24]] + // DEFAULT-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA24]] + // DEFAULT-NEXT: store ptr null, ptr [[TMP0]], align 8, !tbaa [[TBAA26]] + // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: store ptr null, ptr [[TMP0]], align 8, !tbaa [[TBAA6]] // COMMON-NEXT: ret void // *ptr = 0; @@ -148,14 +149,14 @@ void p2struct2(struct S2 *ptr) { // COMMON-LABEL: define void @p2struct2( // COMMON-SAME: ptr noundef [[PTR:%.+]]) // COMMON: [[PTR_ADDR:%.+]] = alloca ptr, align 8 - // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[P1S2_TAG:!.+]] - // DEFAULT-NEXT: [[BASE:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[P1S2_TAG]] - // DEFAULT-NEXT: [[S:%.+]] = getelementptr inbounds nuw %struct.S2, ptr [[BASE]], i32 0, i32 0 - // DEFAULT-NEXT: store ptr null, ptr [[S]], align 8, !tbaa [[S2_S_TAG:!.+]] - // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: [[BASE:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] - // DISABLE-NEXT: [[S:%.+]] = getelementptr inbounds nuw %struct.S2, ptr [[BASE]], i32 0, i32 0 - // DISABLE-NEXT: store ptr null, ptr [[S]], align 8, !tbaa [[S2_S_TAG:!.+]] + // DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA28:![0-9]+]] + // DEFAULT-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA28]] + // DEFAULT-NEXT: [[S:%.*]] = getelementptr inbounds nuw [[STRUCT_S2:%.*]], ptr [[TMP0]], i32 0, i32 0 + // DEFAULT-NEXT: store ptr null, ptr [[S]], align 8, !tbaa [[TBAA30:![0-9]+]] + // DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] + // DISABLE-NEXT: [[S:%.*]] = getelementptr inbounds nuw [[STRUCT_S2:%.*]], ptr [[TMP0]], i32 0, i32 0 + // DISABLE-NEXT: store ptr null, ptr [[S]], align 8, !tbaa [[TBAA8:![0-9]+]] // COMMON-NEXT: ret void ptr->s = 0; } @@ -167,21 +168,21 @@ void vla1(int n, int ptr[][n], int idx) { // COMMON: [[N_ADDR:%.+]] = alloca i32, align 4 // COMMON-NEXT: [[PTR_ADDR:%.+]] = alloca ptr, align 8 // COMMON-NEXT: [[IDX_ADDR:%.+]] = alloca i32, align 4 -// COMMON-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4, !tbaa [[INT_TY:!.+]] -// DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[P1INT0:!.+]] -// DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] -// COMMON-NEXT: store i32 [[IDX]], ptr [[IDX_ADDR]], align 4, !tbaa [[INT_TY]] -// COMMON-NEXT: [[L:%.+]] = load i32, ptr [[N_ADDR]], align 4, !tbaa [[INT_TY]] -// COMMON-NEXT: [[L_EXT:%.+]] = zext i32 [[L]] to i64 -// DEFAULT-NEXT: [[L_PTR:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[P1INT0]] -// DISABLE-NEXT: [[L_PTR:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] -// COMMON-NEXT: [[L_IDX:%.+]] = load i32, ptr [[IDX_ADDR]], align 4, !tbaa [[INT_TY]] -// COMMON-NEXT: [[IDX_EXT:%.+]] = sext i32 [[L_IDX]] to i64 -// COMMON-NEXT: [[MUL:%.+]] = mul nsw i64 [[IDX_EXT]], [[L_EXT]] -// COMMON-NEXT: [[GEP1:%.+]] = getelementptr inbounds i32, ptr [[L_PTR]], i64 [[MUL]] -// COMMON-NEXT: [[GEP2:%.+]] = getelementptr inbounds i32, ptr [[GEP1]], i64 0 -// COMMON-NEXT: store i32 0, ptr [[GEP2]], align 4, !tbaa [[INT_TAG:!.+]] -// DEFAULT-NEXT: ret void +// COMMON-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4, !tbaa [[TBAA2:![0-9]+]] +// DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA10]] +// DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] +// COMMON-NEXT: store i32 [[IDX]], ptr [[IDX_ADDR]], align 4, !tbaa [[TBAA2]] +// COMMON-NEXT: [[TMP0:%.*]] = load i32, ptr [[N_ADDR]], align 4, !tbaa [[TBAA2]] +// COMMON-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 +// DEFAULT-NEXT: [[TMP2:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA10]] +// DISABLE-NEXT: [[TMP2:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] +// COMMON-NEXT: [[TMP3:%.*]] = load i32, ptr [[IDX_ADDR]], align 4, !tbaa [[TBAA2]] +// COMMON-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP3]] to i64 +// COMMON-NEXT: [[TMP4:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP1]] +// COMMON-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP4]] +// COMMON-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 0 +// COMMON-NEXT: store i32 0, ptr [[ARRAYIDX1]], align 4, !tbaa [[TBAA2]] +// COMMON-NEXT: ret void ptr[idx][0] = 0; } @@ -194,11 +195,13 @@ void unamed_struct_typedef(TypedefS *ptr) { // COMMON-LABEL: define void @unamed_struct_typedef( // COMMON-SAME: ptr noundef [[PTRA:%.+]]) // COMMON: [[PTR_ADDR:%.+]] = alloca ptr, align 8 -// DISABLE-NEXT: store ptr [[PTRA]], ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] -// DEFAULT-NEXT: store ptr [[PTRA]], ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR:!.+]] -// COMMON-NEXT: [[L0:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] -// COMMON-NEXT: [[GEP:%.+]] = getelementptr inbounds nuw %struct.TypedefS, ptr [[L0]], i32 0, i32 0 -// COMMON-NEXT: store i32 0, ptr [[GEP]], align 4 +// DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] +// DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA32:![0-9]+]] +// DISABLE-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] +// DEFAULT-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA32]] +// COMMON-NEXT: [[I1:%.*]] = getelementptr inbounds nuw [[STRUCT_TYPEDEFS:%.*]], ptr [[TMP0]], i32 0, i32 0 +// DISABLE-NEXT: store i32 0, ptr [[I1]], align 4, !tbaa [[TBAA10:![0-9]+]] +// DEFAULT-NEXT: store i32 0, ptr [[I1]], align 4, !tbaa [[TBAA33:![0-9]+]] // COMMON-NEXT: ret void ptr->i1 = 0; @@ -208,52 +211,62 @@ int void_ptrs(void **ptr) { // COMMON-LABEL: define i32 @void_ptrs( // COMMON-SAME: ptr noundef [[PTRA:%.+]]) // COMMON: [[PTR_ADDR:%.+]] = alloca ptr, align 8 -// DISABLE-NEXT: store ptr [[PTRA]], ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] -// DEFAULT-NEXT: store ptr [[PTRA]], ptr [[PTR_ADDR]], align 8, !tbaa [[ANYP2:!.+]] -// DISABLE-NEXT: [[L0:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[ANYPTR]] -// DEFAULT-NEXT: [[L0:%.+]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[ANYP2]] -// COMMON-NEXT: [[L1:%.+]] = load ptr, ptr [[L0]], align 8, !tbaa [[ANYPTR]] -// COMMON-NEXT: [[BOOL:%.+]] = icmp ne ptr [[L1]], null -// COMMON-NEXT: [[BOOL_EXT:%.+]] = zext i1 [[BOOL]] to i64 -// COMMON-NEXT: [[COND:%.+]] = select i1 [[BOOL]], i32 0, i32 1 +// DISABLE-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] +// DEFAULT-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA35:![0-9]+]] +// DISABLE-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA6]] +// DEFAULT-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR]], align 8, !tbaa [[TBAA35]] +// DISABLE-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA6]] +// DEFAULT-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA32]] +// COMMON-NEXT: [[TOBOOL:%.*]] = icmp ne ptr [[TMP1]], null +// COMMON-NEXT: [[TMP2:%.*]] = zext i1 [[TOBOOL]] to i64 +// COMMON-NEXT: [[COND:%.*]] = select i1 [[TOBOOL]], i32 0, i32 1 // COMMON-NEXT: ret i32 [[COND]] return *ptr ? 0 : 1; } -// DEFAULT: [[P2INT_0]] = !{[[P2INT:!.+]], [[P2INT]], i64 0} -// DEFAULT: [[P2INT]] = !{!"p2 int", [[ANY_P2_POINTER:!.+]], i64 0} -// DEFAULT: [[ANY_P2_POINTER]] = !{!"any p2 pointer", [[ANY_POINTER:!.+]], i64 0} -// DISABLE: [[ANYPTR]] = !{[[ANY_POINTER:!.+]], [[ANY_POINTER]], i64 0} -// COMMON: [[ANY_POINTER]] = !{!"any pointer", [[CHAR:!.+]], i64 0} -// COMMON: [[CHAR]] = !{!"omnipotent char", [[TBAA_ROOT:!.+]], i64 0} -// COMMON: [[TBAA_ROOT]] = !{!"Simple C/C++ TBAA"} -// DEFAULT: [[P1INT_0]] = !{[[P1INT:!.+]], [[P1INT]], i64 0} -// DEFAULT: [[P1INT]] = !{!"p1 int", [[ANY_POINTER]], i64 0} -// DEFAULT: [[P3INT_0]] = !{[[P3INT:!.+]], [[P3INT]], i64 0} -// DEFAULT: [[P3INT]] = !{!"p3 int", [[ANY_P3_POINTER:!.+]], i64 0} -// DEFAULT: [[ANY_P3_POINTER]] = !{!"any p3 pointer", [[ANY_P2_POINTER]], i64 0} -// DEFAULT: [[P4CHAR_0]] = !{[[P4CHAR:!.+]], [[P4CHAR]], i64 0} -// DEFAULT: [[P4CHAR]] = !{!"p4 omnipotent char", [[ANY_P4_POINTER:!.*]], i64 0} -// DEFAULT: [[ANY_P4_POINTER]] = !{!"any p4 pointer", [[ANY_P3_POINTER]], i64 0} -// DEFAULT: [[P3CHAR_0]] = !{[[P3CHAR:!.+]], [[P3CHAR]], i64 0} -// DEFAULT: [[P3CHAR]] = !{!"p3 omnipotent char", [[ANY_P3_POINTER]], i64 0} -// DEFAULT: [[P2CHAR_0]] = !{[[P2CHAR:!.+]], [[P2CHAR]], i64 0} -// DEFAULT: [[P2CHAR]] = !{!"p2 omnipotent char", [[ANY_P2_POINTER]], i64 0} -// DEFAULT: [[P1CHAR_0]] = !{[[P1CHAR:!.+]], [[P1CHAR]], i64 0} -// DEFAULT: [[P1CHAR]] = !{!"p1 omnipotent char", [[ANY_POINTER]], i64 0} -// DEFAULT: [[P2S1_TAG]] = !{[[P2S1:!.+]], [[P2S1]], i64 0} -// DEFAULT: [[P2S1]] = !{!"p2 _ZTS2S1", [[ANY_P2_POINTER]], i64 0} -// DEFAULT: [[P1S1_TAG:!.+]] = !{[[P1S1:!.+]], [[P1S1]], i64 0} -// DEFAULT: [[P1S1]] = !{!"p1 _ZTS2S1", [[ANY_POINTER]], i64 0} -// DEFAULT: [[P1S2_TAG]] = !{[[P1S2:!.+]], [[P1S2]], i64 0} -// DEFAULT: [[P1S2]] = !{!"p1 _ZTS2S2", [[ANY_POINTER]], i64 0} - -// DEFAULT: [[S2_S_TAG]] = !{[[S2_TY:!.+]], [[P1S1]], i64 0} -// DEFAULT: [[S2_TY]] = !{!"S2", [[P1S1]], i64 0} -// DISABLE: [[S2_S_TAG]] = !{[[S2_TY:!.+]], [[ANY_POINTER]], i64 0} -// DISABLE: [[S2_TY]] = !{!"S2", [[ANY_POINTER]], i64 0} -// COMMON: [[INT_TAG]] = !{[[INT_TY:!.+]], [[INT_TY]], i64 0} -// COMMON: [[INT_TY]] = !{!"int", [[CHAR]], i64 0} -// DEFAULT: [[ANYPTR]] = !{[[ANY_POINTER]], [[ANY_POINTER]], i64 0} -// DEFAULT: [[ANYP2]] = !{[[ANY_P2_POINTER]], [[ANY_P2_POINTER]], i64 0} +// DISABLE: [[TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// DISABLE: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// DISABLE: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// DISABLE: [[META5]] = !{!"Simple C/C++ TBAA"} +// DISABLE: [[TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// DISABLE: [[META7]] = !{!"any pointer", [[META4]], i64 0} +// DISABLE: [[TBAA8]] = !{[[META9:![0-9]+]], [[META7]], i64 0} +// DISABLE: [[META9]] = !{!"S2", [[META7]], i64 0} +// DISABLE: [[TBAA10]] = !{[[META11:![0-9]+]], [[META3]], i64 0} +// DISABLE: [[META11]] = !{!"", [[META3]], i64 0} +// DEFAULT: [[TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// DEFAULT: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// DEFAULT: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// DEFAULT: [[META5]] = !{!"Simple C/C++ TBAA"} +// DEFAULT: [[TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// DEFAULT: [[META7]] = !{!"p2 int", [[META8:![0-9]+]], i64 0} +// DEFAULT: [[META8]] = !{!"any p2 pointer", [[META9:![0-9]+]], i64 0} +// DEFAULT: [[META9]] = !{!"any pointer", [[META4]], i64 0} +// DEFAULT: [[TBAA10]] = !{[[META11:![0-9]+]], [[META11]], i64 0} +// DEFAULT: [[META11]] = !{!"p1 int", [[META9]], i64 0} +// DEFAULT: [[TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} +// DEFAULT: [[META13]] = !{!"p3 int", [[META14:![0-9]+]], i64 0} +// DEFAULT: [[META14]] = !{!"any p3 pointer", [[META8]], i64 0} +// DEFAULT: [[TBAA15]] = !{[[META16:![0-9]+]], [[META16]], i64 0} +// DEFAULT: [[META16]] = !{!"p4 omnipotent char", [[META17:![0-9]+]], i64 0} +// DEFAULT: [[META17]] = !{!"any p4 pointer", [[META14]], i64 0} +// DEFAULT: [[TBAA18]] = !{[[META19:![0-9]+]], [[META19]], i64 0} +// DEFAULT: [[META19]] = !{!"p3 omnipotent char", [[META14]], i64 0} +// DEFAULT: [[TBAA20]] = !{[[META21:![0-9]+]], [[META21]], i64 0} +// DEFAULT: [[META21]] = !{!"p2 omnipotent char", [[META8]], i64 0} +// DEFAULT: [[TBAA22]] = !{[[META23:![0-9]+]], [[META23]], i64 0} +// DEFAULT: [[META23]] = !{!"p1 omnipotent char", [[META9]], i64 0} +// DEFAULT: [[TBAA24]] = !{[[META25:![0-9]+]], [[META25]], i64 0} +// DEFAULT: [[META25]] = !{!"p2 _ZTS2S1", [[META8]], i64 0} +// DEFAULT: [[TBAA26]] = !{[[META27:![0-9]+]], [[META27]], i64 0} +// DEFAULT: [[META27]] = !{!"p1 _ZTS2S1", [[META9]], i64 0} +// DEFAULT: [[TBAA28]] = !{[[META29:![0-9]+]], [[META29]], i64 0} +// DEFAULT: [[META29]] = !{!"p1 _ZTS2S2", [[META9]], i64 0} +// DEFAULT: [[TBAA30]] = !{[[META31:![0-9]+]], [[META27]], i64 0} +// DEFAULT: [[META31]] = !{!"S2", [[META27]], i64 0} +// DEFAULT: [[TBAA32]] = !{[[META9]], [[META9]], i64 0} +// DEFAULT: [[TBAA33]] = !{[[META34:![0-9]+]], [[META3]], i64 0} +// DEFAULT: [[META34]] = !{!"", [[META3]], i64 0} +// DEFAULT: [[TBAA35]] = !{[[META8]], [[META8]], i64 0} +//. diff --git a/clang/test/CodeGen/tbaa-struct-bitfield-endianness.cpp b/clang/test/CodeGen/tbaa-struct-bitfield-endianness.cpp index 1177691..0775f99 100644 --- a/clang/test/CodeGen/tbaa-struct-bitfield-endianness.cpp +++ b/clang/test/CodeGen/tbaa-struct-bitfield-endianness.cpp @@ -18,18 +18,18 @@ struct NamedBitfields { // CHECK-LABEL: _Z4copyP14NamedBitfieldsS0_ // CHECK-SAME: ptr noundef writeonly captures(none) initializes((0, 16)) [[A1:%.*]], ptr noundef readonly captures(none) [[A2:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: entry: -// CHECK-NEXT: tail call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 8 dereferenceable(16) [[A1]], ptr noundef nonnull align 8 dereferenceable(16) [[A2]], i64 16, i1 false), !tbaa.struct [[TBAA_STRUCT2:![0-9]+]] +// CHECK-NEXT: tail call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 8 dereferenceable(16) [[A1]], ptr noundef nonnull align 8 dereferenceable(16) [[A2]], i64 16, i1 false), !tbaa.struct [[TBAA_STRUCT6:![0-9]+]] // CHECK-NEXT: ret void // void copy(NamedBitfields *a1, NamedBitfields *a2) { *a1 = *a2; } -// CHECK: [[TBAA_STRUCT2]] = !{i64 0, i64 4, [[META3:![0-9]+]], i64 4, i64 4, [[META6:![0-9]+]], i64 8, i64 8, [[META8:![0-9]+]]} -// CHECK: [[META3]] = !{[[META4:![0-9]+]], [[META4]], i64 0} +// CHECK: [[META2:![0-9]+]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// CHECK: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} // CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK: [[META5]] = !{!"Simple C++ TBAA"} -// CHECK: [[META6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK: [[META7]] = !{!"int", [[META4]], i64 0} +// CHECK: [[TBAA_STRUCT6]] = !{i64 0, i64 4, [[META7:![0-9]+]], i64 4, i64 4, [[META2:![0-9]+]], i64 8, i64 8, [[META8:![0-9]+]]} +// CHECK: [[META7]] = !{[[META4:![0-9]+]], [[META4]], i64 0} // CHECK: [[META8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} // CHECK: [[META9]] = !{!"double", [[META4]], i64 0} diff --git a/clang/test/CodeGen/tbaa-struct.cpp b/clang/test/CodeGen/tbaa-struct.cpp index ca076ce..48f8492 100644 --- a/clang/test/CodeGen/tbaa-struct.cpp +++ b/clang/test/CodeGen/tbaa-struct.cpp @@ -1,3 +1,4 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 // RUN: %clang_cc1 -triple x86_64-apple-darwin -emit-llvm -o - -O1 %s | \ // RUN: FileCheck -check-prefixes=CHECK,CHECK-OLD %s // RUN: %clang_cc1 -triple x86_64-apple-darwin -new-struct-path-tbaa \ @@ -18,9 +19,9 @@ typedef A __attribute__((may_alias)) AA; void copy(A *a1, A *a2) { // CHECK-LABEL: _Z4copyP1AS0_ // CHECK: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 4 dereferenceable(16) %{{.*}}, ptr noundef nonnull align 4 dereferenceable(16) %{{.*}}, i64 16, i1 false) -// CHECK-OLD-SAME: !tbaa.struct [[TS:!.*]] -// CHECK-NEW-SAME: !tbaa [[TAG_A:![0-9]*]] - *a1 = *a2; +// CHECK-OLD-SAME: !tbaa.struct [[TBAA_STRUCT6:![0-9]+]] +// CHECK-NEW-SAME: !tbaa [[TBAA6:![0-9]+]] +*a1 = *a2; } struct B { @@ -30,10 +31,10 @@ struct B { }; void copy2(B *b1, B *b2) { -// CHECK-LABEL: _Z5copy2P1BS0_ +// CHECK-LABEL: _Z5copy2P1BS0_ // CHECK: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 4 dereferenceable(24) %{{.*}}, ptr noundef nonnull align 4 dereferenceable(24) %{{.*}}, i64 24, i1 false) -// CHECK-OLD-SAME: !tbaa.struct [[TS2:!.*]] -// CHECK-NEW-SAME: !tbaa [[TAG_B:![0-9]*]] +// CHECK-OLD-SAME: !tbaa.struct [[TBAA_STRUCT10:![0-9]+]] +// CHECK-NEW-SAME: !tbaa [[TBAA12:![0-9]+]] *b1 = *b2; } @@ -50,9 +51,9 @@ union U { void copy3(U *u1, U *u2) { // CHECK-LABEL: _Z5copy3P1US0_ // CHECK: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 4 dereferenceable(12) %{{.*}}, ptr noundef nonnull align 4 dereferenceable(12) %{{.*}}, i64 12, i1 false) -// CHECK-OLD-SAME: !tbaa.struct [[TS3:!.*]] -// CHECK-NEW-SAME: !tbaa [[TAG_U:![0-9]*]] - *u1 = *u2; +// CHECK-OLD-SAME: !tbaa.struct [[TBAA_STRUCT11:![0-9]+]] +// CHECK-NEW-SAME: !tbaa [[TBAA15:![0-9]+]] +*u1 = *u2; } // Make sure that zero-length bitfield works. @@ -66,8 +67,8 @@ struct C { void copy4(C *c1, C *c2) { // CHECK-LABEL: _Z5copy4P1CS0_ // CHECK: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 1 dereferenceable(3) {{.*}}, ptr noundef nonnull align 1 dereferenceable(3) {{.*}}, i64 3, i1 false) -// CHECK-OLD-SAME: !tbaa.struct [[TS4:!.*]] -// CHECK-NEW-SAME: !tbaa [[TAG_C:![0-9]*]] +// CHECK-OLD-SAME: !tbaa.struct [[TBAA_STRUCT12:![0-9]+]] +// CHECK-NEW-SAME: !tbaa [[TBAA17:![0-9]+]] *c1 = *c2; } @@ -81,24 +82,24 @@ struct D { void copy5(D *d1, D *d2) { // CHECK-LABEL: _Z5copy5P1DS0_ // CHECK: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 1 dereferenceable(6) {{.*}}, ptr noundef nonnull align 1 dereferenceable(6) {{.*}}, i64 6, i1 false) -// CHECK-OLD-SAME: !tbaa.struct [[TS5:!.*]] -// CHECK-NEW-SAME: !tbaa [[TAG_D:![0-9]*]] +// CHECK-OLD-SAME: !tbaa.struct [[TBAA_STRUCT13:![0-9]+]] +// CHECK-NEW-SAME: !tbaa [[TBAA20:![0-9]+]] *d1 = *d2; } void copy6(AA *a1, A *a2) { // CHECK-LABEL: _Z5copy6P1AS0_ // CHECK: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 4 dereferenceable(16) %{{.*}}, ptr noundef nonnull align 4 dereferenceable(16) %{{.*}}, i64 16, i1 false) -// CHECK-OLD-SAME: !tbaa.struct [[TS]] -// CHECK-NEW-SAME: !tbaa [[TAG_char:![0-9]*]] +// CHECK-OLD-SAME: !tbaa.struct [[TBAA_STRUCT6]] +// CHECK-NEW-SAME: !tbaa [[TBAA23:![0-9]+]] *a1 = *a2; } void copy7(A *a1, AA *a2) { // CHECK-LABEL: _Z5copy7P1AS0_ // CHECK: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 4 dereferenceable(16) %{{.*}}, ptr noundef nonnull align 4 dereferenceable(16) %{{.*}}, i64 16, i1 false) -// CHECK-OLD-SAME: !tbaa.struct [[TS]] -// CHECK-NEW-SAME: !tbaa [[TAG_char]] +// CHECK-OLD-SAME: !tbaa.struct [[TBAA_STRUCT6]] +// CHECK-NEW-SAME: !tbaa [[TBAA23]] *a1 = *a2; } @@ -112,8 +113,8 @@ struct NamedBitfields { void copy8(NamedBitfields *a1, NamedBitfields *a2) { // CHECK-LABEL: _Z5copy8P14NamedBitfieldsS0_ // CHECK: tail call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 8 dereferenceable(16) %a1, ptr noundef nonnull align 8 dereferenceable(16) %a2, i64 16, i1 false), -// CHECK-OLD-SAME: !tbaa.struct [[TS6:!.*]] -// CHECK-NEW-SAME: !tbaa [[TAG_NamedBitfields:!.+]], !tbaa.struct +// CHECK-OLD-SAME: !tbaa.struct [[TBAA_STRUCT14:![0-9]+]] +// CHECK-NEW-SAME: !tbaa [[TBAA24:![0-9]+]], !tbaa.struct *a1 = *a2; } @@ -129,8 +130,8 @@ struct NamedBitfields2 { void copy9(NamedBitfields2 *a1, NamedBitfields2 *a2) { // CHECK-LABEL: _Z5copy9P15NamedBitfields2S0_ // CHECK: tail call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 8 dereferenceable(24) %a1, ptr noundef nonnull align 8 dereferenceable(24) %a2, i64 24, i1 false), -// CHECK-OLD-SAME: !tbaa.struct [[TS7:!.*]] -// CHECK-NEW-SAME: !tbaa [[TAG_NamedBitfields2:!.+]], !tbaa.struct +// CHECK-OLD-SAME: !tbaa.struct [[TBAA_STRUCT17:![0-9]+]] +// CHECK-NEW-SAME: !tbaa [[TBAA30:![0-9]+]], !tbaa.struct *a1 = *a2; } @@ -146,8 +147,8 @@ struct NamedBitfields3 { void copy10(NamedBitfields3 *a1, NamedBitfields3 *a2) { // CHECK-LABEL: _Z6copy10P15NamedBitfields3S0_ // CHECK: tail call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 8 dereferenceable(16) %a1, ptr noundef nonnull align 8 dereferenceable(16) %a2, i64 16, i1 false), -// CHECK-OLD-SAME: !tbaa.struct [[TS8:!.*]] -// CHECK-NEW-SAME: !tbaa [[TAG_NamedBitfields3:!.+]], !tbaa.struct +// CHECK-OLD-SAME: !tbaa.struct [[TBAA_STRUCT18:![0-9]+]] +// CHECK-NEW-SAME: !tbaa [[TBAA33:![0-9]+]], !tbaa.struct *a1 = *a2; } @@ -164,8 +165,8 @@ struct UnionMember1 { void copy11(UnionMember1 *a1, UnionMember1 *a2) { // CHECK-LABEL: _Z6copy11P12UnionMember1S0_ // CHECK: tail call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 8 dereferenceable(16) %a1, ptr noundef nonnull align 8 dereferenceable(16) %a2, i64 16, i1 false), -// CHECK-OLD-SAME: !tbaa.struct [[TS9:!.*]] -// CHECK-NEW-SAME: !tbaa [[TAG_UnionMember1:!.+]], !tbaa.struct +// CHECK-OLD-SAME: !tbaa.struct [[TBAA_STRUCT19:![0-9]+]] +// CHECK-NEW-SAME: !tbaa [[TBAA37:![0-9]+]], !tbaa.struct *a1 = *a2; } @@ -177,52 +178,61 @@ struct UnionMember2 { void copy12(UnionMember2 *a1, UnionMember2 *a2) { // CHECK-LABEL: _Z6copy12P12UnionMember2S0_ // CHECK: tail call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 8 dereferenceable(16) %a1, ptr noundef nonnull align 8 dereferenceable(16) %a2, i64 16, i1 false), -// CHECK-OLD-SAME: !tbaa.struct [[TS10:!.*]] -// CHECK-NEW-SAME: !tbaa [[TAG_UnionMember2:!.+]], !tbaa.struct +// CHECK-OLD-SAME: !tbaa.struct [[TBAA_STRUCT20:![0-9]+]] +// CHECK-NEW-SAME: !tbaa [[TBAA41:![0-9]+]], !tbaa.struct *a1 = *a2; } -// CHECK-OLD: [[TS]] = !{i64 0, i64 2, !{{.*}}, i64 4, i64 4, !{{.*}}, i64 8, i64 1, !{{.*}}, i64 12, i64 4, !{{.*}}} -// CHECK-OLD: [[CHAR:!.*]] = !{!"omnipotent char", !{{.*}}} -// CHECK-OLD: [[TAG_INT:!.*]] = !{[[INT:!.*]], [[INT]], i64 0} -// CHECK-OLD: [[INT]] = !{!"int", [[CHAR]] -// CHECK-OLD: [[TAG_CHAR:!.*]] = !{[[CHAR]], [[CHAR]], i64 0} // (offset, size) = (0,1) char; (4,2) short; (8,4) int; (12,1) char; (16,4) int; (20,4) int -// CHECK-OLD: [[TS2]] = !{i64 0, i64 1, !{{.*}}, i64 4, i64 2, !{{.*}}, i64 8, i64 4, !{{.*}}, i64 12, i64 1, !{{.*}}, i64 16, i64 4, {{.*}}, i64 20, i64 4, {{.*}}} // (offset, size) = (0,8) char; (0,2) char; (4,8) char -// CHECK-OLD: [[TS3]] = !{i64 0, i64 12, [[TAG_CHAR]]} -// CHECK-OLD: [[TS4]] = !{i64 0, i64 1, [[TAG_CHAR]], i64 1, i64 1, [[TAG_CHAR]], i64 2, i64 1, [[TAG_CHAR]]} -// CHECK-OLD: [[TS5]] = !{i64 0, i64 1, [[TAG_CHAR]], i64 4, i64 1, [[TAG_CHAR]], i64 5, i64 1, [[TAG_CHAR]]} -// CHECK-OLD: [[TS6]] = !{i64 0, i64 2, [[TAG_CHAR]], i64 2, i64 1, [[TAG_CHAR]], i64 8, i64 8, [[TAG_DOUBLE:!.+]]} -// CHECK-OLD: [[TAG_DOUBLE]] = !{[[DOUBLE:!.+]], [[DOUBLE]], i64 0} // CHECK-OLD [[DOUBLE]] = !{!"double", [[CHAR]], i64 0} -// CHECK-OLD: [[TS7]] = !{i64 0, i64 1, [[TAG_CHAR]], i64 1, i64 1, [[TAG_CHAR]], i64 2, i64 1, [[TAG_CHAR]], i64 3, i64 2, [[TAG_CHAR]], i64 8, i64 8, [[TAG_DOUBLE]], i64 16, i64 1, [[TAG_CHAR]]} -// CHECK-OLD: [[TS8]] = !{i64 0, i64 4, [[TAG_CHAR]], i64 8, i64 8, [[TAG_DOUBLE]]} -// CHECK-OLD: [[TS9]] = !{i64 0, i64 8, [[TAG_CHAR]], i64 8, i64 4, [[TAG_INT]]} -// CHECK-OLD: [[TS10]] = !{i64 0, i64 4, [[TAG_INT]], i64 8, i64 8, [[TAG_CHAR]]} - -// CHECK-NEW-DAG: [[TYPE_char:!.*]] = !{{{.*}}, i64 1, !"omnipotent char"} -// CHECK-NEW-DAG: [[TAG_char]] = !{[[TYPE_char]], [[TYPE_char]], i64 0, i64 0} -// CHECK-NEW-DAG: [[TYPE_short:!.*]] = !{[[TYPE_char]], i64 2, !"short"} -// CHECK-NEW-DAG: [[TYPE_int:!.*]] = !{[[TYPE_char]], i64 4, !"int"} -// CHECK-NEW-DAG: [[TYPE_A:!.*]] = !{[[TYPE_char]], i64 16, !"_ZTS1A", [[TYPE_short]], i64 0, i64 2, [[TYPE_int]], i64 4, i64 4, [[TYPE_char]], i64 8, i64 1, [[TYPE_int]], i64 12, i64 4} -// CHECK-NEW-DAG: [[TAG_A]] = !{[[TYPE_A]], [[TYPE_A]], i64 0, i64 16} -// CHECK-NEW-DAG: [[TYPE_B:!.*]] = !{[[TYPE_char]], i64 24, !"_ZTS1B", [[TYPE_char]], i64 0, i64 1, [[TYPE_A]], i64 4, i64 16, [[TYPE_int]], i64 20, i64 4} -// CHECK-NEW-DAG: [[TAG_B]] = !{[[TYPE_B]], [[TYPE_B]], i64 0, i64 24} -// CHECK-NEW-DAG: [[TAG_U]] = !{[[TYPE_char]], [[TYPE_char]], i64 0, i64 12} -// CHECK-NEW-DAG: [[TYPE_C:!.*]] = !{[[TYPE_char]], i64 3, !"_ZTS1C", [[TYPE_char]], i64 0, i64 1, [[TYPE_char]], i64 1, i64 1, [[TYPE_char]], i64 2, i64 1} -// CHECK-NEW-DAG: [[TAG_C]] = !{[[TYPE_C]], [[TYPE_C]], i64 0, i64 3} -// CHECK-NEW-DAG: [[TYPE_D:!.*]] = !{[[TYPE_char]], i64 6, !"_ZTS1D", [[TYPE_char]], i64 0, i64 1, [[TYPE_char]], i64 4, i64 1, [[TYPE_char]], i64 5, i64 1} -// CHECK-NEW-DAG: [[TAG_D]] = !{[[TYPE_D]], [[TYPE_D]], i64 0, i64 6} -// CHECK-NEW-DAG: [[TAG_NamedBitfields]] = !{[[TYPE_NamedBitfields:!.+]], [[TYPE_NamedBitfields]], i64 0, i64 16} -// CHECK-NEW-DAG: [[TYPE_NamedBitfields]] = !{[[TYPE_char]], i64 16, !"_ZTS14NamedBitfields", [[TYPE_int]], i64 0, i64 4, [[TYPE_int]], i64 1, i64 4, [[TYPE_char]], i64 2, i64 1, [[TYPE_double:!.+]], i64 8, i64 8} -// CHECK-NEW-DAG: [[TYPE_double]] = !{[[TYPE_char]], i64 8, !"double"} -// CHECK-NEW-DAG: [[TAG_NamedBitfields2]] = !{[[TYPE_NamedBitfields2:!.+]], [[TYPE_NamedBitfields2]], i64 0, i64 24} -// CHECK-NEW-DAG: [[TYPE_NamedBitfields2]] = !{[[TYPE_char]], i64 24, !"_ZTS15NamedBitfields2", [[TYPE_char]], i64 0, i64 1, [[TYPE_char]], i64 1, i64 1, [[TYPE_char]], i64 2, i64 1, [[TYPE_int]], i64 3, i64 4, [[TYPE_int]], i64 3, i64 4, [[TYPE_char]], i64 4, i64 1, [[TYPE_double]], i64 8, i64 8, [[TYPE_int]], i64 16, i64 4} -// CHECK-NEW-DAG: [[TAG_NamedBitfields3]] = !{[[TYPE_NamedBitfields3:!.+]], [[TYPE_NamedBitfields3]], i64 0, i64 16} -// CHECK-NEW-DAG: [[TYPE_NamedBitfields3]] = !{[[TYPE_char]], i64 16, !"_ZTS15NamedBitfields3", [[TYPE_int]], i64 1, i64 4, [[TYPE_int]], i64 2, i64 4, [[TYPE_double]], i64 8, i64 8} -// CHECK-NEW-DAG: [[TAG_UnionMember1]] = !{[[TYPE_UnionMember1:!.+]], [[TYPE_UnionMember1]], i64 0, i64 16} -// CHECK-NEW-DAG: [[TYPE_UnionMember1]] = !{[[TYPE_char]], i64 16, !"_ZTS12UnionMember1", [[TYPE_char]], i64 0, i64 8, [[TYPE_int]], i64 8, i64 4} -// CHECK-NEW-DAG: [[TAG_UnionMember2]] = !{[[TYPE_UnionMember2:!.+]], [[TYPE_UnionMember2]], i64 0, i64 16} -// CHECK-NEW-DAG: [[TYPE_UnionMember2]] = !{[[TYPE_char]], i64 16, !"_ZTS12UnionMember2", [[TYPE_int]], i64 0, i64 4, [[TYPE_char]], i64 8, i64 8} + +//. +// CHECK-OLD: [[META2:![0-9]+]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// CHECK-OLD: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// CHECK-OLD: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK-OLD: [[META5]] = !{!"Simple C++ TBAA"} +// CHECK-OLD: [[TBAA_STRUCT6]] = !{i64 0, i64 2, [[META7:![0-9]+]], i64 4, i64 4, [[META2]], i64 8, i64 1, [[META9:![0-9]+]], i64 12, i64 4, [[META2]]} +// CHECK-OLD: [[META7]] = !{[[META8:![0-9]+]], [[META8]], i64 0} +// CHECK-OLD: [[META8]] = !{!"short", [[META4]], i64 0} +// CHECK-OLD: [[META9]] = !{[[META4]], [[META4]], i64 0} +// CHECK-OLD: [[TBAA_STRUCT10]] = !{i64 0, i64 1, [[META9]], i64 4, i64 2, [[META7]], i64 8, i64 4, [[META2]], i64 12, i64 1, [[META9]], i64 16, i64 4, [[META2]], i64 20, i64 4, [[META2]]} +// CHECK-OLD: [[TBAA_STRUCT11]] = !{i64 0, i64 12, [[META9]]} +// CHECK-OLD: [[TBAA_STRUCT12]] = !{i64 0, i64 1, [[META9]], i64 1, i64 1, [[META9]], i64 2, i64 1, [[META9]]} +// CHECK-OLD: [[TBAA_STRUCT13]] = !{i64 0, i64 1, [[META9]], i64 4, i64 1, [[META9]], i64 5, i64 1, [[META9]]} +// CHECK-OLD: [[TBAA_STRUCT14]] = !{i64 0, i64 2, [[META9]], i64 2, i64 1, [[META9]], i64 8, i64 8, [[META15:![0-9]+]]} +// CHECK-OLD: [[META15]] = !{[[META16:![0-9]+]], [[META16]], i64 0} +// CHECK-OLD: [[META16]] = !{!"double", [[META4]], i64 0} +// CHECK-OLD: [[TBAA_STRUCT17]] = !{i64 0, i64 1, [[META9]], i64 1, i64 1, [[META9]], i64 2, i64 1, [[META9]], i64 3, i64 2, [[META9]], i64 8, i64 8, [[META15]], i64 16, i64 1, [[META9]]} +// CHECK-OLD: [[TBAA_STRUCT18]] = !{i64 0, i64 4, [[META9]], i64 8, i64 8, [[META15]]} +// CHECK-OLD: [[TBAA_STRUCT19]] = !{i64 0, i64 8, [[META9]], i64 8, i64 4, [[META2]]} +// CHECK-OLD: [[TBAA_STRUCT20]] = !{i64 0, i64 4, [[META2]], i64 8, i64 8, [[META9]]} +//. +// CHECK-NEW: [[META2:![0-9]+]] = !{[[META3:![0-9]+]], [[META3]], i64 0, i64 4} +// CHECK-NEW: [[META3]] = !{[[META4:![0-9]+]], i64 4, !"int"} +// CHECK-NEW: [[META4]] = !{[[META5:![0-9]+]], i64 1, !"omnipotent char"} +// CHECK-NEW: [[META5]] = !{!"Simple C++ TBAA"} +// CHECK-NEW: [[TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0, i64 16} +// CHECK-NEW: [[META7]] = !{[[META4]], i64 16, !"_ZTS1A", [[META8:![0-9]+]], i64 0, i64 2, [[META3]], i64 4, i64 4, [[META4]], i64 8, i64 1, [[META3]], i64 12, i64 4} +// CHECK-NEW: [[META8]] = !{[[META4]], i64 2, !"short"} +// CHECK-NEW: [[TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0, i64 24} +// CHECK-NEW: [[META13]] = !{[[META4]], i64 24, !"_ZTS1B", [[META4]], i64 0, i64 1, [[META7]], i64 4, i64 16, [[META3]], i64 20, i64 4} +// CHECK-NEW: [[TBAA15]] = !{[[META4]], [[META4]], i64 0, i64 12} +// CHECK-NEW: [[TBAA17]] = !{[[META18:![0-9]+]], [[META18]], i64 0, i64 3} +// CHECK-NEW: [[META18]] = !{[[META4]], i64 3, !"_ZTS1C", [[META4]], i64 0, i64 1, [[META4]], i64 1, i64 1, [[META4]], i64 2, i64 1} +// CHECK-NEW: [[TBAA20]] = !{[[META21:![0-9]+]], [[META21]], i64 0, i64 6} +// CHECK-NEW: [[META21]] = !{[[META4]], i64 6, !"_ZTS1D", [[META4]], i64 0, i64 1, [[META4]], i64 4, i64 1, [[META4]], i64 5, i64 1} +// CHECK-NEW: [[TBAA23]] = !{[[META4]], [[META4]], i64 0, i64 0} +// CHECK-NEW: [[TBAA24]] = !{[[META25:![0-9]+]], [[META25]], i64 0, i64 16} +// CHECK-NEW: [[META25]] = !{[[META4]], i64 16, !"_ZTS14NamedBitfields", [[META3]], i64 0, i64 4, [[META3]], i64 1, i64 4, [[META4]], i64 2, i64 1, [[META26:![0-9]+]], i64 8, i64 8} +// CHECK-NEW: [[META26]] = !{[[META4]], i64 8, !"double"} +// CHECK-NEW: [[TBAA30]] = !{[[META31:![0-9]+]], [[META31]], i64 0, i64 24} +// CHECK-NEW: [[META31]] = !{[[META4]], i64 24, !"_ZTS15NamedBitfields2", [[META4]], i64 0, i64 1, [[META4]], i64 1, i64 1, [[META4]], i64 2, i64 1, [[META3]], i64 3, i64 4, [[META3]], i64 3, i64 4, [[META4]], i64 4, i64 1, [[META26]], i64 8, i64 8, [[META3]], i64 16, i64 4} +// CHECK-NEW: [[TBAA33]] = !{[[META34:![0-9]+]], [[META34]], i64 0, i64 16} +// CHECK-NEW: [[META34]] = !{[[META4]], i64 16, !"_ZTS15NamedBitfields3", [[META3]], i64 1, i64 4, [[META3]], i64 2, i64 4, [[META26]], i64 8, i64 8} +// CHECK-NEW: [[TBAA37]] = !{[[META38:![0-9]+]], [[META38]], i64 0, i64 16} +// CHECK-NEW: [[META38]] = !{[[META4]], i64 16, !"_ZTS12UnionMember1", [[META4]], i64 0, i64 8, [[META3]], i64 8, i64 4} +// CHECK-NEW: [[TBAA41]] = !{[[META42:![0-9]+]], [[META42]], i64 0, i64 16} +// CHECK-NEW: [[META42]] = !{[[META4]], i64 16, !"_ZTS12UnionMember2", [[META3]], i64 0, i64 4, [[META4]], i64 8, i64 8} +//. diff --git a/clang/test/CodeGen/tbaa.c b/clang/test/CodeGen/tbaa.c index 0ab81f6..a719c03 100644 --- a/clang/test/CodeGen/tbaa.c +++ b/clang/test/CodeGen/tbaa.c @@ -1,3 +1,4 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 // RUN: %clang_cc1 -triple x86_64-apple-darwin -O1 -no-struct-path-tbaa -disable-llvm-passes %s -emit-llvm -o - | FileCheck %s // RUN: %clang_cc1 -triple x86_64-apple-darwin -O1 -disable-llvm-passes %s -emit-llvm -o - | FileCheck %s -check-prefixes=PATH // RUN: %clang_cc1 -triple x86_64-apple-darwin -O0 -disable-llvm-passes %s -emit-llvm -o - | FileCheck %s -check-prefix=NO-TBAA @@ -37,13 +38,13 @@ typedef enum : uint8_t { uint32_t g0(EnumAuto32 *E, uint32_t *val) { // CHECK-LABEL: define{{.*}} i32 @g0( -// CHECK: store i32 5, ptr %{{.*}}, align 4, !tbaa [[TAG_i32:!.*]] -// CHECK: store i32 0, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// CHECK: load i32, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] +// CHECK: store i32 5, ptr %{{.*}}, align 4, !tbaa [[TBAA2:![0-9]+]] +// CHECK: store i32 0, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// CHECK: load i32, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] // PATH-LABEL: define{{.*}} i32 @g0( -// PATH: store i32 5, ptr %{{.*}}, align 4, !tbaa [[TAG_i32:!.*]] -// PATH: store i32 0, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// PATH: load i32, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] +// PATH: store i32 5, ptr %{{.*}}, align 4, !tbaa [[TBAA2:![0-9]+]] +// PATH: store i32 0, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// PATH: load i32, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] *val = 5; *E = RED_AUTO_32; return *val; @@ -51,13 +52,13 @@ uint32_t g0(EnumAuto32 *E, uint32_t *val) { uint64_t g1(EnumAuto64 *E, uint64_t *val) { // CHECK-LABEL: define{{.*}} i64 @g1( -// CHECK: store i64 5, ptr %{{.*}}, align 8, !tbaa [[TAG_i64:!.*]] -// CHECK: store i64 0, ptr %{{.*}}, align 8, !tbaa [[TAG_long:!.*]] -// CHECK: load i64, ptr %{{.*}}, align 8, !tbaa [[TAG_i64]] +// CHECK: store i64 5, ptr %{{.*}}, align 8, !tbaa [[TBAA12:![0-9]+]] +// CHECK: store i64 0, ptr %{{.*}}, align 8, !tbaa [[TBAA14:![0-9]+]] +// CHECK: load i64, ptr %{{.*}}, align 8, !tbaa [[TBAA12]] // PATH-LABEL: define{{.*}} i64 @g1( -// PATH: store i64 5, ptr %{{.*}}, align 8, !tbaa [[TAG_i64:!.*]] -// PATH: store i64 0, ptr %{{.*}}, align 8, !tbaa [[TAG_long:!.*]] -// PATH: load i64, ptr %{{.*}}, align 8, !tbaa [[TAG_i64]] +// PATH: store i64 5, ptr %{{.*}}, align 8, !tbaa [[TBAA12:![0-9]+]] +// PATH: store i64 0, ptr %{{.*}}, align 8, !tbaa [[TBAA14:![0-9]+]] +// PATH: load i64, ptr %{{.*}}, align 8, !tbaa [[TBAA12]] *val = 5; *E = RED_AUTO_64; return *val; @@ -65,13 +66,13 @@ uint64_t g1(EnumAuto64 *E, uint64_t *val) { uint16_t g2(Enum16 *E, uint16_t *val) { // CHECK-LABEL: define{{.*}} i16 @g2( -// CHECK: store i16 5, ptr %{{.*}}, align 2, !tbaa [[TAG_i16:!.*]] -// CHECK: store i16 0, ptr %{{.*}}, align 2, !tbaa [[TAG_i16]] -// CHECK: load i16, ptr %{{.*}}, align 2, !tbaa [[TAG_i16]] +// CHECK: store i16 5, ptr %{{.*}}, align 2, !tbaa [[TBAA18:![0-9]+]] +// CHECK: store i16 0, ptr %{{.*}}, align 2, !tbaa [[TBAA18]] +// CHECK: load i16, ptr %{{.*}}, align 2, !tbaa [[TBAA18]] // PATH-LABEL: define{{.*}} i16 @g2( -// PATH: store i16 5, ptr %{{.*}}, align 2, !tbaa [[TAG_i16:!.*]] -// PATH: store i16 0, ptr %{{.*}}, align 2, !tbaa [[TAG_i16]] -// PATH: load i16, ptr %{{.*}}, align 2, !tbaa [[TAG_i16]] +// PATH: store i16 5, ptr %{{.*}}, align 2, !tbaa [[TBAA18:![0-9]+]] +// PATH: store i16 0, ptr %{{.*}}, align 2, !tbaa [[TBAA18]] +// PATH: load i16, ptr %{{.*}}, align 2, !tbaa [[TBAA18]] *val = 5; *E = RED_16; return *val; @@ -79,38 +80,40 @@ uint16_t g2(Enum16 *E, uint16_t *val) { uint8_t g3(Enum8 *E, uint8_t *val) { // CHECK-LABEL: define{{.*}} i8 @g3( -// CHECK: store i8 5, ptr %{{.*}}, align 1, !tbaa [[TAG_i8:!.*]] -// CHECK: store i8 0, ptr %{{.*}}, align 1, !tbaa [[TAG_i8]] -// CHECK: load i8, ptr %{{.*}}, align 1, !tbaa [[TAG_i8]] +// CHECK: store i8 5, ptr %{{.*}}, align 1, !tbaa [[TBAA22:![0-9]+]] +// CHECK: store i8 0, ptr %{{.*}}, align 1, !tbaa [[TBAA22]] +// CHECK: load i8, ptr %{{.*}}, align 1, !tbaa [[TBAA22]] // PATH-LABEL: define{{.*}} i8 @g3( -// PATH: store i8 5, ptr %{{.*}}, align 1, !tbaa [[TAG_i8:!.*]] -// PATH: store i8 0, ptr %{{.*}}, align 1, !tbaa [[TAG_i8]] -// PATH: load i8, ptr %{{.*}}, align 1, !tbaa [[TAG_i8]] +// PATH: store i8 5, ptr %{{.*}}, align 1, !tbaa [[TBAA22:![0-9]+]] +// PATH: store i8 0, ptr %{{.*}}, align 1, !tbaa [[TBAA22]] +// PATH: load i8, ptr %{{.*}}, align 1, !tbaa [[TBAA22]] *val = 5; *E = RED_8; return *val; } -// CHECK: [[TYPE_char:!.*]] = !{!"omnipotent char", [[TAG_c_tbaa:!.*]], -// CHECK: [[TAG_c_tbaa]] = !{!"Simple C/C++ TBAA"} -// CHECK: [[TAG_i32]] = !{[[TYPE_i32:!.*]], [[TYPE_i32]], i64 0} -// CHECK: [[TYPE_i32]] = !{!"int", [[TYPE_char]], -// CHECK: [[TAG_i64]] = !{[[TYPE_i64:!.*]], [[TYPE_i64]], i64 0} -// CHECK: [[TYPE_i64]] = !{!"long long", [[TYPE_char]], -// CHECK: [[TAG_long]] = !{[[TYPE_long:!.*]], [[TYPE_long]], i64 0} -// CHECK: [[TYPE_long]] = !{!"long", [[TYPE_char]], -// CHECK: [[TAG_i16]] = !{[[TYPE_i16:!.*]], [[TYPE_i16]], i64 0} -// CHECK: [[TYPE_i16]] = !{!"short", [[TYPE_char]], -// CHECK: [[TAG_i8]] = !{[[TYPE_i8:!.*]], [[TYPE_char]], i64 0} - -// PATH: [[TYPE_char:!.*]] = !{!"omnipotent char", [[TAG_c_tbaa:!.*]], -// PATH: [[TAG_c_tbaa]] = !{!"Simple C/C++ TBAA"} -// PATH: [[TAG_i32]] = !{[[TYPE_i32:!.*]], [[TYPE_i32]], i64 0} -// PATH: [[TYPE_i32]] = !{!"int", [[TYPE_char]], -// PATH: [[TAG_i64]] = !{[[TYPE_i64:!.*]], [[TYPE_i64]], i64 0} -// PATH: [[TYPE_i64]] = !{!"long long", [[TYPE_char]], -// PATH: [[TAG_long]] = !{[[TYPE_long:!.*]], [[TYPE_long]], i64 0} -// PATH: [[TYPE_long]] = !{!"long", [[TYPE_char]], -// PATH: [[TAG_i16]] = !{[[TYPE_i16:!.*]], [[TYPE_i16]], i64 0} -// PATH: [[TYPE_i16]] = !{!"short", [[TYPE_char]], -// PATH: [[TAG_i8]] = !{[[TYPE_i8:!.*]], [[TYPE_char]], i64 0} +//. +// CHECK: [[TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// CHECK: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} +// CHECK: [[META13]] = !{!"long long", [[META4]], i64 0} +// CHECK: [[TBAA14]] = !{[[META15:![0-9]+]], [[META15]], i64 0} +// CHECK: [[META15]] = !{!"long", [[META4]], i64 0} +// CHECK: [[TBAA18]] = !{[[META19:![0-9]+]], [[META19]], i64 0} +// CHECK: [[META19]] = !{!"short", [[META4]], i64 0} +// CHECK: [[TBAA22]] = !{[[META4]], [[META4]], i64 0} +//. +// PATH: [[TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// PATH: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// PATH: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// PATH: [[META5]] = !{!"Simple C/C++ TBAA"} +// PATH: [[TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} +// PATH: [[META13]] = !{!"long long", [[META4]], i64 0} +// PATH: [[TBAA14]] = !{[[META15:![0-9]+]], [[META15]], i64 0} +// PATH: [[META15]] = !{!"long", [[META4]], i64 0} +// PATH: [[TBAA18]] = !{[[META19:![0-9]+]], [[META19]], i64 0} +// PATH: [[META19]] = !{!"short", [[META4]], i64 0} +// PATH: [[TBAA22]] = !{[[META4]], [[META4]], i64 0} +//. diff --git a/clang/test/CodeGen/tbaa.cpp b/clang/test/CodeGen/tbaa.cpp index 3e92d1e..29c0c58 100644 --- a/clang/test/CodeGen/tbaa.cpp +++ b/clang/test/CodeGen/tbaa.cpp @@ -1,3 +1,4 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 // RUN: %clang_cc1 -triple x86_64-apple-darwin -O1 -no-struct-path-tbaa -disable-llvm-passes %s -emit-llvm -o - | FileCheck %s // RUN: %clang_cc1 -triple x86_64-apple-darwin -O1 -disable-llvm-passes %s -emit-llvm -o - | FileCheck %s -check-prefixes=PATH,OLD-PATH // RUN: %clang_cc1 -triple x86_64-apple-darwin -O1 -new-struct-path-tbaa -disable-llvm-passes %s -emit-llvm -o - | FileCheck %s -check-prefixes=PATH,NEW-PATH @@ -51,11 +52,11 @@ typedef struct uint32_t g(uint32_t *s, StructA *A, uint64_t count) { // CHECK-LABEL: define{{.*}} i32 @_Z1g -// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32:!.*]] -// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] +// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2:![0-9]+]] +// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] // PATH-LABEL: define{{.*}} i32 @_Z1g -// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32:!.*]] -// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_A_f32:!.*]] +// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2:![0-9]+]] +// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA12:![0-9]+]] *s = 1; A->f32 = 4; return *s; @@ -63,11 +64,11 @@ uint32_t g(uint32_t *s, StructA *A, uint64_t count) { uint32_t g2(uint32_t *s, StructA *A, uint64_t count) { // CHECK-LABEL: define{{.*}} i32 @_Z2g2 -// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// CHECK: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TAG_i16:!.*]] +// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// CHECK: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TBAA12:![0-9]+]] // PATH-LABEL: define{{.*}} i32 @_Z2g2 -// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// PATH: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TAG_A_f16:!.*]] +// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// PATH: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TBAA15:![0-9]+]] *s = 1; A->f16 = 4; return *s; @@ -75,11 +76,11 @@ uint32_t g2(uint32_t *s, StructA *A, uint64_t count) { uint32_t g3(StructA *A, StructB *B, uint64_t count) { // CHECK-LABEL: define{{.*}} i32 @_Z2g3 -// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] +// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] // PATH-LABEL: define{{.*}} i32 @_Z2g3 -// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_A_f32]] -// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_B_a_f32:!.*]] +// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA12]] +// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA16:![0-9]+]] A->f32 = 1; B->a.f32 = 4; return A->f32; @@ -87,11 +88,11 @@ uint32_t g3(StructA *A, StructB *B, uint64_t count) { uint32_t g4(StructA *A, StructB *B, uint64_t count) { // CHECK-LABEL: define{{.*}} i32 @_Z2g4 -// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// CHECK: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TAG_i16]] +// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// CHECK: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TBAA12]] // PATH-LABEL: define{{.*}} i32 @_Z2g4 -// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_A_f32]] -// PATH: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TAG_B_a_f16:!.*]] +// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA12]] +// PATH: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TBAA18:![0-9]+]] A->f32 = 1; B->a.f16 = 4; return A->f32; @@ -99,11 +100,11 @@ uint32_t g4(StructA *A, StructB *B, uint64_t count) { uint32_t g5(StructA *A, StructB *B, uint64_t count) { // CHECK-LABEL: define{{.*}} i32 @_Z2g5 -// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] +// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] // PATH-LABEL: define{{.*}} i32 @_Z2g5 -// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_A_f32]] -// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_B_f32:!.*]] +// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA12]] +// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA19:![0-9]+]] A->f32 = 1; B->f32 = 4; return A->f32; @@ -111,11 +112,11 @@ uint32_t g5(StructA *A, StructB *B, uint64_t count) { uint32_t g6(StructA *A, StructB *B, uint64_t count) { // CHECK-LABEL: define{{.*}} i32 @_Z2g6 -// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] +// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] // PATH-LABEL: define{{.*}} i32 @_Z2g6 -// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_A_f32]] -// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_B_a_f32_2:!.*]] +// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA12]] +// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA20:![0-9]+]] A->f32 = 1; B->a.f32_2 = 4; return A->f32; @@ -123,11 +124,11 @@ uint32_t g6(StructA *A, StructB *B, uint64_t count) { uint32_t g7(StructA *A, StructS *S, uint64_t count) { // CHECK-LABEL: define{{.*}} i32 @_Z2g7 -// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] +// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] // PATH-LABEL: define{{.*}} i32 @_Z2g7 -// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_A_f32]] -// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_S_f32:!.*]] +// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA12]] +// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA21:![0-9]+]] A->f32 = 1; S->f32 = 4; return A->f32; @@ -135,11 +136,11 @@ uint32_t g7(StructA *A, StructS *S, uint64_t count) { uint32_t g8(StructA *A, StructS *S, uint64_t count) { // CHECK-LABEL: define{{.*}} i32 @_Z2g8 -// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// CHECK: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TAG_i16]] +// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// CHECK: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TBAA12]] // PATH-LABEL: define{{.*}} i32 @_Z2g8 -// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_A_f32]] -// PATH: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TAG_S_f16:!.*]] +// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA12]] +// PATH: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TBAA23:![0-9]+]] A->f32 = 1; S->f16 = 4; return A->f32; @@ -147,11 +148,11 @@ uint32_t g8(StructA *A, StructS *S, uint64_t count) { uint32_t g9(StructS *S, StructS2 *S2, uint64_t count) { // CHECK-LABEL: define{{.*}} i32 @_Z2g9 -// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] +// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] // PATH-LABEL: define{{.*}} i32 @_Z2g9 -// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_S_f32]] -// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_S2_f32:!.*]] +// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA21]] +// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA24:![0-9]+]] S->f32 = 1; S2->f32 = 4; return S->f32; @@ -159,11 +160,11 @@ uint32_t g9(StructS *S, StructS2 *S2, uint64_t count) { uint32_t g10(StructS *S, StructS2 *S2, uint64_t count) { // CHECK-LABEL: define{{.*}} i32 @_Z3g10 -// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// CHECK: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TAG_i16]] +// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// CHECK: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TBAA12]] // PATH-LABEL: define{{.*}} i32 @_Z3g10 -// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_S_f32]] -// PATH: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TAG_S2_f16:!.*]] +// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA21]] +// PATH: store i16 4, ptr %{{.*}}, align 4, !tbaa [[TBAA26:![0-9]+]] S->f32 = 1; S2->f16 = 4; return S->f32; @@ -171,11 +172,11 @@ uint32_t g10(StructS *S, StructS2 *S2, uint64_t count) { uint32_t g11(StructC *C, StructD *D, uint64_t count) { // CHECK-LABEL: define{{.*}} i32 @_Z3g11 -// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] +// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] // PATH-LABEL: define{{.*}} i32 @_Z3g11 -// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_C_b_a_f32:!.*]] -// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_D_b_a_f32:!.*]] +// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA27:![0-9]+]] +// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA29:![0-9]+]] C->b.a.f32 = 1; D->b.a.f32 = 4; return C->b.a.f32; @@ -183,12 +184,12 @@ uint32_t g11(StructC *C, StructD *D, uint64_t count) { uint32_t g12(StructC *C, StructD *D, uint64_t count) { // CHECK-LABEL: define{{.*}} i32 @_Z3g12 -// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] +// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] // TODO: differentiate the two accesses. // PATH-LABEL: define{{.*}} i32 @_Z3g12 -// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_B_a_f32]] -// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_B_a_f32]] +// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA16]] +// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA16]] StructB *b1 = &(C->b); StructB *b2 = &(D->b); // b1, b2 have different context. @@ -208,9 +209,9 @@ struct five { char g13(struct five *a, struct five *b) { return a->b; // CHECK-LABEL: define{{.*}} signext i8 @_Z3g13 -// CHECK: load i8, ptr %{{.*}}, align 1, !tbaa [[TAG_char:!.*]] +// CHECK: load i8, ptr %{{.*}}, align 1, !tbaa [[TBAA16:![0-9]+]] // PATH-LABEL: define{{.*}} signext i8 @_Z3g13 -// PATH: load i8, ptr %{{.*}}, align 1, !tbaa [[TAG_five_b:!.*]] +// PATH: load i8, ptr %{{.*}}, align 1, !tbaa [[TBAA33:![0-9]+]] } struct six { @@ -221,9 +222,9 @@ struct six { }; char g14(struct six *a, struct six *b) { // CHECK-LABEL: define{{.*}} signext i8 @_Z3g14 -// CHECK: load i8, ptr %{{.*}}, align 1, !tbaa [[TAG_char]] +// CHECK: load i8, ptr %{{.*}}, align 1, !tbaa [[TBAA16]] // PATH-LABEL: define{{.*}} signext i8 @_Z3g14 -// PATH: load i8, ptr %{{.*}}, align 1, !tbaa [[TAG_six_b:!.*]] +// PATH: load i8, ptr %{{.*}}, align 1, !tbaa [[TBAA37:![0-9]+]] return a->b; } @@ -231,75 +232,78 @@ char g14(struct six *a, struct six *b) { typedef StructS StructS3; uint32_t g15(StructS *S, StructS3 *S3, uint64_t count) { // CHECK-LABEL: define{{.*}} i32 @_Z3g15 -// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] -// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_i32]] +// CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] +// CHECK: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA2]] // PATH-LABEL: define{{.*}} i32 @_Z3g15 -// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TAG_S_f32]] -// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TAG_S_f32]] +// PATH: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA21]] +// PATH: store i32 4, ptr %{{.*}}, align 4, !tbaa [[TBAA21]] S->f32 = 1; S3->f32 = 4; return S->f32; } -// CHECK: [[TYPE_char:!.*]] = !{!"omnipotent char", [[TAG_cxx_tbaa:!.*]], -// CHECK: [[TAG_cxx_tbaa]] = !{!"Simple C++ TBAA"} -// CHECK: [[TAG_i32]] = !{[[TYPE_i32:!.*]], [[TYPE_i32]], i64 0} -// CHECK: [[TYPE_i32]] = !{!"int", [[TYPE_char]], -// CHECK: [[TAG_i16]] = !{[[TYPE_i16:!.*]], [[TYPE_i16]], i64 0} -// CHECK: [[TYPE_i16]] = !{!"short", [[TYPE_char]], -// CHECK: [[TAG_char]] = !{[[TYPE_char]], [[TYPE_char]], i64 0} - -// OLD-PATH: [[TYPE_CHAR:!.*]] = !{!"omnipotent char", ! -// OLD-PATH: [[TAG_i32]] = !{[[TYPE_INT:!.*]], [[TYPE_INT]], i64 0} -// OLD-PATH: [[TYPE_INT]] = !{!"int", [[TYPE_CHAR]] -// OLD-PATH: [[TAG_A_f32]] = !{[[TYPE_A:!.*]], [[TYPE_INT]], i64 4} -// OLD-PATH: [[TYPE_A]] = !{!"_ZTS7StructA", [[TYPE_SHORT:!.*]], i64 0, [[TYPE_INT]], i64 4, [[TYPE_SHORT]], i64 8, [[TYPE_INT]], i64 12} -// OLD-PATH: [[TYPE_SHORT:!.*]] = !{!"short", [[TYPE_CHAR]] -// OLD-PATH: [[TAG_A_f16]] = !{[[TYPE_A]], [[TYPE_SHORT]], i64 0} -// OLD-PATH: [[TAG_B_a_f32]] = !{[[TYPE_B:!.*]], [[TYPE_INT]], i64 8} -// OLD-PATH: [[TYPE_B]] = !{!"_ZTS7StructB", [[TYPE_SHORT]], i64 0, [[TYPE_A]], i64 4, [[TYPE_INT]], i64 20} -// OLD-PATH: [[TAG_B_a_f16]] = !{[[TYPE_B]], [[TYPE_SHORT]], i64 4} -// OLD-PATH: [[TAG_B_f32]] = !{[[TYPE_B]], [[TYPE_INT]], i64 20} -// OLD-PATH: [[TAG_B_a_f32_2]] = !{[[TYPE_B]], [[TYPE_INT]], i64 16} -// OLD-PATH: [[TAG_S_f32]] = !{[[TYPE_S:!.*]], [[TYPE_INT]], i64 4} -// OLD-PATH: [[TYPE_S]] = !{!"_ZTS7StructS", [[TYPE_SHORT]], i64 0, [[TYPE_INT]], i64 4} -// OLD-PATH: [[TAG_S_f16]] = !{[[TYPE_S]], [[TYPE_SHORT]], i64 0} -// OLD-PATH: [[TAG_S2_f32]] = !{[[TYPE_S2:!.*]], [[TYPE_INT]], i64 4} -// OLD-PATH: [[TYPE_S2]] = !{!"_ZTS8StructS2", [[TYPE_SHORT]], i64 0, [[TYPE_INT]], i64 4} -// OLD-PATH: [[TAG_S2_f16]] = !{[[TYPE_S2]], [[TYPE_SHORT]], i64 0} -// OLD-PATH: [[TAG_C_b_a_f32]] = !{[[TYPE_C:!.*]], [[TYPE_INT]], i64 12} -// OLD-PATH: [[TYPE_C]] = !{!"_ZTS7StructC", [[TYPE_SHORT]], i64 0, [[TYPE_B]], i64 4, [[TYPE_INT]], i64 28} -// OLD-PATH: [[TAG_D_b_a_f32]] = !{[[TYPE_D:!.*]], [[TYPE_INT]], i64 12} -// OLD-PATH: [[TYPE_D]] = !{!"_ZTS7StructD", [[TYPE_SHORT]], i64 0, [[TYPE_B]], i64 4, [[TYPE_INT]], i64 28, [[TYPE_CHAR]], i64 32} -// OLD-PATH: [[TAG_five_b]] = !{[[TYPE_five:!.*]], [[TYPE_CHAR]], i64 1} -// OLD-PATH: [[TYPE_five]] = !{!"_ZTS4five", [[TYPE_CHAR]], i64 0, [[TYPE_CHAR]], i64 1, [[TYPE_CHAR]], i64 2} -// OLD-PATH: [[TAG_six_b]] = !{[[TYPE_six:!.*]], [[TYPE_CHAR]], i64 4} -// OLD-PATH: [[TYPE_six]] = !{!"_ZTS3six", [[TYPE_CHAR]], i64 0, [[TYPE_CHAR]], i64 4, [[TYPE_CHAR]], i64 5} - -// NEW-PATH-DAG: [[ROOT:!.*]] = !{!"Simple C++ TBAA"} -// NEW-PATH-DAG: [[TYPE_char:!.*]] = !{[[ROOT]], i64 1, !"omnipotent char"} -// NEW-PATH-DAG: [[TYPE_short:!.*]] = !{[[TYPE_char]], i64 2, !"short"} -// NEW-PATH-DAG: [[TYPE_int:!.*]] = !{[[TYPE_char]], i64 4, !"int"} -// NEW-PATH-DAG: [[TAG_i32:!.*]] = !{[[TYPE_int]], [[TYPE_int]], i64 0, i64 4} -// NEW-PATH-DAG: [[TYPE_A:!.*]] = !{[[TYPE_char]], i64 16, !"_ZTS7StructA", [[TYPE_short]], i64 0, i64 2, [[TYPE_int]], i64 4, i64 4, [[TYPE_short]], i64 8, i64 2, [[TYPE_int]], i64 12, i64 4} -// NEW-PATH-DAG: [[TAG_A_f16]] = !{[[TYPE_A]], [[TYPE_short]], i64 0, i64 2} -// NEW-PATH-DAG: [[TAG_A_f32]] = !{[[TYPE_A]], [[TYPE_int]], i64 4, i64 4} -// NEW-PATH-DAG: [[TYPE_B:!.*]] = !{[[TYPE_char]], i64 24, !"_ZTS7StructB", [[TYPE_short]], i64 0, i64 2, [[TYPE_A]], i64 4, i64 16, [[TYPE_int]], i64 20, i64 4} -// NEW-PATH-DAG: [[TAG_B_a_f16]] = !{[[TYPE_B]], [[TYPE_short]], i64 4, i64 2} -// NEW-PATH-DAG: [[TAG_B_a_f32]] = !{[[TYPE_B]], [[TYPE_int]], i64 8, i64 4} -// NEW-PATH-DAG: [[TAG_B_f32]] = !{[[TYPE_B]], [[TYPE_int]], i64 20, i64 4} -// NEW-PATH-DAG: [[TAG_B_a_f32_2]] = !{[[TYPE_B]], [[TYPE_int]], i64 16, i64 4} -// NEW-PATH-DAG: [[TYPE_S:!.*]] = !{[[TYPE_char]], i64 8, !"_ZTS7StructS", [[TYPE_short]], i64 0, i64 2, [[TYPE_int]], i64 4, i64 4} -// NEW-PATH-DAG: [[TAG_S_f16]] = !{[[TYPE_S]], [[TYPE_short]], i64 0, i64 2} -// NEW-PATH-DAG: [[TAG_S_f32]] = !{[[TYPE_S]], [[TYPE_int]], i64 4, i64 4} -// NEW-PATH-DAG: [[TYPE_S2:!.*]] = !{[[TYPE_char]], i64 8, !"_ZTS8StructS2", [[TYPE_short]], i64 0, i64 2, [[TYPE_int]], i64 4, i64 4} -// NEW-PATH-DAG: [[TAG_S2_f16]] = !{[[TYPE_S2]], [[TYPE_short]], i64 0, i64 2} -// NEW-PATH-DAG: [[TAG_S2_f32]] = !{[[TYPE_S2]], [[TYPE_int]], i64 4, i64 4} -// NEW-PATH-DAG: [[TYPE_C:!.*]] = !{[[TYPE_char]], i64 32, !"_ZTS7StructC", [[TYPE_short]], i64 0, i64 2, [[TYPE_B]], i64 4, i64 24, [[TYPE_int]], i64 28, i64 4} -// NEW-PATH-DAG: [[TAG_C_b_a_f32]] = !{[[TYPE_C]], [[TYPE_int]], i64 12, i64 4} -// NEW-PATH-DAG: [[TYPE_D:!.*]] = !{[[TYPE_char]], i64 36, !"_ZTS7StructD", [[TYPE_short]], i64 0, i64 2, [[TYPE_B]], i64 4, i64 24, [[TYPE_int]], i64 28, i64 4, [[TYPE_char]], i64 32, i64 1} -// NEW-PATH-DAG: [[TAG_D_b_a_f32]] = !{[[TYPE_D]], [[TYPE_int]], i64 12, i64 4} -// NEW-PATH-DAG: [[TYPE_five:!.*]] = !{[[TYPE_char]], i64 3, !"_ZTS4five", [[TYPE_char]], i64 0, i64 1, [[TYPE_char]], i64 1, i64 1, [[TYPE_char]], i64 2, i64 1} -// NEW-PATH-DAG: [[TAG_five_b]] = !{[[TYPE_five]], [[TYPE_char]], i64 1, i64 1} -// NEW-PATH-DAG: [[TYPE_six:!.*]] = !{[[TYPE_char]], i64 6, !"_ZTS3six", [[TYPE_char]], i64 0, i64 1, [[TYPE_char]], i64 4, i64 1, [[TYPE_char]], i64 5, i64 1} -// NEW-PATH-DAG: [[TAG_six_b]] = !{[[TYPE_six]], [[TYPE_char]], i64 4, i64 1} +//. +// CHECK: [[TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// CHECK: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C++ TBAA"} +// CHECK: [[TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} +// CHECK: [[META13]] = !{!"short", [[META4]], i64 0} +// CHECK: [[TBAA16]] = !{[[META4]], [[META4]], i64 0} +//. +// OLD-PATH: [[TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// OLD-PATH: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// OLD-PATH: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// OLD-PATH: [[META5]] = !{!"Simple C++ TBAA"} +// OLD-PATH: [[TBAA12]] = !{[[META13:![0-9]+]], [[META3]], i64 4} +// OLD-PATH: [[META13]] = !{!"_ZTS7StructA", [[META14:![0-9]+]], i64 0, [[META3]], i64 4, [[META14]], i64 8, [[META3]], i64 12} +// OLD-PATH: [[META14]] = !{!"short", [[META4]], i64 0} +// OLD-PATH: [[TBAA15]] = !{[[META13]], [[META14]], i64 0} +// OLD-PATH: [[TBAA16]] = !{[[META17:![0-9]+]], [[META3]], i64 8} +// OLD-PATH: [[META17]] = !{!"_ZTS7StructB", [[META14]], i64 0, [[META13]], i64 4, [[META3]], i64 20} +// OLD-PATH: [[TBAA18]] = !{[[META17]], [[META14]], i64 4} +// OLD-PATH: [[TBAA19]] = !{[[META17]], [[META3]], i64 20} +// OLD-PATH: [[TBAA20]] = !{[[META17]], [[META3]], i64 16} +// OLD-PATH: [[TBAA21]] = !{[[META22:![0-9]+]], [[META3]], i64 4} +// OLD-PATH: [[META22]] = !{!"_ZTS7StructS", [[META14]], i64 0, [[META3]], i64 4} +// OLD-PATH: [[TBAA23]] = !{[[META22]], [[META14]], i64 0} +// OLD-PATH: [[TBAA24]] = !{[[META25:![0-9]+]], [[META3]], i64 4} +// OLD-PATH: [[META25]] = !{!"_ZTS8StructS2", [[META14]], i64 0, [[META3]], i64 4} +// OLD-PATH: [[TBAA26]] = !{[[META25]], [[META14]], i64 0} +// OLD-PATH: [[TBAA27]] = !{[[META28:![0-9]+]], [[META3]], i64 12} +// OLD-PATH: [[META28]] = !{!"_ZTS7StructC", [[META14]], i64 0, [[META17]], i64 4, [[META3]], i64 28} +// OLD-PATH: [[TBAA29]] = !{[[META30:![0-9]+]], [[META3]], i64 12} +// OLD-PATH: [[META30]] = !{!"_ZTS7StructD", [[META14]], i64 0, [[META17]], i64 4, [[META3]], i64 28, [[META4]], i64 32} +// OLD-PATH: [[TBAA33]] = !{[[META34:![0-9]+]], [[META4]], i64 1} +// OLD-PATH: [[META34]] = !{!"_ZTS4five", [[META4]], i64 0, [[META4]], i64 1, [[META4]], i64 2} +// OLD-PATH: [[TBAA37]] = !{[[META38:![0-9]+]], [[META4]], i64 4} +// OLD-PATH: [[META38]] = !{!"_ZTS3six", [[META4]], i64 0, [[META4]], i64 4, [[META4]], i64 5} +//. +// NEW-PATH: [[TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0, i64 4} +// NEW-PATH: [[META3]] = !{[[META4:![0-9]+]], i64 4, !"int"} +// NEW-PATH: [[META4]] = !{[[META5:![0-9]+]], i64 1, !"omnipotent char"} +// NEW-PATH: [[META5]] = !{!"Simple C++ TBAA"} +// NEW-PATH: [[TBAA12]] = !{[[META13:![0-9]+]], [[META3]], i64 4, i64 4} +// NEW-PATH: [[META13]] = !{[[META4]], i64 16, !"_ZTS7StructA", [[META14:![0-9]+]], i64 0, i64 2, [[META3]], i64 4, i64 4, [[META14]], i64 8, i64 2, [[META3]], i64 12, i64 4} +// NEW-PATH: [[META14]] = !{[[META4]], i64 2, !"short"} +// NEW-PATH: [[TBAA15]] = !{[[META13]], [[META14]], i64 0, i64 2} +// NEW-PATH: [[TBAA16]] = !{[[META17:![0-9]+]], [[META3]], i64 8, i64 4} +// NEW-PATH: [[META17]] = !{[[META4]], i64 24, !"_ZTS7StructB", [[META14]], i64 0, i64 2, [[META13]], i64 4, i64 16, [[META3]], i64 20, i64 4} +// NEW-PATH: [[TBAA18]] = !{[[META17]], [[META14]], i64 4, i64 2} +// NEW-PATH: [[TBAA19]] = !{[[META17]], [[META3]], i64 20, i64 4} +// NEW-PATH: [[TBAA20]] = !{[[META17]], [[META3]], i64 16, i64 4} +// NEW-PATH: [[TBAA21]] = !{[[META22:![0-9]+]], [[META3]], i64 4, i64 4} +// NEW-PATH: [[META22]] = !{[[META4]], i64 8, !"_ZTS7StructS", [[META14]], i64 0, i64 2, [[META3]], i64 4, i64 4} +// NEW-PATH: [[TBAA23]] = !{[[META22]], [[META14]], i64 0, i64 2} +// NEW-PATH: [[TBAA24]] = !{[[META25:![0-9]+]], [[META3]], i64 4, i64 4} +// NEW-PATH: [[META25]] = !{[[META4]], i64 8, !"_ZTS8StructS2", [[META14]], i64 0, i64 2, [[META3]], i64 4, i64 4} +// NEW-PATH: [[TBAA26]] = !{[[META25]], [[META14]], i64 0, i64 2} +// NEW-PATH: [[TBAA27]] = !{[[META28:![0-9]+]], [[META3]], i64 12, i64 4} +// NEW-PATH: [[META28]] = !{[[META4]], i64 32, !"_ZTS7StructC", [[META14]], i64 0, i64 2, [[META17]], i64 4, i64 24, [[META3]], i64 28, i64 4} +// NEW-PATH: [[TBAA29]] = !{[[META30:![0-9]+]], [[META3]], i64 12, i64 4} +// NEW-PATH: [[META30]] = !{[[META4]], i64 36, !"_ZTS7StructD", [[META14]], i64 0, i64 2, [[META17]], i64 4, i64 24, [[META3]], i64 28, i64 4, [[META4]], i64 32, i64 1} +// NEW-PATH: [[TBAA33]] = !{[[META34:![0-9]+]], [[META4]], i64 1, i64 1} +// NEW-PATH: [[META34]] = !{[[META4]], i64 3, !"_ZTS4five", [[META4]], i64 0, i64 1, [[META4]], i64 1, i64 1, [[META4]], i64 2, i64 1} +// NEW-PATH: [[TBAA37]] = !{[[META38:![0-9]+]], [[META4]], i64 4, i64 1} +// NEW-PATH: [[META38]] = !{[[META4]], i64 6, !"_ZTS3six", [[META4]], i64 0, i64 1, [[META4]], i64 4, i64 1, [[META4]], i64 5, i64 1} +//. diff --git a/clang/test/CodeGenCXX/attr-likelihood-if-branch-weights.cpp b/clang/test/CodeGenCXX/attr-likelihood-if-branch-weights.cpp index 8969e12..2af2929 100644 --- a/clang/test/CodeGenCXX/attr-likelihood-if-branch-weights.cpp +++ b/clang/test/CodeGenCXX/attr-likelihood-if-branch-weights.cpp @@ -10,7 +10,7 @@ extern bool B(); // CHECK-SAME: ) #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[RETVAL:%.*]] = alloca i1, align 1 -// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2:![0-9]+]], !range [[RNG6:![0-9]+]], !noundef [[META7:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6:![0-9]+]], !range [[RNG8:![0-9]+]], !noundef [[META9:![0-9]+]] // CHECK-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP0]] to i1 // CHECK-NEXT: [[LOADEDV_EXPVAL:%.*]] = call i1 @llvm.expect.i1(i1 [[LOADEDV]], i1 true) // CHECK-NEXT: br i1 [[LOADEDV_EXPVAL]], label %[[IF_THEN:.*]], label %[[IF_END:.*]] @@ -38,7 +38,7 @@ bool f() { // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[RETVAL:%.*]] = alloca i1, align 1 -// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP0]] to i1 // CHECK-NEXT: [[LOADEDV_EXPVAL:%.*]] = call i1 @llvm.expect.i1(i1 [[LOADEDV]], i1 false) // CHECK-NEXT: br i1 [[LOADEDV_EXPVAL]], label %[[IF_THEN:.*]], label %[[IF_END:.*]] @@ -67,7 +67,7 @@ bool g() { // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[RETVAL:%.*]] = alloca i1, align 1 -// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP0]] to i1 // CHECK-NEXT: [[LOADEDV_EXPVAL:%.*]] = call i1 @llvm.expect.i1(i1 [[LOADEDV]], i1 false) // CHECK-NEXT: br i1 [[LOADEDV_EXPVAL]], label %[[IF_THEN:.*]], label %[[IF_END:.*]] @@ -93,14 +93,14 @@ bool h() { // CHECK-LABEL: define dso_local void @_Z8NullStmtv( // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP0]] to i1 // CHECK-NEXT: [[LOADEDV_EXPVAL:%.*]] = call i1 @llvm.expect.i1(i1 [[LOADEDV]], i1 false) // CHECK-NEXT: br i1 [[LOADEDV_EXPVAL]], label %[[IF_THEN:.*]], label %[[IF_ELSE:.*]] // CHECK: [[IF_THEN]]: // CHECK-NEXT: br label %[[IF_END:.*]] // CHECK: [[IF_ELSE]]: -// CHECK-NEXT: store volatile i8 1, ptr @b, align 1, !tbaa [[BOOL_TBAA2]] +// CHECK-NEXT: store volatile i8 1, ptr @b, align 1, !tbaa [[BOOL_TBAA6]] // CHECK-NEXT: br label %[[IF_END]] // CHECK: [[IF_END]]: // CHECK-NEXT: ret void @@ -117,7 +117,7 @@ void NullStmt() { // CHECK-LABEL: define dso_local void @_Z6IfStmtv( // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP0]] to i1 // CHECK-NEXT: [[LOADEDV_EXPVAL:%.*]] = call i1 @llvm.expect.i1(i1 [[LOADEDV]], i1 false) // CHECK-NEXT: br i1 [[LOADEDV_EXPVAL]], label %[[IF_THEN:.*]], label %[[IF_END2:.*]] @@ -129,7 +129,7 @@ void NullStmt() { // CHECK: [[IF_END]]: // CHECK-NEXT: br label %[[IF_END2]] // CHECK: [[IF_END2]]: -// CHECK-NEXT: [[TMP1:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV3:%.*]] = trunc i8 [[TMP1]] to i1 // CHECK-NEXT: br i1 [[LOADEDV3]], label %[[IF_THEN4:.*]], label %[[IF_END8:.*]] // CHECK: [[IF_THEN4]]: @@ -137,7 +137,7 @@ void NullStmt() { // CHECK-NEXT: [[CALL5_EXPVAL:%.*]] = call i1 @llvm.expect.i1(i1 [[CALL5]], i1 false) // CHECK-NEXT: br i1 [[CALL5_EXPVAL]], label %[[IF_THEN6:.*]], label %[[IF_END7:.*]] // CHECK: [[IF_THEN6]]: -// CHECK-NEXT: store volatile i8 0, ptr @b, align 1, !tbaa [[BOOL_TBAA2]] +// CHECK-NEXT: store volatile i8 0, ptr @b, align 1, !tbaa [[BOOL_TBAA6]] // CHECK-NEXT: br label %[[IF_END7]] // CHECK: [[IF_END7]]: // CHECK-NEXT: br label %[[IF_END8]] @@ -157,7 +157,7 @@ void IfStmt() { // CHECK-LABEL: define dso_local void @_Z9WhileStmtv( // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP0]] to i1 // CHECK-NEXT: [[LOADEDV_EXPVAL:%.*]] = call i1 @llvm.expect.i1(i1 [[LOADEDV]], i1 false) // CHECK-NEXT: br i1 [[LOADEDV_EXPVAL]], label %[[IF_THEN:.*]], label %[[IF_END:.*]] @@ -167,11 +167,11 @@ void IfStmt() { // CHECK-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @_Z1Bv() // CHECK-NEXT: br i1 [[CALL]], label %[[WHILE_BODY:.*]], label %[[WHILE_END:.*]] // CHECK: [[WHILE_BODY]]: -// CHECK-NEXT: br label %[[WHILE_COND]], !llvm.loop [[LOOP8:![0-9]+]] +// CHECK-NEXT: br label %[[WHILE_COND]], !llvm.loop [[LOOP10:![0-9]+]] // CHECK: [[WHILE_END]]: // CHECK-NEXT: br label %[[IF_END]] // CHECK: [[IF_END]]: -// CHECK-NEXT: [[TMP1:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV1:%.*]] = trunc i8 [[TMP1]] to i1 // CHECK-NEXT: br i1 [[LOADEDV1]], label %[[IF_THEN2:.*]], label %[[IF_END7:.*]] // CHECK: [[IF_THEN2]]: @@ -181,8 +181,8 @@ void IfStmt() { // CHECK-NEXT: [[CALL4_EXPVAL:%.*]] = call i1 @llvm.expect.i1(i1 [[CALL4]], i1 false) // CHECK-NEXT: br i1 [[CALL4_EXPVAL]], label %[[WHILE_BODY5:.*]], label %[[WHILE_END6:.*]] // CHECK: [[WHILE_BODY5]]: -// CHECK-NEXT: store volatile i8 0, ptr @b, align 1, !tbaa [[BOOL_TBAA2]] -// CHECK-NEXT: br label %[[WHILE_COND3]], !llvm.loop [[LOOP11:![0-9]+]] +// CHECK-NEXT: store volatile i8 0, ptr @b, align 1, !tbaa [[BOOL_TBAA6]] +// CHECK-NEXT: br label %[[WHILE_COND3]], !llvm.loop [[LOOP13:![0-9]+]] // CHECK: [[WHILE_END6]]: // CHECK-NEXT: br label %[[IF_END7]] // CHECK: [[IF_END7]]: @@ -200,7 +200,7 @@ void WhileStmt() { // CHECK-LABEL: define dso_local void @_Z6DoStmtv( // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP0]] to i1 // CHECK-NEXT: [[LOADEDV_EXPVAL:%.*]] = call i1 @llvm.expect.i1(i1 [[LOADEDV]], i1 false) // CHECK-NEXT: br i1 [[LOADEDV_EXPVAL]], label %[[IF_THEN:.*]], label %[[IF_END:.*]] @@ -210,11 +210,11 @@ void WhileStmt() { // CHECK-NEXT: br label %[[DO_COND:.*]] // CHECK: [[DO_COND]]: // CHECK-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @_Z1Bv() -// CHECK-NEXT: br i1 [[CALL]], label %[[DO_BODY]], label %[[DO_END:.*]], !llvm.loop [[LOOP12:![0-9]+]] +// CHECK-NEXT: br i1 [[CALL]], label %[[DO_BODY]], label %[[DO_END:.*]], !llvm.loop [[LOOP14:![0-9]+]] // CHECK: [[DO_END]]: // CHECK-NEXT: br label %[[IF_END]] // CHECK: [[IF_END]]: -// CHECK-NEXT: [[TMP1:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV1:%.*]] = trunc i8 [[TMP1]] to i1 // CHECK-NEXT: br i1 [[LOADEDV1]], label %[[IF_THEN2:.*]], label %[[IF_END7:.*]] // CHECK: [[IF_THEN2]]: @@ -223,7 +223,7 @@ void WhileStmt() { // CHECK-NEXT: br label %[[DO_COND4:.*]] // CHECK: [[DO_COND4]]: // CHECK-NEXT: [[CALL5:%.*]] = call noundef zeroext i1 @_Z1Bv() -// CHECK-NEXT: br i1 [[CALL5]], label %[[DO_BODY3]], label %[[DO_END6:.*]], !llvm.loop [[LOOP13:![0-9]+]] +// CHECK-NEXT: br i1 [[CALL5]], label %[[DO_BODY3]], label %[[DO_END6:.*]], !llvm.loop [[LOOP15:![0-9]+]] // CHECK: [[DO_END6]]: // CHECK-NEXT: br label %[[IF_END7]] // CHECK: [[IF_END7]]: @@ -244,7 +244,7 @@ void DoStmt() { // CHECK-LABEL: define dso_local void @_Z7ForStmtv( // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP0]] to i1 // CHECK-NEXT: [[LOADEDV_EXPVAL:%.*]] = call i1 @llvm.expect.i1(i1 [[LOADEDV]], i1 false) // CHECK-NEXT: br i1 [[LOADEDV_EXPVAL]], label %[[IF_THEN:.*]], label %[[IF_END:.*]] @@ -254,11 +254,11 @@ void DoStmt() { // CHECK-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @_Z1Bv() // CHECK-NEXT: br i1 [[CALL]], label %[[FOR_BODY:.*]], label %[[FOR_END:.*]] // CHECK: [[FOR_BODY]]: -// CHECK-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]] +// CHECK-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]] // CHECK: [[FOR_END]]: // CHECK-NEXT: br label %[[IF_END]] // CHECK: [[IF_END]]: -// CHECK-NEXT: [[TMP1:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV1:%.*]] = trunc i8 [[TMP1]] to i1 // CHECK-NEXT: br i1 [[LOADEDV1]], label %[[IF_THEN2:.*]], label %[[IF_END7:.*]] // CHECK: [[IF_THEN2]]: @@ -268,7 +268,7 @@ void DoStmt() { // CHECK-NEXT: [[CALL4_EXPVAL:%.*]] = call i1 @llvm.expect.i1(i1 [[CALL4]], i1 false) // CHECK-NEXT: br i1 [[CALL4_EXPVAL]], label %[[FOR_BODY5:.*]], label %[[FOR_END6:.*]] // CHECK: [[FOR_BODY5]]: -// CHECK-NEXT: br label %[[FOR_COND3]], !llvm.loop [[LOOP15:![0-9]+]] +// CHECK-NEXT: br label %[[FOR_COND3]], !llvm.loop [[LOOP17:![0-9]+]] // CHECK: [[FOR_END6]]: // CHECK-NEXT: br label %[[IF_END7]] // CHECK: [[IF_END7]]: @@ -286,14 +286,14 @@ void ForStmt() { // CHECK-LABEL: define dso_local void @_Z8GotoStmtv( // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP0]] to i1 // CHECK-NEXT: [[LOADEDV_EXPVAL:%.*]] = call i1 @llvm.expect.i1(i1 [[LOADEDV]], i1 false) // CHECK-NEXT: br i1 [[LOADEDV_EXPVAL]], label %[[IF_THEN:.*]], label %[[IF_ELSE:.*]] // CHECK: [[IF_THEN]]: // CHECK-NEXT: br label %[[END:.*]] // CHECK: [[IF_ELSE]]: -// CHECK-NEXT: store volatile i8 1, ptr @b, align 1, !tbaa [[BOOL_TBAA2]] +// CHECK-NEXT: store volatile i8 1, ptr @b, align 1, !tbaa [[BOOL_TBAA6]] // CHECK-NEXT: br label %[[IF_END:.*]] // CHECK: [[IF_END]]: // CHECK-NEXT: br label %[[END]] @@ -313,14 +313,14 @@ end:; // CHECK-LABEL: define dso_local void @_Z10ReturnStmtv( // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP0]] to i1 // CHECK-NEXT: [[LOADEDV_EXPVAL:%.*]] = call i1 @llvm.expect.i1(i1 [[LOADEDV]], i1 false) // CHECK-NEXT: br i1 [[LOADEDV_EXPVAL]], label %[[IF_THEN:.*]], label %[[IF_ELSE:.*]] // CHECK: [[IF_THEN]]: // CHECK-NEXT: br label %[[IF_END:.*]] // CHECK: [[IF_ELSE]]: -// CHECK-NEXT: store volatile i8 1, ptr @b, align 1, !tbaa [[BOOL_TBAA2]] +// CHECK-NEXT: store volatile i8 1, ptr @b, align 1, !tbaa [[BOOL_TBAA6]] // CHECK-NEXT: br label %[[IF_END]] // CHECK: [[IF_END]]: // CHECK-NEXT: ret void @@ -337,31 +337,31 @@ void ReturnStmt() { // CHECK-LABEL: define dso_local void @_Z10SwitchStmtv( // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP0:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP0]] to i1 // CHECK-NEXT: [[LOADEDV_EXPVAL:%.*]] = call i1 @llvm.expect.i1(i1 [[LOADEDV]], i1 false) // CHECK-NEXT: br i1 [[LOADEDV_EXPVAL]], label %[[IF_THEN:.*]], label %[[IF_ELSE:.*]] // CHECK: [[IF_THEN]]: -// CHECK-NEXT: [[TMP1:%.*]] = load volatile i32, ptr @i, align 4, !tbaa [[INT_TBAA16:![0-9]+]] +// CHECK-NEXT: [[TMP1:%.*]] = load volatile i32, ptr @i, align 4, !tbaa [[INT_TBAA2:![0-9]+]] // CHECK-NEXT: switch i32 [[TMP1]], label %[[SW_EPILOG:.*]] [ // CHECK-NEXT: ] // CHECK: [[SW_EPILOG]]: // CHECK-NEXT: br label %[[IF_END:.*]] // CHECK: [[IF_ELSE]]: -// CHECK-NEXT: store volatile i8 1, ptr @b, align 1, !tbaa [[BOOL_TBAA2]] +// CHECK-NEXT: store volatile i8 1, ptr @b, align 1, !tbaa [[BOOL_TBAA6]] // CHECK-NEXT: br label %[[IF_END]] // CHECK: [[IF_END]]: -// CHECK-NEXT: [[TMP2:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA2]], !range [[RNG6]], !noundef [[META7]] +// CHECK-NEXT: [[TMP2:%.*]] = load volatile i8, ptr @b, align 1, !tbaa [[BOOL_TBAA6]], !range [[RNG8]], !noundef [[META9]] // CHECK-NEXT: [[LOADEDV1:%.*]] = trunc i8 [[TMP2]] to i1 // CHECK-NEXT: br i1 [[LOADEDV1]], label %[[IF_THEN2:.*]], label %[[IF_ELSE4:.*]] // CHECK: [[IF_THEN2]]: -// CHECK-NEXT: [[TMP3:%.*]] = load volatile i32, ptr @i, align 4, !tbaa [[INT_TBAA16]] +// CHECK-NEXT: [[TMP3:%.*]] = load volatile i32, ptr @i, align 4, !tbaa [[INT_TBAA2]] // CHECK-NEXT: switch i32 [[TMP3]], label %[[SW_EPILOG3:.*]] [ // CHECK-NEXT: ] // CHECK: [[SW_EPILOG3]]: // CHECK-NEXT: br label %[[IF_END5:.*]] // CHECK: [[IF_ELSE4]]: -// CHECK-NEXT: store volatile i8 1, ptr @b, align 1, !tbaa [[BOOL_TBAA2]] +// CHECK-NEXT: store volatile i8 1, ptr @b, align 1, !tbaa [[BOOL_TBAA6]] // CHECK-NEXT: br label %[[IF_END5]] // CHECK: [[IF_END5]]: // CHECK-NEXT: ret void @@ -383,20 +383,20 @@ void SwitchStmt() { } //. -// CHECK: [[BOOL_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"bool", [[META4:![0-9]+]], i64 0} +// CHECK: [[INT_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// CHECK: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} // CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} // CHECK: [[META5]] = !{!"Simple C++ TBAA"} -// CHECK: [[RNG6]] = !{i8 0, i8 2} -// CHECK: [[META7]] = !{} -// CHECK: [[LOOP8]] = distinct !{[[LOOP8]], [[META9:![0-9]+]], [[META10:![0-9]+]]} -// CHECK: [[META9]] = !{!"llvm.loop.mustprogress"} -// CHECK: [[META10]] = !{!"llvm.loop.unroll.disable"} -// CHECK: [[LOOP11]] = distinct !{[[LOOP11]], [[META9]], [[META10]]} -// CHECK: [[LOOP12]] = distinct !{[[LOOP12]], [[META9]], [[META10]]} -// CHECK: [[LOOP13]] = distinct !{[[LOOP13]], [[META9]], [[META10]]} -// CHECK: [[LOOP14]] = distinct !{[[LOOP14]], [[META9]], [[META10]]} -// CHECK: [[LOOP15]] = distinct !{[[LOOP15]], [[META9]], [[META10]]} -// CHECK: [[INT_TBAA16]] = !{[[META17:![0-9]+]], [[META17]], i64 0} -// CHECK: [[META17]] = !{!"int", [[META4]], i64 0} +// CHECK: [[BOOL_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK: [[META7]] = !{!"bool", [[META4]], i64 0} +// CHECK: [[RNG8]] = !{i8 0, i8 2} +// CHECK: [[META9]] = !{} +// CHECK: [[LOOP10]] = distinct !{[[LOOP10]], [[META11:![0-9]+]], [[META12:![0-9]+]]} +// CHECK: [[META11]] = !{!"llvm.loop.mustprogress"} +// CHECK: [[META12]] = !{!"llvm.loop.unroll.disable"} +// CHECK: [[LOOP13]] = distinct !{[[LOOP13]], [[META11]], [[META12]]} +// CHECK: [[LOOP14]] = distinct !{[[LOOP14]], [[META11]], [[META12]]} +// CHECK: [[LOOP15]] = distinct !{[[LOOP15]], [[META11]], [[META12]]} +// CHECK: [[LOOP16]] = distinct !{[[LOOP16]], [[META11]], [[META12]]} +// CHECK: [[LOOP17]] = distinct !{[[LOOP17]], [[META11]], [[META12]]} //. diff --git a/clang/test/CodeGenCXX/builtin-get-vtable-pointer.cpp b/clang/test/CodeGenCXX/builtin-get-vtable-pointer.cpp index 604fb6c..0bde634 100644 --- a/clang/test/CodeGenCXX/builtin-get-vtable-pointer.cpp +++ b/clang/test/CodeGenCXX/builtin-get-vtable-pointer.cpp @@ -1,3 +1,4 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 // RUN: %clang_cc1 %s -x c++ -std=c++23 -triple x86_64-apple-darwin10 -emit-llvm -O1 -disable-llvm-passes -no-enable-noundef-analysis -o - | FileCheck --check-prefix=CHECK-NOAUTH %s // RUN: %clang_cc1 %s -x c++ -std=c++23 -triple arm64-apple-ios -fptrauth-calls -fptrauth-vtable-pointer-type-discrimination -emit-llvm -O1 -disable-llvm-passes -no-enable-noundef-analysis -o - | FileCheck --check-prefix=CHECK-TYPEAUTH %s // RUN: %clang_cc1 %s -x c++ -std=c++23 -triple arm64-apple-ios -fptrauth-calls -fptrauth-vtable-pointer-address-discrimination -emit-llvm -O1 -disable-llvm-passes -no-enable-noundef-analysis -o - | FileCheck --check-prefix=CHECK-ADDRESSAUTH %s @@ -35,309 +36,1290 @@ template <class A> struct same_type<A, A> { static const bool value = true; }; +// CHECK-NOAUTH-LABEL: define ptr @_ZN5test11aEPNS_1AE( +// CHECK-NOAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-NOAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-NOAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-NOAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA6:![0-9]+]] +// CHECK-NOAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA6]] +// CHECK-NOAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9:![0-9]+]] +// CHECK-NOAUTH-NEXT: ret ptr [[VTABLE]] +// +// CHECK-TYPEAUTH-LABEL: define ptr @_ZN5test11aEPNS_1AE( +// CHECK-TYPEAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-TYPEAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-TYPEAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-TYPEAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA6:![0-9]+]] +// CHECK-TYPEAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA6]] +// CHECK-TYPEAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9:![0-9]+]] +// CHECK-TYPEAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP1]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP3:%.*]] = inttoptr i64 [[TMP2]] to ptr +// CHECK-TYPEAUTH-NEXT: [[TMP4:%.*]] = load volatile i8, ptr [[TMP3]], align 8 +// CHECK-TYPEAUTH-NEXT: ret ptr [[TMP3]] +// +// CHECK-ADDRESSAUTH-LABEL: define ptr @_ZN5test11aEPNS_1AE( +// CHECK-ADDRESSAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-ADDRESSAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-ADDRESSAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-ADDRESSAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA6:![0-9]+]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA6]] +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9:![0-9]+]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP2]], i32 2, i64 [[TMP1]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP4:%.*]] = inttoptr i64 [[TMP3]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[TMP5:%.*]] = load volatile i8, ptr [[TMP4]], align 8 +// CHECK-ADDRESSAUTH-NEXT: ret ptr [[TMP4]] +// +// CHECK-BOTHAUTH-LABEL: define ptr @_ZN5test11aEPNS_1AE( +// CHECK-BOTHAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0:[0-9]+]] { +// CHECK-BOTHAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-BOTHAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-BOTHAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA6:![0-9]+]] +// CHECK-BOTHAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA6]] +// CHECK-BOTHAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9:![0-9]+]] +// CHECK-BOTHAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP1]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP3:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP3]], i32 2, i64 [[TMP2]]) +// CHECK-BOTHAUTH-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP4]] to ptr +// CHECK-BOTHAUTH-NEXT: [[TMP6:%.*]] = load volatile i8, ptr [[TMP5]], align 8 +// CHECK-BOTHAUTH-NEXT: ret ptr [[TMP5]] +// const void *a(A *o) { static_assert(same_type<decltype(__builtin_get_vtable_pointer(o)), const void*>::value); - // CHECK-NOAUTH: define ptr @_ZN5test11aEPNS_1AE(ptr %o) #0 { - // CHECK-TYPEAUTH: define ptr @_ZN5test11aEPNS_1AE(ptr %o) #0 { return __builtin_get_vtable_pointer(o); - // CHECK-NOAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %0 = load ptr, ptr %o.addr, align 8 - // CHECK-TYPEAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %1 = ptrtoint ptr %vtable to i64 - // CHECK-TYPEAUTH: %2 = call i64 @llvm.ptrauth.auth(i64 %1, i32 2, i64 48388) - // CHECK-TYPEAUTH: %3 = inttoptr i64 %2 to ptr - // CHECK-TYPEAUTH: %4 = load volatile i8, ptr %3, align 8 - // CHECK-ADDRESSAUTH: %2 = ptrtoint ptr %vtable to i64 - // CHECK-ADDRESSAUTH: %3 = call i64 @llvm.ptrauth.auth(i64 %2, i32 2, i64 %1) - // CHECK-ADDRESSAUTH: %4 = inttoptr i64 %3 to ptr - // CHECK-ADDRESSAUTH: %5 = load volatile i8, ptr %4, align 8 - // CHECK-BOTHAUTH: [[T1:%.*]] = ptrtoint ptr %0 to i64 - // CHECK-BOTHAUTH: [[T2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[T1]], i64 48388) - // CHECK-BOTHAUTH: [[T3:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-BOTHAUTH: [[T4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[T3]], i32 2, i64 [[T2]]) - // CHECK-BOTHAUTH: [[T5:%.*]] = inttoptr i64 [[T4]] to ptr - // CHECK-BOTHAUTH: [[T6:%.*]] = load volatile i8, ptr [[T5]], align 8 } +// CHECK-NOAUTH-LABEL: define ptr @_ZN5test11bEPNS_1BE( +// CHECK-NOAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-NOAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-NOAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-NOAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA11:![0-9]+]] +// CHECK-NOAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA11]] +// CHECK-NOAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: ret ptr [[VTABLE]] +// +// CHECK-TYPEAUTH-LABEL: define ptr @_ZN5test11bEPNS_1BE( +// CHECK-TYPEAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-TYPEAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-TYPEAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-TYPEAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA11:![0-9]+]] +// CHECK-TYPEAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA11]] +// CHECK-TYPEAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP1]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP3:%.*]] = inttoptr i64 [[TMP2]] to ptr +// CHECK-TYPEAUTH-NEXT: [[TMP4:%.*]] = load volatile i8, ptr [[TMP3]], align 8 +// CHECK-TYPEAUTH-NEXT: ret ptr [[TMP3]] +// +// CHECK-ADDRESSAUTH-LABEL: define ptr @_ZN5test11bEPNS_1BE( +// CHECK-ADDRESSAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-ADDRESSAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-ADDRESSAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-ADDRESSAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA11:![0-9]+]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA11]] +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP2]], i32 2, i64 [[TMP1]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP4:%.*]] = inttoptr i64 [[TMP3]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[TMP5:%.*]] = load volatile i8, ptr [[TMP4]], align 8 +// CHECK-ADDRESSAUTH-NEXT: ret ptr [[TMP4]] +// +// CHECK-BOTHAUTH-LABEL: define ptr @_ZN5test11bEPNS_1BE( +// CHECK-BOTHAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-BOTHAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-BOTHAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-BOTHAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA11:![0-9]+]] +// CHECK-BOTHAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA11]] +// CHECK-BOTHAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP1]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP3:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP3]], i32 2, i64 [[TMP2]]) +// CHECK-BOTHAUTH-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP4]] to ptr +// CHECK-BOTHAUTH-NEXT: [[TMP6:%.*]] = load volatile i8, ptr [[TMP5]], align 8 +// CHECK-BOTHAUTH-NEXT: ret ptr [[TMP5]] +// const void *b(B *o) { static_assert(same_type<decltype(__builtin_get_vtable_pointer(o)), const void*>::value); - // CHECK-TYPEAUTH: define ptr @_ZN5test11bEPNS_1BE(ptr %o) #0 { - // CHECK-NOAUTH: define ptr @_ZN5test11bEPNS_1BE(ptr %o) #0 { return __builtin_get_vtable_pointer(o); - // CHECK-NOAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %1 = ptrtoint ptr %vtable to i64 - // CHECK-TYPEAUTH: %2 = call i64 @llvm.ptrauth.auth(i64 %1, i32 2, i64 48388) - // CHECK-TYPEAUTH: %3 = inttoptr i64 %2 to ptr - // CHECK-TYPEAUTH: %4 = load volatile i8, ptr %3, align 8 - // CHECK-ADDRESSAUTH: %2 = ptrtoint ptr %vtable to i64 - // CHECK-ADDRESSAUTH: %3 = call i64 @llvm.ptrauth.auth(i64 %2, i32 2, i64 %1) - // CHECK-ADDRESSAUTH: %4 = inttoptr i64 %3 to ptr - // CHECK-ADDRESSAUTH: %5 = load volatile i8, ptr %4, align 8 - // CHECK-BOTHAUTH: [[T2:%.*]] = call i64 @llvm.ptrauth.blend(i64 %1, i64 48388) - // CHECK-BOTHAUTH: [[T3:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-BOTHAUTH: [[T4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[T3]], i32 2, i64 [[T2]]) - // CHECK-BOTHAUTH: [[T5:%.*]] = inttoptr i64 [[T4]] to ptr - // CHECK-BOTHAUTH: [[T6:%.*]] = load volatile i8, ptr [[T5]], align 8 } +// CHECK-NOAUTH-LABEL: define ptr @_ZN5test16b_as_AEPNS_1BE( +// CHECK-NOAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-NOAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-NOAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-NOAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA11]] +// CHECK-NOAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA11]] +// CHECK-NOAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: ret ptr [[VTABLE]] +// +// CHECK-TYPEAUTH-LABEL: define ptr @_ZN5test16b_as_AEPNS_1BE( +// CHECK-TYPEAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-TYPEAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-TYPEAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-TYPEAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA11]] +// CHECK-TYPEAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA11]] +// CHECK-TYPEAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP1]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP3:%.*]] = inttoptr i64 [[TMP2]] to ptr +// CHECK-TYPEAUTH-NEXT: [[TMP4:%.*]] = load volatile i8, ptr [[TMP3]], align 8 +// CHECK-TYPEAUTH-NEXT: ret ptr [[TMP3]] +// +// CHECK-ADDRESSAUTH-LABEL: define ptr @_ZN5test16b_as_AEPNS_1BE( +// CHECK-ADDRESSAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-ADDRESSAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-ADDRESSAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-ADDRESSAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA11]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA11]] +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP2]], i32 2, i64 [[TMP1]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP4:%.*]] = inttoptr i64 [[TMP3]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[TMP5:%.*]] = load volatile i8, ptr [[TMP4]], align 8 +// CHECK-ADDRESSAUTH-NEXT: ret ptr [[TMP4]] +// +// CHECK-BOTHAUTH-LABEL: define ptr @_ZN5test16b_as_AEPNS_1BE( +// CHECK-BOTHAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-BOTHAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-BOTHAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-BOTHAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA11]] +// CHECK-BOTHAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA11]] +// CHECK-BOTHAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP1]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP3:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP3]], i32 2, i64 [[TMP2]]) +// CHECK-BOTHAUTH-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP4]] to ptr +// CHECK-BOTHAUTH-NEXT: [[TMP6:%.*]] = load volatile i8, ptr [[TMP5]], align 8 +// CHECK-BOTHAUTH-NEXT: ret ptr [[TMP5]] +// const void *b_as_A(B *o) { static_assert(same_type<decltype(__builtin_get_vtable_pointer(o)), const void*>::value); - // CHECK-NOAUTH: define ptr @_ZN5test16b_as_AEPNS_1BE(ptr %o) #0 { return __builtin_get_vtable_pointer((A *)o); - // CHECK-NOAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %1 = ptrtoint ptr %vtable to i64 - // CHECK-TYPEAUTH: %2 = call i64 @llvm.ptrauth.auth(i64 %1, i32 2, i64 48388) - // CHECK-TYPEAUTH: %3 = inttoptr i64 %2 to ptr - // CHECK-TYPEAUTH: %4 = load volatile i8, ptr %3, align 8 - // CHECK-ADDRESSAUTH: %2 = ptrtoint ptr %vtable to i64 - // CHECK-ADDRESSAUTH: %3 = call i64 @llvm.ptrauth.auth(i64 %2, i32 2, i64 %1) - // CHECK-ADDRESSAUTH: %4 = inttoptr i64 %3 to ptr - // CHECK-ADDRESSAUTH: %5 = load volatile i8, ptr %4, align 8 - // CHECK-BOTHAUTH: [[T2:%.*]] = call i64 @llvm.ptrauth.blend(i64 %1, i64 48388) - // CHECK-BOTHAUTH: [[T3:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-BOTHAUTH: [[T4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[T3]], i32 2, i64 [[T2]]) - // CHECK-BOTHAUTH: [[T5:%.*]] = inttoptr i64 [[T4]] to ptr - // CHECK-BOTHAUTH: [[T6:%.*]] = load volatile i8, ptr [[T5]], align 8 } +// CHECK-NOAUTH-LABEL: define ptr @_ZN5test11cEPNS_1CE( +// CHECK-NOAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-NOAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-NOAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-NOAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA13:![0-9]+]] +// CHECK-NOAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-NOAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: ret ptr [[VTABLE]] +// +// CHECK-TYPEAUTH-LABEL: define ptr @_ZN5test11cEPNS_1CE( +// CHECK-TYPEAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-TYPEAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-TYPEAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-TYPEAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA13:![0-9]+]] +// CHECK-TYPEAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-TYPEAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP1]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP3:%.*]] = inttoptr i64 [[TMP2]] to ptr +// CHECK-TYPEAUTH-NEXT: [[TMP4:%.*]] = load volatile i8, ptr [[TMP3]], align 8 +// CHECK-TYPEAUTH-NEXT: ret ptr [[TMP3]] +// +// CHECK-ADDRESSAUTH-LABEL: define ptr @_ZN5test11cEPNS_1CE( +// CHECK-ADDRESSAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-ADDRESSAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-ADDRESSAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-ADDRESSAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA13:![0-9]+]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP2]], i32 2, i64 [[TMP1]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP4:%.*]] = inttoptr i64 [[TMP3]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[TMP5:%.*]] = load volatile i8, ptr [[TMP4]], align 8 +// CHECK-ADDRESSAUTH-NEXT: ret ptr [[TMP4]] +// +// CHECK-BOTHAUTH-LABEL: define ptr @_ZN5test11cEPNS_1CE( +// CHECK-BOTHAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-BOTHAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-BOTHAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-BOTHAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA13:![0-9]+]] +// CHECK-BOTHAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-BOTHAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP1]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP3:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP3]], i32 2, i64 [[TMP2]]) +// CHECK-BOTHAUTH-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP4]] to ptr +// CHECK-BOTHAUTH-NEXT: [[TMP6:%.*]] = load volatile i8, ptr [[TMP5]], align 8 +// CHECK-BOTHAUTH-NEXT: ret ptr [[TMP5]] +// const void *c(C *o) { static_assert(same_type<decltype(__builtin_get_vtable_pointer(o)), const void*>::value); - // CHECK-NOAUTH: define ptr @_ZN5test11cEPNS_1CE(ptr %o) #0 { return __builtin_get_vtable_pointer(o); - // CHECK-NOAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %1 = ptrtoint ptr %vtable to i64 - // CHECK-TYPEAUTH: %2 = call i64 @llvm.ptrauth.auth(i64 %1, i32 2, i64 48388) - // CHECK-TYPEAUTH: %3 = inttoptr i64 %2 to ptr - // CHECK-TYPEAUTH: %4 = load volatile i8, ptr %3, align 8 - // CHECK-ADDRESSAUTH: %2 = ptrtoint ptr %vtable to i64 - // CHECK-ADDRESSAUTH: %3 = call i64 @llvm.ptrauth.auth(i64 %2, i32 2, i64 %1) - // CHECK-ADDRESSAUTH: %4 = inttoptr i64 %3 to ptr - // CHECK-ADDRESSAUTH: %5 = load volatile i8, ptr %4, align 8 - // CHECK-BOTHAUTH: [[T2:%.*]] = call i64 @llvm.ptrauth.blend(i64 %1, i64 48388) - // CHECK-BOTHAUTH: [[T3:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-BOTHAUTH: [[T4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[T3]], i32 2, i64 [[T2]]) - // CHECK-BOTHAUTH: [[T5:%.*]] = inttoptr i64 [[T4]] to ptr - // CHECK-BOTHAUTH: [[T6:%.*]] = load volatile i8, ptr [[T5]], align 8 } +// CHECK-NOAUTH-LABEL: define ptr @_ZN5test16c_as_ZEPNS_1CE( +// CHECK-NOAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-NOAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-NOAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-NOAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-NOAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-NOAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: ret ptr [[VTABLE]] +// +// CHECK-TYPEAUTH-LABEL: define ptr @_ZN5test16c_as_ZEPNS_1CE( +// CHECK-TYPEAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-TYPEAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-TYPEAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-TYPEAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-TYPEAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-TYPEAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP1]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP3:%.*]] = inttoptr i64 [[TMP2]] to ptr +// CHECK-TYPEAUTH-NEXT: [[TMP4:%.*]] = load volatile i8, ptr [[TMP3]], align 8 +// CHECK-TYPEAUTH-NEXT: ret ptr [[TMP3]] +// +// CHECK-ADDRESSAUTH-LABEL: define ptr @_ZN5test16c_as_ZEPNS_1CE( +// CHECK-ADDRESSAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-ADDRESSAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-ADDRESSAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-ADDRESSAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP2]], i32 2, i64 [[TMP1]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP4:%.*]] = inttoptr i64 [[TMP3]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[TMP5:%.*]] = load volatile i8, ptr [[TMP4]], align 8 +// CHECK-ADDRESSAUTH-NEXT: ret ptr [[TMP4]] +// +// CHECK-BOTHAUTH-LABEL: define ptr @_ZN5test16c_as_ZEPNS_1CE( +// CHECK-BOTHAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-BOTHAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-BOTHAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-BOTHAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-BOTHAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-BOTHAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP1]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP3:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP3]], i32 2, i64 [[TMP2]]) +// CHECK-BOTHAUTH-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP4]] to ptr +// CHECK-BOTHAUTH-NEXT: [[TMP6:%.*]] = load volatile i8, ptr [[TMP5]], align 8 +// CHECK-BOTHAUTH-NEXT: ret ptr [[TMP5]] +// const void *c_as_Z(C *o) { static_assert(same_type<decltype(__builtin_get_vtable_pointer(o)), const void*>::value); - // CHECK-NOAUTH: define ptr @_ZN5test16c_as_ZEPNS_1CE(ptr %o) #0 { return __builtin_get_vtable_pointer((Z *)o); - // CHECK-NOAUTH: %0 = load ptr, ptr %o.addr, align 8 - // CHECK-NOAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %1 = ptrtoint ptr %vtable to i64 - // CHECK-TYPEAUTH: %2 = call i64 @llvm.ptrauth.auth(i64 %1, i32 2, i64 48388) - // CHECK-TYPEAUTH: %3 = inttoptr i64 %2 to ptr - // CHECK-TYPEAUTH: %4 = load volatile i8, ptr %3, align 8 - // CHECK-ADDRESSAUTH: %2 = ptrtoint ptr %vtable to i64 - // CHECK-ADDRESSAUTH: %3 = call i64 @llvm.ptrauth.auth(i64 %2, i32 2, i64 %1) - // CHECK-ADDRESSAUTH: %4 = inttoptr i64 %3 to ptr - // CHECK-ADDRESSAUTH: %5 = load volatile i8, ptr %4, align 8 - // CHECK-BOTHAUTH: [[T2:%.*]] = call i64 @llvm.ptrauth.blend(i64 %1, i64 48388) - // CHECK-BOTHAUTH: [[T3:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-BOTHAUTH: [[T4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[T3]], i32 2, i64 [[T2]]) - // CHECK-BOTHAUTH: [[T5:%.*]] = inttoptr i64 [[T4]] to ptr - // CHECK-BOTHAUTH: [[T6:%.*]] = load volatile i8, ptr [[T5]], align 8 } +// CHECK-NOAUTH-LABEL: define ptr @_ZN5test16c_as_BEPNS_1CE( +// CHECK-NOAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-NOAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-NOAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-NOAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-NOAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-NOAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[TMP0]], null +// CHECK-NOAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-NOAUTH: [[CAST_NOTNULL]]: +// CHECK-NOAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8 +// CHECK-NOAUTH-NEXT: br label %[[CAST_END]] +// CHECK-NOAUTH: [[CAST_END]]: +// CHECK-NOAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-NOAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[CAST_RESULT]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: ret ptr [[VTABLE]] +// +// CHECK-TYPEAUTH-LABEL: define ptr @_ZN5test16c_as_BEPNS_1CE( +// CHECK-TYPEAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-TYPEAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-TYPEAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-TYPEAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-TYPEAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-TYPEAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[TMP0]], null +// CHECK-TYPEAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-TYPEAUTH: [[CAST_NOTNULL]]: +// CHECK-TYPEAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8 +// CHECK-TYPEAUTH-NEXT: br label %[[CAST_END]] +// CHECK-TYPEAUTH: [[CAST_END]]: +// CHECK-TYPEAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-TYPEAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[CAST_RESULT]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP2]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP4:%.*]] = inttoptr i64 [[TMP3]] to ptr +// CHECK-TYPEAUTH-NEXT: [[TMP5:%.*]] = load volatile i8, ptr [[TMP4]], align 8 +// CHECK-TYPEAUTH-NEXT: ret ptr [[TMP4]] +// +// CHECK-ADDRESSAUTH-LABEL: define ptr @_ZN5test16c_as_BEPNS_1CE( +// CHECK-ADDRESSAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-ADDRESSAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-ADDRESSAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-ADDRESSAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[TMP0]], null +// CHECK-ADDRESSAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-ADDRESSAUTH: [[CAST_NOTNULL]]: +// CHECK-ADDRESSAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8 +// CHECK-ADDRESSAUTH-NEXT: br label %[[CAST_END]] +// CHECK-ADDRESSAUTH: [[CAST_END]]: +// CHECK-ADDRESSAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[CAST_RESULT]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[CAST_RESULT]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP3:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP3]], i32 2, i64 [[TMP2]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP4]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[TMP6:%.*]] = load volatile i8, ptr [[TMP5]], align 8 +// CHECK-ADDRESSAUTH-NEXT: ret ptr [[TMP5]] +// +// CHECK-BOTHAUTH-LABEL: define ptr @_ZN5test16c_as_BEPNS_1CE( +// CHECK-BOTHAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-BOTHAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-BOTHAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-BOTHAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-BOTHAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA13]] +// CHECK-BOTHAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[TMP0]], null +// CHECK-BOTHAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-BOTHAUTH: [[CAST_NOTNULL]]: +// CHECK-BOTHAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8 +// CHECK-BOTHAUTH-NEXT: br label %[[CAST_END]] +// CHECK-BOTHAUTH: [[CAST_END]]: +// CHECK-BOTHAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-BOTHAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[CAST_RESULT]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[CAST_RESULT]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP2]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP4:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP5:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP4]], i32 2, i64 [[TMP3]]) +// CHECK-BOTHAUTH-NEXT: [[TMP6:%.*]] = inttoptr i64 [[TMP5]] to ptr +// CHECK-BOTHAUTH-NEXT: [[TMP7:%.*]] = load volatile i8, ptr [[TMP6]], align 8 +// CHECK-BOTHAUTH-NEXT: ret ptr [[TMP6]] +// const void *c_as_B(C *o) { static_assert(same_type<decltype(__builtin_get_vtable_pointer(o)), const void*>::value); - // CHECK-NOAUTH: define ptr @_ZN5test16c_as_BEPNS_1CE(ptr %o) #0 { return __builtin_get_vtable_pointer((B *)o); - // CHECK-NOAUTH: %add.ptr = getelementptr inbounds i8, ptr %0, i64 8 - // CHECK-NOAUTH: br label %cast.end - // CHECK-NOAUTH: %cast.result = phi ptr [ %add.ptr, %cast.notnull ], [ null, %entry ] - // CHECK-NOAUTH: %vtable = load ptr, ptr %cast.result, align 8 - // CHECK-TYPEAUTH: %cast.result = phi ptr [ %add.ptr, %cast.notnull ], [ null, %entry ] - // CHECK-TYPEAUTH: %vtable = load ptr, ptr %cast.result, align 8 - // CHECK-TYPEAUTH: %2 = ptrtoint ptr %vtable to i64 - // CHECK-TYPEAUTH: %3 = call i64 @llvm.ptrauth.auth(i64 %2, i32 2, i64 48388) - // CHECK-TYPEAUTH: %4 = inttoptr i64 %3 to ptr - // CHECK-TYPEAUTH: %5 = load volatile i8, ptr %4, align 8 - // CHECK-ADDRESSAUTH: %2 = ptrtoint ptr %cast.result to i64 - // CHECK-ADDRESSAUTH: %3 = ptrtoint ptr %vtable to i64 - // CHECK-ADDRESSAUTH: %4 = call i64 @llvm.ptrauth.auth(i64 %3, i32 2, i64 %2) - // CHECK-ADDRESSAUTH: %5 = inttoptr i64 %4 to ptr - // CHECK-ADDRESSAUTH: %6 = load volatile i8, ptr %5, align 8 - // CHECK-BOTHAUTH: [[T2:%.*]] = call i64 @llvm.ptrauth.blend(i64 %2, i64 48388) - // CHECK-BOTHAUTH: [[T3:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-BOTHAUTH: [[T4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[T3]], i32 2, i64 [[T2]]) - // CHECK-BOTHAUTH: [[T5:%.*]] = inttoptr i64 [[T4]] to ptr - // CHECK-BOTHAUTH: [[T6:%.*]] = load volatile i8, ptr [[T5]], align 8 } +// CHECK-NOAUTH-LABEL: define ptr @_ZN5test11dEPNS_1DE( +// CHECK-NOAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-NOAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-NOAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-NOAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA15:![0-9]+]] +// CHECK-NOAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA15]] +// CHECK-NOAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: ret ptr [[VTABLE]] +// +// CHECK-TYPEAUTH-LABEL: define ptr @_ZN5test11dEPNS_1DE( +// CHECK-TYPEAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-TYPEAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-TYPEAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-TYPEAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA15:![0-9]+]] +// CHECK-TYPEAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA15]] +// CHECK-TYPEAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP1]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP3:%.*]] = inttoptr i64 [[TMP2]] to ptr +// CHECK-TYPEAUTH-NEXT: [[TMP4:%.*]] = load volatile i8, ptr [[TMP3]], align 8 +// CHECK-TYPEAUTH-NEXT: ret ptr [[TMP3]] +// +// CHECK-ADDRESSAUTH-LABEL: define ptr @_ZN5test11dEPNS_1DE( +// CHECK-ADDRESSAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-ADDRESSAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-ADDRESSAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-ADDRESSAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA15:![0-9]+]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA15]] +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP2]], i32 2, i64 [[TMP1]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP4:%.*]] = inttoptr i64 [[TMP3]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[TMP5:%.*]] = load volatile i8, ptr [[TMP4]], align 8 +// CHECK-ADDRESSAUTH-NEXT: ret ptr [[TMP4]] +// +// CHECK-BOTHAUTH-LABEL: define ptr @_ZN5test11dEPNS_1DE( +// CHECK-BOTHAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-BOTHAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-BOTHAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-BOTHAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA15:![0-9]+]] +// CHECK-BOTHAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA15]] +// CHECK-BOTHAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP1]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP3:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP3]], i32 2, i64 [[TMP2]]) +// CHECK-BOTHAUTH-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP4]] to ptr +// CHECK-BOTHAUTH-NEXT: [[TMP6:%.*]] = load volatile i8, ptr [[TMP5]], align 8 +// CHECK-BOTHAUTH-NEXT: ret ptr [[TMP5]] +// const void *d(D *o) { static_assert(same_type<decltype(__builtin_get_vtable_pointer(o)), const void*>::value); - // CHECK-NOAUTH: define ptr @_ZN5test11dEPNS_1DE(ptr %o) #0 { return __builtin_get_vtable_pointer(o); - // CHECK-NOAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %1 = ptrtoint ptr %vtable to i64 - // CHECK-TYPEAUTH: %2 = call i64 @llvm.ptrauth.auth(i64 %1, i32 2, i64 48388) - // CHECK-TYPEAUTH: %3 = inttoptr i64 %2 to ptr - // CHECK-TYPEAUTH: %4 = load volatile i8, ptr %3, align 8 - // CHECK-ADDRESSAUTH: %1 = ptrtoint ptr %0 to i64 - // CHECK-ADDRESSAUTH: %2 = ptrtoint ptr %vtable to i64 - // CHECK-ADDRESSAUTH: %3 = call i64 @llvm.ptrauth.auth(i64 %2, i32 2, i64 %1) - // CHECK-ADDRESSAUTH: %4 = inttoptr i64 %3 to ptr - // CHECK-ADDRESSAUTH: %5 = load volatile i8, ptr %4, align 8 - // CHECK-BOTHAUTH: [[T1:%.*]] = ptrtoint ptr %0 to i64 - // CHECK-BOTHAUTH: [[T2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[T1]], i64 48388) - // CHECK-BOTHAUTH: [[T3:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-BOTHAUTH: [[T4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[T3]], i32 2, i64 [[T2]]) - // CHECK-BOTHAUTH: [[T5:%.*]] = inttoptr i64 [[T4]] to ptr - // CHECK-BOTHAUTH: [[T6:%.*]] = load volatile i8, ptr [[T5]], align 8 } +// CHECK-NOAUTH-LABEL: define ptr @_ZN5test16d_as_AEPNS_1DE( +// CHECK-NOAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-NOAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-NOAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-NOAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA15]] +// CHECK-NOAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA15]] +// CHECK-NOAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[TMP0]], null +// CHECK-NOAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-NOAUTH: [[CAST_NOTNULL]]: +// CHECK-NOAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: [[VBASE_OFFSET_PTR:%.*]] = getelementptr i8, ptr [[VTABLE]], i64 -32 +// CHECK-NOAUTH-NEXT: [[VBASE_OFFSET:%.*]] = load i64, ptr [[VBASE_OFFSET_PTR]], align 8 +// CHECK-NOAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 [[VBASE_OFFSET]] +// CHECK-NOAUTH-NEXT: br label %[[CAST_END]] +// CHECK-NOAUTH: [[CAST_END]]: +// CHECK-NOAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-NOAUTH-NEXT: [[VTABLE1:%.*]] = load ptr, ptr [[CAST_RESULT]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: ret ptr [[VTABLE1]] +// +// CHECK-TYPEAUTH-LABEL: define ptr @_ZN5test16d_as_AEPNS_1DE( +// CHECK-TYPEAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-TYPEAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-TYPEAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-TYPEAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA15]] +// CHECK-TYPEAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA15]] +// CHECK-TYPEAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[TMP0]], null +// CHECK-TYPEAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-TYPEAUTH: [[CAST_NOTNULL]]: +// CHECK-TYPEAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP2]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP4:%.*]] = inttoptr i64 [[TMP3]] to ptr +// CHECK-TYPEAUTH-NEXT: [[VBASE_OFFSET_PTR:%.*]] = getelementptr i8, ptr [[TMP4]], i64 -32 +// CHECK-TYPEAUTH-NEXT: [[VBASE_OFFSET:%.*]] = load i64, ptr [[VBASE_OFFSET_PTR]], align 8 +// CHECK-TYPEAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 [[VBASE_OFFSET]] +// CHECK-TYPEAUTH-NEXT: br label %[[CAST_END]] +// CHECK-TYPEAUTH: [[CAST_END]]: +// CHECK-TYPEAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-TYPEAUTH-NEXT: [[VTABLE1:%.*]] = load ptr, ptr [[CAST_RESULT]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP5:%.*]] = ptrtoint ptr [[VTABLE1]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP6:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP5]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP7:%.*]] = inttoptr i64 [[TMP6]] to ptr +// CHECK-TYPEAUTH-NEXT: [[TMP8:%.*]] = load volatile i8, ptr [[TMP7]], align 8 +// CHECK-TYPEAUTH-NEXT: ret ptr [[TMP7]] +// +// CHECK-ADDRESSAUTH-LABEL: define ptr @_ZN5test16d_as_AEPNS_1DE( +// CHECK-ADDRESSAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-ADDRESSAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-ADDRESSAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-ADDRESSAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA15]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA15]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[TMP0]], null +// CHECK-ADDRESSAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-ADDRESSAUTH: [[CAST_NOTNULL]]: +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP3:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP3]], i32 2, i64 [[TMP2]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP4]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[VBASE_OFFSET_PTR:%.*]] = getelementptr i8, ptr [[TMP5]], i64 -32 +// CHECK-ADDRESSAUTH-NEXT: [[VBASE_OFFSET:%.*]] = load i64, ptr [[VBASE_OFFSET_PTR]], align 8 +// CHECK-ADDRESSAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 [[VBASE_OFFSET]] +// CHECK-ADDRESSAUTH-NEXT: br label %[[CAST_END]] +// CHECK-ADDRESSAUTH: [[CAST_END]]: +// CHECK-ADDRESSAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE1:%.*]] = load ptr, ptr [[CAST_RESULT]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP6:%.*]] = ptrtoint ptr [[CAST_RESULT]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP7:%.*]] = ptrtoint ptr [[VTABLE1]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP8:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP7]], i32 2, i64 [[TMP6]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP9:%.*]] = inttoptr i64 [[TMP8]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[TMP10:%.*]] = load volatile i8, ptr [[TMP9]], align 8 +// CHECK-ADDRESSAUTH-NEXT: ret ptr [[TMP9]] +// +// CHECK-BOTHAUTH-LABEL: define ptr @_ZN5test16d_as_AEPNS_1DE( +// CHECK-BOTHAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-BOTHAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-BOTHAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-BOTHAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA15]] +// CHECK-BOTHAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA15]] +// CHECK-BOTHAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[TMP0]], null +// CHECK-BOTHAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-BOTHAUTH: [[CAST_NOTNULL]]: +// CHECK-BOTHAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP2]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP4:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP5:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP4]], i32 2, i64 [[TMP3]]) +// CHECK-BOTHAUTH-NEXT: [[TMP6:%.*]] = inttoptr i64 [[TMP5]] to ptr +// CHECK-BOTHAUTH-NEXT: [[VBASE_OFFSET_PTR:%.*]] = getelementptr i8, ptr [[TMP6]], i64 -32 +// CHECK-BOTHAUTH-NEXT: [[VBASE_OFFSET:%.*]] = load i64, ptr [[VBASE_OFFSET_PTR]], align 8 +// CHECK-BOTHAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 [[VBASE_OFFSET]] +// CHECK-BOTHAUTH-NEXT: br label %[[CAST_END]] +// CHECK-BOTHAUTH: [[CAST_END]]: +// CHECK-BOTHAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-BOTHAUTH-NEXT: [[VTABLE1:%.*]] = load ptr, ptr [[CAST_RESULT]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP7:%.*]] = ptrtoint ptr [[CAST_RESULT]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP8:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP7]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP9:%.*]] = ptrtoint ptr [[VTABLE1]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP10:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP9]], i32 2, i64 [[TMP8]]) +// CHECK-BOTHAUTH-NEXT: [[TMP11:%.*]] = inttoptr i64 [[TMP10]] to ptr +// CHECK-BOTHAUTH-NEXT: [[TMP12:%.*]] = load volatile i8, ptr [[TMP11]], align 8 +// CHECK-BOTHAUTH-NEXT: ret ptr [[TMP11]] +// const void *d_as_A(D *o) { static_assert(same_type<decltype(__builtin_get_vtable_pointer(o)), const void*>::value); - // CHECK-NOAUTH: define ptr @_ZN5test16d_as_AEPNS_1DE(ptr %o) #0 { return __builtin_get_vtable_pointer((A *)o); - // CHECK-NOAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-NOAUTH: %vbase.offset.ptr = getelementptr i8, ptr %vtable, i64 -32 - // CHECK-NOAUTH: %vbase.offset = load i64, ptr %vbase.offset.ptr, align 8 - // CHECK-NOAUTH: %add.ptr = getelementptr inbounds i8, ptr %0, i64 %vbase.offset - // CHECK-NOAUTH: %cast.result = phi ptr [ %add.ptr, %cast.notnull ], [ null, %entry ] - // CHECK-NOAUTH: %vtable1 = load ptr, ptr %cast.result, align 8 - // CHECK-TYPEAUTH: %vtable1 = load ptr, ptr %cast.result, align 8 - // CHECK-TYPEAUTH: %5 = ptrtoint ptr %vtable1 to i64 - // CHECK-TYPEAUTH: %6 = call i64 @llvm.ptrauth.auth(i64 %5, i32 2, i64 48388) - // CHECK-TYPEAUTH: %7 = inttoptr i64 %6 to ptr - // CHECK-TYPEAUTH: %8 = load volatile i8, ptr %7, align 8 - // CHECK-ADDRESSAUTH: %6 = ptrtoint ptr %cast.result to i64 - // CHECK-ADDRESSAUTH: %7 = ptrtoint ptr %vtable1 to i64 - // CHECK-ADDRESSAUTH: %8 = call i64 @llvm.ptrauth.auth(i64 %7, i32 2, i64 %6) - // CHECK-ADDRESSAUTH: %9 = inttoptr i64 %8 to ptr - // CHECK-ADDRESSAUTH: %10 = load volatile i8, ptr %9, align 8 - // CHECK-BOTHAUTH: [[T1:%.*]] = ptrtoint ptr %cast.result to i64 - // CHECK-BOTHAUTH: [[T2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[T1]], i64 48388) - // CHECK-BOTHAUTH: [[T3:%.*]] = ptrtoint ptr %vtable1 to i64 - // CHECK-BOTHAUTH: [[T4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[T3]], i32 2, i64 [[T2]]) - // CHECK-BOTHAUTH: [[T5:%.*]] = inttoptr i64 [[T4]] to ptr - // CHECK-BOTHAUTH: [[T6:%.*]] = load volatile i8, ptr [[T5]], align 8 } +// CHECK-NOAUTH-LABEL: define ptr @_ZN5test11eEPNS_1EE( +// CHECK-NOAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-NOAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-NOAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-NOAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA17:![0-9]+]] +// CHECK-NOAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-NOAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: ret ptr [[VTABLE]] +// +// CHECK-TYPEAUTH-LABEL: define ptr @_ZN5test11eEPNS_1EE( +// CHECK-TYPEAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-TYPEAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-TYPEAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-TYPEAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA17:![0-9]+]] +// CHECK-TYPEAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-TYPEAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP1]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP3:%.*]] = inttoptr i64 [[TMP2]] to ptr +// CHECK-TYPEAUTH-NEXT: [[TMP4:%.*]] = load volatile i8, ptr [[TMP3]], align 8 +// CHECK-TYPEAUTH-NEXT: ret ptr [[TMP3]] +// +// CHECK-ADDRESSAUTH-LABEL: define ptr @_ZN5test11eEPNS_1EE( +// CHECK-ADDRESSAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-ADDRESSAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-ADDRESSAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-ADDRESSAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA17:![0-9]+]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP2]], i32 2, i64 [[TMP1]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP4:%.*]] = inttoptr i64 [[TMP3]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[TMP5:%.*]] = load volatile i8, ptr [[TMP4]], align 8 +// CHECK-ADDRESSAUTH-NEXT: ret ptr [[TMP4]] +// +// CHECK-BOTHAUTH-LABEL: define ptr @_ZN5test11eEPNS_1EE( +// CHECK-BOTHAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-BOTHAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-BOTHAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-BOTHAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA17:![0-9]+]] +// CHECK-BOTHAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-BOTHAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP1]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP3:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP3]], i32 2, i64 [[TMP2]]) +// CHECK-BOTHAUTH-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP4]] to ptr +// CHECK-BOTHAUTH-NEXT: [[TMP6:%.*]] = load volatile i8, ptr [[TMP5]], align 8 +// CHECK-BOTHAUTH-NEXT: ret ptr [[TMP5]] +// const void *e(E *o) { static_assert(same_type<decltype(__builtin_get_vtable_pointer(o)), const void*>::value); - // CHECK-NOAUTH: define ptr @_ZN5test11eEPNS_1EE(ptr %o) #0 { return __builtin_get_vtable_pointer(o); - // CHECK-NOAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %1 = ptrtoint ptr %vtable to i64 - // CHECK-TYPEAUTH: %2 = call i64 @llvm.ptrauth.auth(i64 %1, i32 2, i64 48388) - // CHECK-TYPEAUTH: %3 = inttoptr i64 %2 to ptr - // CHECK-TYPEAUTH: %4 = load volatile i8, ptr %3, align 8 - // CHECK-ADDRESSAUTH: [[T1:%.*]] = ptrtoint ptr %0 to i64 - // CHECK-ADDRESSAUTH: [[T2:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-ADDRESSAUTH: [[T3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[T2]], i32 2, i64 [[T1]]) - // CHECK-ADDRESSAUTH: [[T4:%.*]] = inttoptr i64 [[T3]] to ptr - // CHECK-ADDRESSAUTH: [[T5:%.*]] = load volatile i8, ptr [[T4]], align 8 - // CHECK-BOTHAUTH: [[T1:%.*]] = ptrtoint ptr %0 to i64 - // CHECK-BOTHAUTH: [[T2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[T1]], i64 48388) - // CHECK-BOTHAUTH: [[T3:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-BOTHAUTH: [[T4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[T3]], i32 2, i64 [[T2]]) - // CHECK-BOTHAUTH: [[T5:%.*]] = inttoptr i64 [[T4]] to ptr - // CHECK-BOTHAUTH: [[T6:%.*]] = load volatile i8, ptr [[T5]], align 8 } +// CHECK-NOAUTH-LABEL: define ptr @_ZN5test16e_as_BEPNS_1EE( +// CHECK-NOAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-NOAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-NOAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-NOAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-NOAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-NOAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[TMP0]], null +// CHECK-NOAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-NOAUTH: [[CAST_NOTNULL]]: +// CHECK-NOAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8 +// CHECK-NOAUTH-NEXT: br label %[[CAST_END]] +// CHECK-NOAUTH: [[CAST_END]]: +// CHECK-NOAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-NOAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[CAST_RESULT]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: ret ptr [[VTABLE]] +// +// CHECK-TYPEAUTH-LABEL: define ptr @_ZN5test16e_as_BEPNS_1EE( +// CHECK-TYPEAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-TYPEAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-TYPEAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-TYPEAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-TYPEAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-TYPEAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[TMP0]], null +// CHECK-TYPEAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-TYPEAUTH: [[CAST_NOTNULL]]: +// CHECK-TYPEAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8 +// CHECK-TYPEAUTH-NEXT: br label %[[CAST_END]] +// CHECK-TYPEAUTH: [[CAST_END]]: +// CHECK-TYPEAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-TYPEAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[CAST_RESULT]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP2]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP4:%.*]] = inttoptr i64 [[TMP3]] to ptr +// CHECK-TYPEAUTH-NEXT: [[TMP5:%.*]] = load volatile i8, ptr [[TMP4]], align 8 +// CHECK-TYPEAUTH-NEXT: ret ptr [[TMP4]] +// +// CHECK-ADDRESSAUTH-LABEL: define ptr @_ZN5test16e_as_BEPNS_1EE( +// CHECK-ADDRESSAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-ADDRESSAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-ADDRESSAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-ADDRESSAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[TMP0]], null +// CHECK-ADDRESSAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-ADDRESSAUTH: [[CAST_NOTNULL]]: +// CHECK-ADDRESSAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8 +// CHECK-ADDRESSAUTH-NEXT: br label %[[CAST_END]] +// CHECK-ADDRESSAUTH: [[CAST_END]]: +// CHECK-ADDRESSAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[CAST_RESULT]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[CAST_RESULT]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP3:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP3]], i32 2, i64 [[TMP2]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP4]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[TMP6:%.*]] = load volatile i8, ptr [[TMP5]], align 8 +// CHECK-ADDRESSAUTH-NEXT: ret ptr [[TMP5]] +// +// CHECK-BOTHAUTH-LABEL: define ptr @_ZN5test16e_as_BEPNS_1EE( +// CHECK-BOTHAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-BOTHAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-BOTHAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-BOTHAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-BOTHAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-BOTHAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[TMP0]], null +// CHECK-BOTHAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-BOTHAUTH: [[CAST_NOTNULL]]: +// CHECK-BOTHAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8 +// CHECK-BOTHAUTH-NEXT: br label %[[CAST_END]] +// CHECK-BOTHAUTH: [[CAST_END]]: +// CHECK-BOTHAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-BOTHAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[CAST_RESULT]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[CAST_RESULT]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP2]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP4:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP5:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP4]], i32 2, i64 [[TMP3]]) +// CHECK-BOTHAUTH-NEXT: [[TMP6:%.*]] = inttoptr i64 [[TMP5]] to ptr +// CHECK-BOTHAUTH-NEXT: [[TMP7:%.*]] = load volatile i8, ptr [[TMP6]], align 8 +// CHECK-BOTHAUTH-NEXT: ret ptr [[TMP6]] +// const void *e_as_B(E *o) { static_assert(same_type<decltype(__builtin_get_vtable_pointer(o)), const void*>::value); - // CHECK-NOAUTH: define ptr @_ZN5test16e_as_BEPNS_1EE(ptr %o) #0 { return __builtin_get_vtable_pointer((B *)o); - // CHECK-NOAUTH: %add.ptr = getelementptr inbounds i8, ptr %0, i64 8 - // CHECK-NOAUTH: %cast.result = phi ptr [ %add.ptr, %cast.notnull ], [ null, %entry ] - // CHECK-NOAUTH: %vtable = load ptr, ptr %cast.result, align 8 - // CHECK-TYPEAUTH: %vtable = load ptr, ptr %cast.result, align 8 - // CHECK-TYPEAUTH: %2 = ptrtoint ptr %vtable to i64 - // CHECK-TYPEAUTH: %3 = call i64 @llvm.ptrauth.auth(i64 %2, i32 2, i64 48388) - // CHECK-TYPEAUTH: %4 = inttoptr i64 %3 to ptr - // CHECK-TYPEAUTH: %5 = load volatile i8, ptr %4, align 8 - // CHECK-ADDRESSAUTH: [[T1:%.*]] = ptrtoint ptr %cast.result to i64 - // CHECK-ADDRESSAUTH: [[T2:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-ADDRESSAUTH: [[T3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[T2]], i32 2, i64 [[T1]]) - // CHECK-ADDRESSAUTH: [[T4:%.*]] = inttoptr i64 [[T3]] to ptr - // CHECK-ADDRESSAUTH: [[T5:%.*]] = load volatile i8, ptr [[T4]], align 8 - // CHECK-BOTHAUTH: [[T1:%.*]] = ptrtoint ptr %cast.result to i64 - // CHECK-BOTHAUTH: [[T2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[T1]], i64 48388) - // CHECK-BOTHAUTH: [[T3:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-BOTHAUTH: [[T4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[T3]], i32 2, i64 [[T2]]) - // CHECK-BOTHAUTH: [[T5:%.*]] = inttoptr i64 [[T4]] to ptr - // CHECK-BOTHAUTH: [[T6:%.*]] = load volatile i8, ptr [[T5]], align 8 } +// CHECK-NOAUTH-LABEL: define ptr @_ZN5test16e_as_DEPNS_1EE( +// CHECK-NOAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-NOAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-NOAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-NOAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-NOAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-NOAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: ret ptr [[VTABLE]] +// +// CHECK-TYPEAUTH-LABEL: define ptr @_ZN5test16e_as_DEPNS_1EE( +// CHECK-TYPEAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-TYPEAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-TYPEAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-TYPEAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-TYPEAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-TYPEAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP1]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP3:%.*]] = inttoptr i64 [[TMP2]] to ptr +// CHECK-TYPEAUTH-NEXT: [[TMP4:%.*]] = load volatile i8, ptr [[TMP3]], align 8 +// CHECK-TYPEAUTH-NEXT: ret ptr [[TMP3]] +// +// CHECK-ADDRESSAUTH-LABEL: define ptr @_ZN5test16e_as_DEPNS_1EE( +// CHECK-ADDRESSAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-ADDRESSAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-ADDRESSAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-ADDRESSAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP2]], i32 2, i64 [[TMP1]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP4:%.*]] = inttoptr i64 [[TMP3]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[TMP5:%.*]] = load volatile i8, ptr [[TMP4]], align 8 +// CHECK-ADDRESSAUTH-NEXT: ret ptr [[TMP4]] +// +// CHECK-BOTHAUTH-LABEL: define ptr @_ZN5test16e_as_DEPNS_1EE( +// CHECK-BOTHAUTH-SAME: ptr [[O:%.*]]) #[[ATTR0]] { +// CHECK-BOTHAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-BOTHAUTH-NEXT: [[O_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-BOTHAUTH-NEXT: store ptr [[O]], ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-BOTHAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[O_ADDR]], align 8, !tbaa [[TBAA17]] +// CHECK-BOTHAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP1]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP3:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP3]], i32 2, i64 [[TMP2]]) +// CHECK-BOTHAUTH-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP4]] to ptr +// CHECK-BOTHAUTH-NEXT: [[TMP6:%.*]] = load volatile i8, ptr [[TMP5]], align 8 +// CHECK-BOTHAUTH-NEXT: ret ptr [[TMP5]] +// const void *e_as_D(E *o) { static_assert(same_type<decltype(__builtin_get_vtable_pointer(o)), const void*>::value); - // CHECK-NOAUTH: define ptr @_ZN5test16e_as_DEPNS_1EE(ptr %o) #0 { return __builtin_get_vtable_pointer((D *)o); - // CHECK-NOAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %vtable = load ptr, ptr %0, align 8 - // CHECK-TYPEAUTH: %1 = ptrtoint ptr %vtable to i64 - // CHECK-TYPEAUTH: %2 = call i64 @llvm.ptrauth.auth(i64 %1, i32 2, i64 48388) - // CHECK-TYPEAUTH: %3 = inttoptr i64 %2 to ptr - // CHECK-TYPEAUTH: %4 = load volatile i8, ptr %3, align 8 - // CHECK-ADDRESSAUTH: [[T1:%.*]] = ptrtoint ptr %0 to i64 - // CHECK-ADDRESSAUTH: [[T2:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-ADDRESSAUTH: [[T3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[T2]], i32 2, i64 [[T1]]) - // CHECK-ADDRESSAUTH: [[T4:%.*]] = inttoptr i64 [[T3]] to ptr - // CHECK-ADDRESSAUTH: [[T5:%.*]] = load volatile i8, ptr [[T4]], align 8 - // CHECK-BOTHAUTH: [[T1:%.*]] = ptrtoint ptr %0 to i64 - // CHECK-BOTHAUTH: [[T2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[T1]], i64 48388) - // CHECK-BOTHAUTH: [[T3:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-BOTHAUTH: [[T4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[T3]], i32 2, i64 [[T2]]) - // CHECK-BOTHAUTH: [[T5:%.*]] = inttoptr i64 [[T4]] to ptr - // CHECK-BOTHAUTH: [[T6:%.*]] = load volatile i8, ptr [[T5]], align 8 } +// CHECK-NOAUTH-LABEL: define ptr @aArrayParameter( +// CHECK-NOAUTH-SAME: ptr [[AARRAY:%.*]]) #[[ATTR0]] { +// CHECK-NOAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-NOAUTH-NEXT: [[AARRAY_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-NOAUTH-NEXT: store ptr [[AARRAY]], ptr [[AARRAY_ADDR]], align 8, !tbaa [[TBAA6]] +// CHECK-NOAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AARRAY_ADDR]], align 8, !tbaa [[TBAA6]] +// CHECK-NOAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: ret ptr [[VTABLE]] +// +// CHECK-TYPEAUTH-LABEL: define ptr @aArrayParameter( +// CHECK-TYPEAUTH-SAME: ptr [[AARRAY:%.*]]) #[[ATTR0]] { +// CHECK-TYPEAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-TYPEAUTH-NEXT: [[AARRAY_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-TYPEAUTH-NEXT: store ptr [[AARRAY]], ptr [[AARRAY_ADDR]], align 8, !tbaa [[TBAA6]] +// CHECK-TYPEAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AARRAY_ADDR]], align 8, !tbaa [[TBAA6]] +// CHECK-TYPEAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP1]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP3:%.*]] = inttoptr i64 [[TMP2]] to ptr +// CHECK-TYPEAUTH-NEXT: [[TMP4:%.*]] = load volatile i8, ptr [[TMP3]], align 8 +// CHECK-TYPEAUTH-NEXT: ret ptr [[TMP3]] +// +// CHECK-ADDRESSAUTH-LABEL: define ptr @aArrayParameter( +// CHECK-ADDRESSAUTH-SAME: ptr [[AARRAY:%.*]]) #[[ATTR0]] { +// CHECK-ADDRESSAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-ADDRESSAUTH-NEXT: [[AARRAY_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-ADDRESSAUTH-NEXT: store ptr [[AARRAY]], ptr [[AARRAY_ADDR]], align 8, !tbaa [[TBAA6]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AARRAY_ADDR]], align 8, !tbaa [[TBAA6]] +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP2]], i32 2, i64 [[TMP1]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP4:%.*]] = inttoptr i64 [[TMP3]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[TMP5:%.*]] = load volatile i8, ptr [[TMP4]], align 8 +// CHECK-ADDRESSAUTH-NEXT: ret ptr [[TMP4]] +// +// CHECK-BOTHAUTH-LABEL: define ptr @aArrayParameter( +// CHECK-BOTHAUTH-SAME: ptr [[AARRAY:%.*]]) #[[ATTR0]] { +// CHECK-BOTHAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-BOTHAUTH-NEXT: [[AARRAY_ADDR:%.*]] = alloca ptr, align 8 +// CHECK-BOTHAUTH-NEXT: store ptr [[AARRAY]], ptr [[AARRAY_ADDR]], align 8, !tbaa [[TBAA6]] +// CHECK-BOTHAUTH-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AARRAY_ADDR]], align 8, !tbaa [[TBAA6]] +// CHECK-BOTHAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP1]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP3:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP3]], i32 2, i64 [[TMP2]]) +// CHECK-BOTHAUTH-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP4]] to ptr +// CHECK-BOTHAUTH-NEXT: [[TMP6:%.*]] = load volatile i8, ptr [[TMP5]], align 8 +// CHECK-BOTHAUTH-NEXT: ret ptr [[TMP5]] +// extern "C" const void *aArrayParameter(A aArray[]) { static_assert(same_type<decltype(__builtin_get_vtable_pointer(aArray)), const void*>::value); - // CHECK-NOAUTH: [[THIS_OBJ:%.*]] = load ptr, ptr %aArray.addr - // CHECK-NOAUTH: %vtable = load ptr, ptr [[THIS_OBJ]] - // CHECK-TYPEAUTH: [[THIS_OBJ:%.*]] = load ptr, ptr %aArray.addr - // CHECK-TYPEAUTH: %vtable = load ptr, ptr [[THIS_OBJ]] - // CHECK-TYPEAUTH: [[VTABLEI:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-TYPEAUTH: [[AUTHENTICATED:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[VTABLEI]], i32 2, i64 48388) - // CHECK-ADDRESSAUTH: [[VTABLE_ADDR:%.*]] = load ptr, ptr %aArray.addr, align 8, !tbaa !2 - // CHECK-ADDRESSAUTH: %vtable = load ptr, ptr %0, align 8, !tbaa !7 - // CHECK-ADDRESSAUTH: [[VTABLE_ADDRI:%.*]] = ptrtoint ptr [[VTABLE_ADDR]] to i64 - // CHECK-ADDRESSAUTH: [[VTABLEI:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-ADDRESSAUTH: [[AUTHENTICATED:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[VTABLEI]], i32 2, i64 [[VTABLE_ADDRI]]) - // CHECK-BOTHAUTH: [[VTABLE_ADDR:%.*]] = load ptr, ptr %aArray.addr, align 8, !tbaa !2 - // CHECK-BOTHAUTH: %vtable = load ptr, ptr [[VTABLE_ADDR]], align 8, !tbaa !7 - // CHECK-BOTHAUTH: [[VTABLE_ADDRI:%.*]] = ptrtoint ptr [[VTABLE_ADDR]] to i64 - // CHECK-BOTHAUTH: [[VTABLE_DISC:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[VTABLE_ADDRI]], i64 48388) - // CHECK-BOTHAUTH: [[VTABLE_PTR:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-BOTHAUTH: [[AUTHENTICATED:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[VTABLE_PTR]], i32 2, i64 [[VTABLE_DISC]]) return __builtin_get_vtable_pointer(aArray); } +// CHECK-NOAUTH-LABEL: define ptr @aArrayLocal( +// CHECK-NOAUTH-SAME: ) #[[ATTR0]] { +// CHECK-NOAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-NOAUTH-NEXT: [[ARRAY:%.*]] = alloca [1 x %"struct.test1::A"], align 8 +// CHECK-NOAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[ARRAY]]) #[[ATTR5:[0-9]+]] +// CHECK-NOAUTH-NEXT: call void @_ZN5test11AC1Ev(ptr nonnull align 8 dereferenceable(8) [[ARRAY]]) +// CHECK-NOAUTH-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [1 x %"struct.test1::A"], ptr [[ARRAY]], i64 0, i64 0 +// CHECK-NOAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[ARRAYDECAY]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[ARRAY]]) #[[ATTR5]] +// CHECK-NOAUTH-NEXT: ret ptr [[VTABLE]] +// +// CHECK-TYPEAUTH-LABEL: define ptr @aArrayLocal( +// CHECK-TYPEAUTH-SAME: ) #[[ATTR0]] { +// CHECK-TYPEAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-TYPEAUTH-NEXT: [[ARRAY:%.*]] = alloca [1 x %"struct.test1::A"], align 8 +// CHECK-TYPEAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[ARRAY]]) #[[ATTR7:[0-9]+]] +// CHECK-TYPEAUTH-NEXT: [[CALL:%.*]] = call ptr @_ZN5test11AC1Ev(ptr nonnull align 8 dereferenceable(8) [[ARRAY]]) +// CHECK-TYPEAUTH-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [1 x %"struct.test1::A"], ptr [[ARRAY]], i64 0, i64 0 +// CHECK-TYPEAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[ARRAYDECAY]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP0:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP1:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP0]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP2:%.*]] = inttoptr i64 [[TMP1]] to ptr +// CHECK-TYPEAUTH-NEXT: [[TMP3:%.*]] = load volatile i8, ptr [[TMP2]], align 8 +// CHECK-TYPEAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[ARRAY]]) #[[ATTR7]] +// CHECK-TYPEAUTH-NEXT: ret ptr [[TMP2]] +// +// CHECK-ADDRESSAUTH-LABEL: define ptr @aArrayLocal( +// CHECK-ADDRESSAUTH-SAME: ) #[[ATTR0]] { +// CHECK-ADDRESSAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-ADDRESSAUTH-NEXT: [[ARRAY:%.*]] = alloca [1 x %"struct.test1::A"], align 8 +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[ARRAY]]) #[[ATTR7:[0-9]+]] +// CHECK-ADDRESSAUTH-NEXT: [[CALL:%.*]] = call ptr @_ZN5test11AC1Ev(ptr nonnull align 8 dereferenceable(8) [[ARRAY]]) +// CHECK-ADDRESSAUTH-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [1 x %"struct.test1::A"], ptr [[ARRAY]], i64 0, i64 0 +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[ARRAYDECAY]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP0:%.*]] = ptrtoint ptr [[ARRAYDECAY]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP2:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP1]], i32 2, i64 [[TMP0]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP3:%.*]] = inttoptr i64 [[TMP2]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[TMP4:%.*]] = load volatile i8, ptr [[TMP3]], align 8 +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[ARRAY]]) #[[ATTR7]] +// CHECK-ADDRESSAUTH-NEXT: ret ptr [[TMP3]] +// +// CHECK-BOTHAUTH-LABEL: define ptr @aArrayLocal( +// CHECK-BOTHAUTH-SAME: ) #[[ATTR0]] { +// CHECK-BOTHAUTH-NEXT: [[ENTRY:.*:]] +// CHECK-BOTHAUTH-NEXT: [[ARRAY:%.*]] = alloca [1 x %"struct.test1::A"], align 8 +// CHECK-BOTHAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[ARRAY]]) #[[ATTR7:[0-9]+]] +// CHECK-BOTHAUTH-NEXT: [[CALL:%.*]] = call ptr @_ZN5test11AC1Ev(ptr nonnull align 8 dereferenceable(8) [[ARRAY]]) +// CHECK-BOTHAUTH-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [1 x %"struct.test1::A"], ptr [[ARRAY]], i64 0, i64 0 +// CHECK-BOTHAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[ARRAYDECAY]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP0:%.*]] = ptrtoint ptr [[ARRAYDECAY]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP1:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP0]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP2]], i32 2, i64 [[TMP1]]) +// CHECK-BOTHAUTH-NEXT: [[TMP4:%.*]] = inttoptr i64 [[TMP3]] to ptr +// CHECK-BOTHAUTH-NEXT: [[TMP5:%.*]] = load volatile i8, ptr [[TMP4]], align 8 +// CHECK-BOTHAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[ARRAY]]) #[[ATTR7]] +// CHECK-BOTHAUTH-NEXT: ret ptr [[TMP4]] +// extern "C" const void *aArrayLocal() { A array[] = { A() }; static_assert(same_type<decltype(__builtin_get_vtable_pointer(array)), const void*>::value); - // CHECK-NOAUTH: [[THIS_OBJ:%.*]] = getelementptr inbounds [1 x %"struct.test1::A"], ptr %array - // CHECK-NOAUTH: %vtable = load ptr, ptr %arraydecay - // CHECK-TYPEAUTH: %arraydecay = getelementptr inbounds [1 x %"struct.test1::A"] - // CHECK-TYPEAUTH: %vtable = load ptr, ptr %arraydecay - // CHECK-TYPEAUTH: [[VTABLEI:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-TYPEAUTH: [[AUTHENTICATED:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[VTABLEI]], i32 2, i64 48388) - // CHECK-ADDRESSAUTH: %arraydecay = getelementptr inbounds [1 x %"struct.test1::A"], ptr %array, i64 0, i64 0 - // CHECK-ADDRESSAUTH: %vtable = load ptr, ptr %arraydecay, align 8, !tbaa !7 - // CHECK-ADDRESSAUTH: [[VTABLE_ADDRI:%.*]] = ptrtoint ptr %arraydecay to i64 - // CHECK-ADDRESSAUTH: [[VTABLEI:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-ADDRESSAUTH: [[AUTHENTICATED:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[VTABLEI]], i32 2, i64 [[VTABLE_ADDRI]]) - // CHECK-BOTHAUTH: %arraydecay = getelementptr inbounds [1 x %"struct.test1::A"], ptr %array, i64 0, i64 0 - // CHECK-BOTHAUTH: %vtable = load ptr, ptr %arraydecay, align 8, !tbaa !7 - // CHECK-BOTHAUTH: [[VTABLE_ADDRI:%.*]] = ptrtoint ptr %arraydecay to i64 - // CHECK-BOTHAUTH: [[VTABLE_DISC:%.*]] = call i64 @llvm.ptrauth.blend(i64 %0, i64 48388) - // CHECK-BOTHAUTH: [[VTABLEI:%.*]] = ptrtoint ptr %vtable to i64 - // CHECK-BOTHAUTH: [[AUTHENTICATED:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[VTABLEI]], i32 2, i64 [[VTABLE_DISC]]) return __builtin_get_vtable_pointer(array); } +// CHECK-NOAUTH-LABEL: define void @_ZN5test14testEv( +// CHECK-NOAUTH-SAME: ) #[[ATTR0]] { +// CHECK-NOAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-NOAUTH-NEXT: [[AINSTANCE:%.*]] = alloca %"struct.test1::A", align 8 +// CHECK-NOAUTH-NEXT: [[BINSTANCE:%.*]] = alloca %"struct.test1::B", align 8 +// CHECK-NOAUTH-NEXT: [[CINSTANCE:%.*]] = alloca %"struct.test1::C", align 8 +// CHECK-NOAUTH-NEXT: [[DINSTANCE:%.*]] = alloca %"struct.test1::D", align 8 +// CHECK-NOAUTH-NEXT: [[EINSTANCE:%.*]] = alloca %"struct.test1::E", align 8 +// CHECK-NOAUTH-NEXT: [[EARRAY:%.*]] = alloca [1 x %"struct.test1::E"], align 16 +// CHECK-NOAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[AINSTANCE]]) #[[ATTR5]] +// CHECK-NOAUTH-NEXT: call void @_ZN5test11AC1Ev(ptr nonnull align 8 dereferenceable(8) [[AINSTANCE]]) +// CHECK-NOAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[BINSTANCE]]) #[[ATTR5]] +// CHECK-NOAUTH-NEXT: call void @_ZN5test11BC1Ev(ptr nonnull align 8 dereferenceable(8) [[BINSTANCE]]) +// CHECK-NOAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[CINSTANCE]]) #[[ATTR5]] +// CHECK-NOAUTH-NEXT: call void @_ZN5test11CC1Ev(ptr nonnull align 8 dereferenceable(16) [[CINSTANCE]]) +// CHECK-NOAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[DINSTANCE]]) #[[ATTR5]] +// CHECK-NOAUTH-NEXT: call void @_ZN5test11DC1Ev(ptr nonnull align 8 dereferenceable(8) [[DINSTANCE]]) +// CHECK-NOAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[EINSTANCE]]) #[[ATTR5]] +// CHECK-NOAUTH-NEXT: call void @_ZN5test11EC1Ev(ptr nonnull align 8 dereferenceable(16) [[EINSTANCE]]) +// CHECK-NOAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[EARRAY]]) #[[ATTR5]] +// CHECK-NOAUTH-NEXT: call void @llvm.memset.p0.i64(ptr align 16 [[EARRAY]], i8 0, i64 16, i1 false) +// CHECK-NOAUTH-NEXT: call void @_ZN5test11EC1Ev(ptr nonnull align 8 dereferenceable(16) [[EARRAY]]) +// CHECK-NOAUTH-NEXT: [[CALL:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[AINSTANCE]]) +// CHECK-NOAUTH-NEXT: [[CALL1:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[BINSTANCE]]) +// CHECK-NOAUTH-NEXT: [[TMP0:%.*]] = icmp eq ptr [[CINSTANCE]], null +// CHECK-NOAUTH-NEXT: br i1 [[TMP0]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-NOAUTH: [[CAST_NOTNULL]]: +// CHECK-NOAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[CINSTANCE]], i64 8 +// CHECK-NOAUTH-NEXT: br label %[[CAST_END]] +// CHECK-NOAUTH: [[CAST_END]]: +// CHECK-NOAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-NOAUTH-NEXT: [[CALL2:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT]]) +// CHECK-NOAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[DINSTANCE]], null +// CHECK-NOAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END5:.*]], label %[[CAST_NOTNULL3:.*]] +// CHECK-NOAUTH: [[CAST_NOTNULL3]]: +// CHECK-NOAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[DINSTANCE]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: [[VBASE_OFFSET_PTR:%.*]] = getelementptr i8, ptr [[VTABLE]], i64 -32 +// CHECK-NOAUTH-NEXT: [[VBASE_OFFSET:%.*]] = load i64, ptr [[VBASE_OFFSET_PTR]], align 8 +// CHECK-NOAUTH-NEXT: [[ADD_PTR4:%.*]] = getelementptr inbounds i8, ptr [[DINSTANCE]], i64 [[VBASE_OFFSET]] +// CHECK-NOAUTH-NEXT: br label %[[CAST_END5]] +// CHECK-NOAUTH: [[CAST_END5]]: +// CHECK-NOAUTH-NEXT: [[CAST_RESULT6:%.*]] = phi ptr [ [[ADD_PTR4]], %[[CAST_NOTNULL3]] ], [ null, %[[CAST_END]] ] +// CHECK-NOAUTH-NEXT: [[CALL7:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT6]]) +// CHECK-NOAUTH-NEXT: [[TMP2:%.*]] = icmp eq ptr [[EINSTANCE]], null +// CHECK-NOAUTH-NEXT: br i1 [[TMP2]], label %[[CAST_END13:.*]], label %[[CAST_NOTNULL8:.*]] +// CHECK-NOAUTH: [[CAST_NOTNULL8]]: +// CHECK-NOAUTH-NEXT: [[VTABLE9:%.*]] = load ptr, ptr [[EINSTANCE]], align 8, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: [[VBASE_OFFSET_PTR10:%.*]] = getelementptr i8, ptr [[VTABLE9]], i64 -32 +// CHECK-NOAUTH-NEXT: [[VBASE_OFFSET11:%.*]] = load i64, ptr [[VBASE_OFFSET_PTR10]], align 8 +// CHECK-NOAUTH-NEXT: [[ADD_PTR12:%.*]] = getelementptr inbounds i8, ptr [[EINSTANCE]], i64 [[VBASE_OFFSET11]] +// CHECK-NOAUTH-NEXT: br label %[[CAST_END13]] +// CHECK-NOAUTH: [[CAST_END13]]: +// CHECK-NOAUTH-NEXT: [[CAST_RESULT14:%.*]] = phi ptr [ [[ADD_PTR12]], %[[CAST_NOTNULL8]] ], [ null, %[[CAST_END5]] ] +// CHECK-NOAUTH-NEXT: [[CALL15:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT14]]) +// CHECK-NOAUTH-NEXT: [[TMP3:%.*]] = icmp eq ptr [[EINSTANCE]], null +// CHECK-NOAUTH-NEXT: br i1 [[TMP3]], label %[[CAST_END18:.*]], label %[[CAST_NOTNULL16:.*]] +// CHECK-NOAUTH: [[CAST_NOTNULL16]]: +// CHECK-NOAUTH-NEXT: [[ADD_PTR17:%.*]] = getelementptr inbounds i8, ptr [[EINSTANCE]], i64 8 +// CHECK-NOAUTH-NEXT: br label %[[CAST_END18]] +// CHECK-NOAUTH: [[CAST_END18]]: +// CHECK-NOAUTH-NEXT: [[CAST_RESULT19:%.*]] = phi ptr [ [[ADD_PTR17]], %[[CAST_NOTNULL16]] ], [ null, %[[CAST_END13]] ] +// CHECK-NOAUTH-NEXT: [[CALL20:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT19]]) +// CHECK-NOAUTH-NEXT: [[CALL21:%.*]] = call ptr @_ZN5test11bEPNS_1BE(ptr [[BINSTANCE]]) +// CHECK-NOAUTH-NEXT: [[TMP4:%.*]] = icmp eq ptr [[CINSTANCE]], null +// CHECK-NOAUTH-NEXT: br i1 [[TMP4]], label %[[CAST_END24:.*]], label %[[CAST_NOTNULL22:.*]] +// CHECK-NOAUTH: [[CAST_NOTNULL22]]: +// CHECK-NOAUTH-NEXT: [[ADD_PTR23:%.*]] = getelementptr inbounds i8, ptr [[CINSTANCE]], i64 8 +// CHECK-NOAUTH-NEXT: br label %[[CAST_END24]] +// CHECK-NOAUTH: [[CAST_END24]]: +// CHECK-NOAUTH-NEXT: [[CAST_RESULT25:%.*]] = phi ptr [ [[ADD_PTR23]], %[[CAST_NOTNULL22]] ], [ null, %[[CAST_END18]] ] +// CHECK-NOAUTH-NEXT: [[CALL26:%.*]] = call ptr @_ZN5test11bEPNS_1BE(ptr [[CAST_RESULT25]]) +// CHECK-NOAUTH-NEXT: [[TMP5:%.*]] = icmp eq ptr [[EINSTANCE]], null +// CHECK-NOAUTH-NEXT: br i1 [[TMP5]], label %[[CAST_END29:.*]], label %[[CAST_NOTNULL27:.*]] +// CHECK-NOAUTH: [[CAST_NOTNULL27]]: +// CHECK-NOAUTH-NEXT: [[ADD_PTR28:%.*]] = getelementptr inbounds i8, ptr [[EINSTANCE]], i64 8 +// CHECK-NOAUTH-NEXT: br label %[[CAST_END29]] +// CHECK-NOAUTH: [[CAST_END29]]: +// CHECK-NOAUTH-NEXT: [[CAST_RESULT30:%.*]] = phi ptr [ [[ADD_PTR28]], %[[CAST_NOTNULL27]] ], [ null, %[[CAST_END24]] ] +// CHECK-NOAUTH-NEXT: [[CALL31:%.*]] = call ptr @_ZN5test11bEPNS_1BE(ptr [[CAST_RESULT30]]) +// CHECK-NOAUTH-NEXT: [[CALL32:%.*]] = call ptr @_ZN5test16b_as_AEPNS_1BE(ptr [[BINSTANCE]]) +// CHECK-NOAUTH-NEXT: [[CALL33:%.*]] = call ptr @_ZN5test11cEPNS_1CE(ptr [[CINSTANCE]]) +// CHECK-NOAUTH-NEXT: [[CALL34:%.*]] = call ptr @_ZN5test16c_as_ZEPNS_1CE(ptr [[CINSTANCE]]) +// CHECK-NOAUTH-NEXT: [[CALL35:%.*]] = call ptr @_ZN5test16c_as_BEPNS_1CE(ptr [[CINSTANCE]]) +// CHECK-NOAUTH-NEXT: [[CALL36:%.*]] = call ptr @_ZN5test11dEPNS_1DE(ptr [[DINSTANCE]]) +// CHECK-NOAUTH-NEXT: [[CALL37:%.*]] = call ptr @_ZN5test11dEPNS_1DE(ptr [[EINSTANCE]]) +// CHECK-NOAUTH-NEXT: [[CALL38:%.*]] = call ptr @_ZN5test16d_as_AEPNS_1DE(ptr [[DINSTANCE]]) +// CHECK-NOAUTH-NEXT: [[CALL39:%.*]] = call ptr @_ZN5test16d_as_AEPNS_1DE(ptr [[EINSTANCE]]) +// CHECK-NOAUTH-NEXT: [[CALL40:%.*]] = call ptr @_ZN5test11eEPNS_1EE(ptr [[EINSTANCE]]) +// CHECK-NOAUTH-NEXT: [[CALL41:%.*]] = call ptr @_ZN5test16e_as_BEPNS_1EE(ptr [[EINSTANCE]]) +// CHECK-NOAUTH-NEXT: [[CALL42:%.*]] = call ptr @_ZN5test16e_as_DEPNS_1EE(ptr [[EINSTANCE]]) +// CHECK-NOAUTH-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [1 x %"struct.test1::E"], ptr [[EARRAY]], i64 0, i64 0 +// CHECK-NOAUTH-NEXT: [[VTABLE43:%.*]] = load ptr, ptr [[ARRAYDECAY]], align 16, !tbaa [[TBAA9]] +// CHECK-NOAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[EARRAY]]) #[[ATTR5]] +// CHECK-NOAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[EINSTANCE]]) #[[ATTR5]] +// CHECK-NOAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[DINSTANCE]]) #[[ATTR5]] +// CHECK-NOAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[CINSTANCE]]) #[[ATTR5]] +// CHECK-NOAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[BINSTANCE]]) #[[ATTR5]] +// CHECK-NOAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[AINSTANCE]]) #[[ATTR5]] +// CHECK-NOAUTH-NEXT: ret void +// +// CHECK-TYPEAUTH-LABEL: define void @_ZN5test14testEv( +// CHECK-TYPEAUTH-SAME: ) #[[ATTR0]] { +// CHECK-TYPEAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-TYPEAUTH-NEXT: [[AINSTANCE:%.*]] = alloca %"struct.test1::A", align 8 +// CHECK-TYPEAUTH-NEXT: [[BINSTANCE:%.*]] = alloca %"struct.test1::B", align 8 +// CHECK-TYPEAUTH-NEXT: [[CINSTANCE:%.*]] = alloca %"struct.test1::C", align 8 +// CHECK-TYPEAUTH-NEXT: [[DINSTANCE:%.*]] = alloca %"struct.test1::D", align 8 +// CHECK-TYPEAUTH-NEXT: [[EINSTANCE:%.*]] = alloca %"struct.test1::E", align 8 +// CHECK-TYPEAUTH-NEXT: [[EARRAY:%.*]] = alloca [1 x %"struct.test1::E"], align 8 +// CHECK-TYPEAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[AINSTANCE]]) #[[ATTR7]] +// CHECK-TYPEAUTH-NEXT: [[CALL:%.*]] = call ptr @_ZN5test11AC1Ev(ptr nonnull align 8 dereferenceable(8) [[AINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[BINSTANCE]]) #[[ATTR7]] +// CHECK-TYPEAUTH-NEXT: [[CALL1:%.*]] = call ptr @_ZN5test11BC1Ev(ptr nonnull align 8 dereferenceable(8) [[BINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[CINSTANCE]]) #[[ATTR7]] +// CHECK-TYPEAUTH-NEXT: [[CALL2:%.*]] = call ptr @_ZN5test11CC1Ev(ptr nonnull align 8 dereferenceable(16) [[CINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[DINSTANCE]]) #[[ATTR7]] +// CHECK-TYPEAUTH-NEXT: [[CALL3:%.*]] = call ptr @_ZN5test11DC1Ev(ptr nonnull align 8 dereferenceable(8) [[DINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[EINSTANCE]]) #[[ATTR7]] +// CHECK-TYPEAUTH-NEXT: [[CALL4:%.*]] = call ptr @_ZN5test11EC1Ev(ptr nonnull align 8 dereferenceable(16) [[EINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[EARRAY]]) #[[ATTR7]] +// CHECK-TYPEAUTH-NEXT: call void @llvm.memset.p0.i64(ptr align 8 [[EARRAY]], i8 0, i64 16, i1 false) +// CHECK-TYPEAUTH-NEXT: [[CALL5:%.*]] = call ptr @_ZN5test11EC1Ev(ptr nonnull align 8 dereferenceable(16) [[EARRAY]]) +// CHECK-TYPEAUTH-NEXT: [[CALL6:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[AINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: [[CALL7:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[BINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: [[TMP0:%.*]] = icmp eq ptr [[CINSTANCE]], null +// CHECK-TYPEAUTH-NEXT: br i1 [[TMP0]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-TYPEAUTH: [[CAST_NOTNULL]]: +// CHECK-TYPEAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[CINSTANCE]], i64 8 +// CHECK-TYPEAUTH-NEXT: br label %[[CAST_END]] +// CHECK-TYPEAUTH: [[CAST_END]]: +// CHECK-TYPEAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-TYPEAUTH-NEXT: [[CALL8:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT]]) +// CHECK-TYPEAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[DINSTANCE]], null +// CHECK-TYPEAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END11:.*]], label %[[CAST_NOTNULL9:.*]] +// CHECK-TYPEAUTH: [[CAST_NOTNULL9]]: +// CHECK-TYPEAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[DINSTANCE]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP2]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP4:%.*]] = inttoptr i64 [[TMP3]] to ptr +// CHECK-TYPEAUTH-NEXT: [[VBASE_OFFSET_PTR:%.*]] = getelementptr i8, ptr [[TMP4]], i64 -32 +// CHECK-TYPEAUTH-NEXT: [[VBASE_OFFSET:%.*]] = load i64, ptr [[VBASE_OFFSET_PTR]], align 8 +// CHECK-TYPEAUTH-NEXT: [[ADD_PTR10:%.*]] = getelementptr inbounds i8, ptr [[DINSTANCE]], i64 [[VBASE_OFFSET]] +// CHECK-TYPEAUTH-NEXT: br label %[[CAST_END11]] +// CHECK-TYPEAUTH: [[CAST_END11]]: +// CHECK-TYPEAUTH-NEXT: [[CAST_RESULT12:%.*]] = phi ptr [ [[ADD_PTR10]], %[[CAST_NOTNULL9]] ], [ null, %[[CAST_END]] ] +// CHECK-TYPEAUTH-NEXT: [[CALL13:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT12]]) +// CHECK-TYPEAUTH-NEXT: [[TMP5:%.*]] = icmp eq ptr [[EINSTANCE]], null +// CHECK-TYPEAUTH-NEXT: br i1 [[TMP5]], label %[[CAST_END19:.*]], label %[[CAST_NOTNULL14:.*]] +// CHECK-TYPEAUTH: [[CAST_NOTNULL14]]: +// CHECK-TYPEAUTH-NEXT: [[VTABLE15:%.*]] = load ptr, ptr [[EINSTANCE]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP6:%.*]] = ptrtoint ptr [[VTABLE15]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP7:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP6]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP8:%.*]] = inttoptr i64 [[TMP7]] to ptr +// CHECK-TYPEAUTH-NEXT: [[VBASE_OFFSET_PTR16:%.*]] = getelementptr i8, ptr [[TMP8]], i64 -32 +// CHECK-TYPEAUTH-NEXT: [[VBASE_OFFSET17:%.*]] = load i64, ptr [[VBASE_OFFSET_PTR16]], align 8 +// CHECK-TYPEAUTH-NEXT: [[ADD_PTR18:%.*]] = getelementptr inbounds i8, ptr [[EINSTANCE]], i64 [[VBASE_OFFSET17]] +// CHECK-TYPEAUTH-NEXT: br label %[[CAST_END19]] +// CHECK-TYPEAUTH: [[CAST_END19]]: +// CHECK-TYPEAUTH-NEXT: [[CAST_RESULT20:%.*]] = phi ptr [ [[ADD_PTR18]], %[[CAST_NOTNULL14]] ], [ null, %[[CAST_END11]] ] +// CHECK-TYPEAUTH-NEXT: [[CALL21:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT20]]) +// CHECK-TYPEAUTH-NEXT: [[TMP9:%.*]] = icmp eq ptr [[EINSTANCE]], null +// CHECK-TYPEAUTH-NEXT: br i1 [[TMP9]], label %[[CAST_END24:.*]], label %[[CAST_NOTNULL22:.*]] +// CHECK-TYPEAUTH: [[CAST_NOTNULL22]]: +// CHECK-TYPEAUTH-NEXT: [[ADD_PTR23:%.*]] = getelementptr inbounds i8, ptr [[EINSTANCE]], i64 8 +// CHECK-TYPEAUTH-NEXT: br label %[[CAST_END24]] +// CHECK-TYPEAUTH: [[CAST_END24]]: +// CHECK-TYPEAUTH-NEXT: [[CAST_RESULT25:%.*]] = phi ptr [ [[ADD_PTR23]], %[[CAST_NOTNULL22]] ], [ null, %[[CAST_END19]] ] +// CHECK-TYPEAUTH-NEXT: [[CALL26:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT25]]) +// CHECK-TYPEAUTH-NEXT: [[CALL27:%.*]] = call ptr @_ZN5test11bEPNS_1BE(ptr [[BINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: [[TMP10:%.*]] = icmp eq ptr [[CINSTANCE]], null +// CHECK-TYPEAUTH-NEXT: br i1 [[TMP10]], label %[[CAST_END30:.*]], label %[[CAST_NOTNULL28:.*]] +// CHECK-TYPEAUTH: [[CAST_NOTNULL28]]: +// CHECK-TYPEAUTH-NEXT: [[ADD_PTR29:%.*]] = getelementptr inbounds i8, ptr [[CINSTANCE]], i64 8 +// CHECK-TYPEAUTH-NEXT: br label %[[CAST_END30]] +// CHECK-TYPEAUTH: [[CAST_END30]]: +// CHECK-TYPEAUTH-NEXT: [[CAST_RESULT31:%.*]] = phi ptr [ [[ADD_PTR29]], %[[CAST_NOTNULL28]] ], [ null, %[[CAST_END24]] ] +// CHECK-TYPEAUTH-NEXT: [[CALL32:%.*]] = call ptr @_ZN5test11bEPNS_1BE(ptr [[CAST_RESULT31]]) +// CHECK-TYPEAUTH-NEXT: [[TMP11:%.*]] = icmp eq ptr [[EINSTANCE]], null +// CHECK-TYPEAUTH-NEXT: br i1 [[TMP11]], label %[[CAST_END35:.*]], label %[[CAST_NOTNULL33:.*]] +// CHECK-TYPEAUTH: [[CAST_NOTNULL33]]: +// CHECK-TYPEAUTH-NEXT: [[ADD_PTR34:%.*]] = getelementptr inbounds i8, ptr [[EINSTANCE]], i64 8 +// CHECK-TYPEAUTH-NEXT: br label %[[CAST_END35]] +// CHECK-TYPEAUTH: [[CAST_END35]]: +// CHECK-TYPEAUTH-NEXT: [[CAST_RESULT36:%.*]] = phi ptr [ [[ADD_PTR34]], %[[CAST_NOTNULL33]] ], [ null, %[[CAST_END30]] ] +// CHECK-TYPEAUTH-NEXT: [[CALL37:%.*]] = call ptr @_ZN5test11bEPNS_1BE(ptr [[CAST_RESULT36]]) +// CHECK-TYPEAUTH-NEXT: [[CALL38:%.*]] = call ptr @_ZN5test16b_as_AEPNS_1BE(ptr [[BINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: [[CALL39:%.*]] = call ptr @_ZN5test11cEPNS_1CE(ptr [[CINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: [[CALL40:%.*]] = call ptr @_ZN5test16c_as_ZEPNS_1CE(ptr [[CINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: [[CALL41:%.*]] = call ptr @_ZN5test16c_as_BEPNS_1CE(ptr [[CINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: [[CALL42:%.*]] = call ptr @_ZN5test11dEPNS_1DE(ptr [[DINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: [[CALL43:%.*]] = call ptr @_ZN5test11dEPNS_1DE(ptr [[EINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: [[CALL44:%.*]] = call ptr @_ZN5test16d_as_AEPNS_1DE(ptr [[DINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: [[CALL45:%.*]] = call ptr @_ZN5test16d_as_AEPNS_1DE(ptr [[EINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: [[CALL46:%.*]] = call ptr @_ZN5test11eEPNS_1EE(ptr [[EINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: [[CALL47:%.*]] = call ptr @_ZN5test16e_as_BEPNS_1EE(ptr [[EINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: [[CALL48:%.*]] = call ptr @_ZN5test16e_as_DEPNS_1EE(ptr [[EINSTANCE]]) +// CHECK-TYPEAUTH-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [1 x %"struct.test1::E"], ptr [[EARRAY]], i64 0, i64 0 +// CHECK-TYPEAUTH-NEXT: [[VTABLE49:%.*]] = load ptr, ptr [[ARRAYDECAY]], align 8, !tbaa [[TBAA9]] +// CHECK-TYPEAUTH-NEXT: [[TMP12:%.*]] = ptrtoint ptr [[VTABLE49]] to i64 +// CHECK-TYPEAUTH-NEXT: [[TMP13:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP12]], i32 2, i64 48388) +// CHECK-TYPEAUTH-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP13]] to ptr +// CHECK-TYPEAUTH-NEXT: [[TMP15:%.*]] = load volatile i8, ptr [[TMP14]], align 8 +// CHECK-TYPEAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[EARRAY]]) #[[ATTR7]] +// CHECK-TYPEAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[EINSTANCE]]) #[[ATTR7]] +// CHECK-TYPEAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[DINSTANCE]]) #[[ATTR7]] +// CHECK-TYPEAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[CINSTANCE]]) #[[ATTR7]] +// CHECK-TYPEAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[BINSTANCE]]) #[[ATTR7]] +// CHECK-TYPEAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[AINSTANCE]]) #[[ATTR7]] +// CHECK-TYPEAUTH-NEXT: ret void +// +// CHECK-ADDRESSAUTH-LABEL: define void @_ZN5test14testEv( +// CHECK-ADDRESSAUTH-SAME: ) #[[ATTR0]] { +// CHECK-ADDRESSAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-ADDRESSAUTH-NEXT: [[AINSTANCE:%.*]] = alloca %"struct.test1::A", align 8 +// CHECK-ADDRESSAUTH-NEXT: [[BINSTANCE:%.*]] = alloca %"struct.test1::B", align 8 +// CHECK-ADDRESSAUTH-NEXT: [[CINSTANCE:%.*]] = alloca %"struct.test1::C", align 8 +// CHECK-ADDRESSAUTH-NEXT: [[DINSTANCE:%.*]] = alloca %"struct.test1::D", align 8 +// CHECK-ADDRESSAUTH-NEXT: [[EINSTANCE:%.*]] = alloca %"struct.test1::E", align 8 +// CHECK-ADDRESSAUTH-NEXT: [[EARRAY:%.*]] = alloca [1 x %"struct.test1::E"], align 8 +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[AINSTANCE]]) #[[ATTR7]] +// CHECK-ADDRESSAUTH-NEXT: [[CALL:%.*]] = call ptr @_ZN5test11AC1Ev(ptr nonnull align 8 dereferenceable(8) [[AINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[BINSTANCE]]) #[[ATTR7]] +// CHECK-ADDRESSAUTH-NEXT: [[CALL1:%.*]] = call ptr @_ZN5test11BC1Ev(ptr nonnull align 8 dereferenceable(8) [[BINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[CINSTANCE]]) #[[ATTR7]] +// CHECK-ADDRESSAUTH-NEXT: [[CALL2:%.*]] = call ptr @_ZN5test11CC1Ev(ptr nonnull align 8 dereferenceable(16) [[CINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[DINSTANCE]]) #[[ATTR7]] +// CHECK-ADDRESSAUTH-NEXT: [[CALL3:%.*]] = call ptr @_ZN5test11DC1Ev(ptr nonnull align 8 dereferenceable(8) [[DINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[EINSTANCE]]) #[[ATTR7]] +// CHECK-ADDRESSAUTH-NEXT: [[CALL4:%.*]] = call ptr @_ZN5test11EC1Ev(ptr nonnull align 8 dereferenceable(16) [[EINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[EARRAY]]) #[[ATTR7]] +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.memset.p0.i64(ptr align 8 [[EARRAY]], i8 0, i64 16, i1 false) +// CHECK-ADDRESSAUTH-NEXT: [[CALL5:%.*]] = call ptr @_ZN5test11EC1Ev(ptr nonnull align 8 dereferenceable(16) [[EARRAY]]) +// CHECK-ADDRESSAUTH-NEXT: [[CALL6:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[AINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: [[CALL7:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[BINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP0:%.*]] = icmp eq ptr [[CINSTANCE]], null +// CHECK-ADDRESSAUTH-NEXT: br i1 [[TMP0]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-ADDRESSAUTH: [[CAST_NOTNULL]]: +// CHECK-ADDRESSAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[CINSTANCE]], i64 8 +// CHECK-ADDRESSAUTH-NEXT: br label %[[CAST_END]] +// CHECK-ADDRESSAUTH: [[CAST_END]]: +// CHECK-ADDRESSAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-ADDRESSAUTH-NEXT: [[CALL8:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[DINSTANCE]], null +// CHECK-ADDRESSAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END11:.*]], label %[[CAST_NOTNULL9:.*]] +// CHECK-ADDRESSAUTH: [[CAST_NOTNULL9]]: +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[DINSTANCE]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[DINSTANCE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP3:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP4:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP3]], i32 2, i64 [[TMP2]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP5:%.*]] = inttoptr i64 [[TMP4]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[VBASE_OFFSET_PTR:%.*]] = getelementptr i8, ptr [[TMP5]], i64 -32 +// CHECK-ADDRESSAUTH-NEXT: [[VBASE_OFFSET:%.*]] = load i64, ptr [[VBASE_OFFSET_PTR]], align 8 +// CHECK-ADDRESSAUTH-NEXT: [[ADD_PTR10:%.*]] = getelementptr inbounds i8, ptr [[DINSTANCE]], i64 [[VBASE_OFFSET]] +// CHECK-ADDRESSAUTH-NEXT: br label %[[CAST_END11]] +// CHECK-ADDRESSAUTH: [[CAST_END11]]: +// CHECK-ADDRESSAUTH-NEXT: [[CAST_RESULT12:%.*]] = phi ptr [ [[ADD_PTR10]], %[[CAST_NOTNULL9]] ], [ null, %[[CAST_END]] ] +// CHECK-ADDRESSAUTH-NEXT: [[CALL13:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT12]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP6:%.*]] = icmp eq ptr [[EINSTANCE]], null +// CHECK-ADDRESSAUTH-NEXT: br i1 [[TMP6]], label %[[CAST_END19:.*]], label %[[CAST_NOTNULL14:.*]] +// CHECK-ADDRESSAUTH: [[CAST_NOTNULL14]]: +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE15:%.*]] = load ptr, ptr [[EINSTANCE]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP7:%.*]] = ptrtoint ptr [[EINSTANCE]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP8:%.*]] = ptrtoint ptr [[VTABLE15]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP9:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP8]], i32 2, i64 [[TMP7]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP10:%.*]] = inttoptr i64 [[TMP9]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[VBASE_OFFSET_PTR16:%.*]] = getelementptr i8, ptr [[TMP10]], i64 -32 +// CHECK-ADDRESSAUTH-NEXT: [[VBASE_OFFSET17:%.*]] = load i64, ptr [[VBASE_OFFSET_PTR16]], align 8 +// CHECK-ADDRESSAUTH-NEXT: [[ADD_PTR18:%.*]] = getelementptr inbounds i8, ptr [[EINSTANCE]], i64 [[VBASE_OFFSET17]] +// CHECK-ADDRESSAUTH-NEXT: br label %[[CAST_END19]] +// CHECK-ADDRESSAUTH: [[CAST_END19]]: +// CHECK-ADDRESSAUTH-NEXT: [[CAST_RESULT20:%.*]] = phi ptr [ [[ADD_PTR18]], %[[CAST_NOTNULL14]] ], [ null, %[[CAST_END11]] ] +// CHECK-ADDRESSAUTH-NEXT: [[CALL21:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT20]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP11:%.*]] = icmp eq ptr [[EINSTANCE]], null +// CHECK-ADDRESSAUTH-NEXT: br i1 [[TMP11]], label %[[CAST_END24:.*]], label %[[CAST_NOTNULL22:.*]] +// CHECK-ADDRESSAUTH: [[CAST_NOTNULL22]]: +// CHECK-ADDRESSAUTH-NEXT: [[ADD_PTR23:%.*]] = getelementptr inbounds i8, ptr [[EINSTANCE]], i64 8 +// CHECK-ADDRESSAUTH-NEXT: br label %[[CAST_END24]] +// CHECK-ADDRESSAUTH: [[CAST_END24]]: +// CHECK-ADDRESSAUTH-NEXT: [[CAST_RESULT25:%.*]] = phi ptr [ [[ADD_PTR23]], %[[CAST_NOTNULL22]] ], [ null, %[[CAST_END19]] ] +// CHECK-ADDRESSAUTH-NEXT: [[CALL26:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT25]]) +// CHECK-ADDRESSAUTH-NEXT: [[CALL27:%.*]] = call ptr @_ZN5test11bEPNS_1BE(ptr [[BINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP12:%.*]] = icmp eq ptr [[CINSTANCE]], null +// CHECK-ADDRESSAUTH-NEXT: br i1 [[TMP12]], label %[[CAST_END30:.*]], label %[[CAST_NOTNULL28:.*]] +// CHECK-ADDRESSAUTH: [[CAST_NOTNULL28]]: +// CHECK-ADDRESSAUTH-NEXT: [[ADD_PTR29:%.*]] = getelementptr inbounds i8, ptr [[CINSTANCE]], i64 8 +// CHECK-ADDRESSAUTH-NEXT: br label %[[CAST_END30]] +// CHECK-ADDRESSAUTH: [[CAST_END30]]: +// CHECK-ADDRESSAUTH-NEXT: [[CAST_RESULT31:%.*]] = phi ptr [ [[ADD_PTR29]], %[[CAST_NOTNULL28]] ], [ null, %[[CAST_END24]] ] +// CHECK-ADDRESSAUTH-NEXT: [[CALL32:%.*]] = call ptr @_ZN5test11bEPNS_1BE(ptr [[CAST_RESULT31]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP13:%.*]] = icmp eq ptr [[EINSTANCE]], null +// CHECK-ADDRESSAUTH-NEXT: br i1 [[TMP13]], label %[[CAST_END35:.*]], label %[[CAST_NOTNULL33:.*]] +// CHECK-ADDRESSAUTH: [[CAST_NOTNULL33]]: +// CHECK-ADDRESSAUTH-NEXT: [[ADD_PTR34:%.*]] = getelementptr inbounds i8, ptr [[EINSTANCE]], i64 8 +// CHECK-ADDRESSAUTH-NEXT: br label %[[CAST_END35]] +// CHECK-ADDRESSAUTH: [[CAST_END35]]: +// CHECK-ADDRESSAUTH-NEXT: [[CAST_RESULT36:%.*]] = phi ptr [ [[ADD_PTR34]], %[[CAST_NOTNULL33]] ], [ null, %[[CAST_END30]] ] +// CHECK-ADDRESSAUTH-NEXT: [[CALL37:%.*]] = call ptr @_ZN5test11bEPNS_1BE(ptr [[CAST_RESULT36]]) +// CHECK-ADDRESSAUTH-NEXT: [[CALL38:%.*]] = call ptr @_ZN5test16b_as_AEPNS_1BE(ptr [[BINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: [[CALL39:%.*]] = call ptr @_ZN5test11cEPNS_1CE(ptr [[CINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: [[CALL40:%.*]] = call ptr @_ZN5test16c_as_ZEPNS_1CE(ptr [[CINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: [[CALL41:%.*]] = call ptr @_ZN5test16c_as_BEPNS_1CE(ptr [[CINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: [[CALL42:%.*]] = call ptr @_ZN5test11dEPNS_1DE(ptr [[DINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: [[CALL43:%.*]] = call ptr @_ZN5test11dEPNS_1DE(ptr [[EINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: [[CALL44:%.*]] = call ptr @_ZN5test16d_as_AEPNS_1DE(ptr [[DINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: [[CALL45:%.*]] = call ptr @_ZN5test16d_as_AEPNS_1DE(ptr [[EINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: [[CALL46:%.*]] = call ptr @_ZN5test11eEPNS_1EE(ptr [[EINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: [[CALL47:%.*]] = call ptr @_ZN5test16e_as_BEPNS_1EE(ptr [[EINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: [[CALL48:%.*]] = call ptr @_ZN5test16e_as_DEPNS_1EE(ptr [[EINSTANCE]]) +// CHECK-ADDRESSAUTH-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [1 x %"struct.test1::E"], ptr [[EARRAY]], i64 0, i64 0 +// CHECK-ADDRESSAUTH-NEXT: [[VTABLE49:%.*]] = load ptr, ptr [[ARRAYDECAY]], align 8, !tbaa [[TBAA9]] +// CHECK-ADDRESSAUTH-NEXT: [[TMP14:%.*]] = ptrtoint ptr [[ARRAYDECAY]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP15:%.*]] = ptrtoint ptr [[VTABLE49]] to i64 +// CHECK-ADDRESSAUTH-NEXT: [[TMP16:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP15]], i32 2, i64 [[TMP14]]) +// CHECK-ADDRESSAUTH-NEXT: [[TMP17:%.*]] = inttoptr i64 [[TMP16]] to ptr +// CHECK-ADDRESSAUTH-NEXT: [[TMP18:%.*]] = load volatile i8, ptr [[TMP17]], align 8 +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[EARRAY]]) #[[ATTR7]] +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[EINSTANCE]]) #[[ATTR7]] +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[DINSTANCE]]) #[[ATTR7]] +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[CINSTANCE]]) #[[ATTR7]] +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[BINSTANCE]]) #[[ATTR7]] +// CHECK-ADDRESSAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[AINSTANCE]]) #[[ATTR7]] +// CHECK-ADDRESSAUTH-NEXT: ret void +// +// CHECK-BOTHAUTH-LABEL: define void @_ZN5test14testEv( +// CHECK-BOTHAUTH-SAME: ) #[[ATTR0]] { +// CHECK-BOTHAUTH-NEXT: [[ENTRY:.*]]: +// CHECK-BOTHAUTH-NEXT: [[AINSTANCE:%.*]] = alloca %"struct.test1::A", align 8 +// CHECK-BOTHAUTH-NEXT: [[BINSTANCE:%.*]] = alloca %"struct.test1::B", align 8 +// CHECK-BOTHAUTH-NEXT: [[CINSTANCE:%.*]] = alloca %"struct.test1::C", align 8 +// CHECK-BOTHAUTH-NEXT: [[DINSTANCE:%.*]] = alloca %"struct.test1::D", align 8 +// CHECK-BOTHAUTH-NEXT: [[EINSTANCE:%.*]] = alloca %"struct.test1::E", align 8 +// CHECK-BOTHAUTH-NEXT: [[EARRAY:%.*]] = alloca [1 x %"struct.test1::E"], align 8 +// CHECK-BOTHAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[AINSTANCE]]) #[[ATTR7]] +// CHECK-BOTHAUTH-NEXT: [[CALL:%.*]] = call ptr @_ZN5test11AC1Ev(ptr nonnull align 8 dereferenceable(8) [[AINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[BINSTANCE]]) #[[ATTR7]] +// CHECK-BOTHAUTH-NEXT: [[CALL1:%.*]] = call ptr @_ZN5test11BC1Ev(ptr nonnull align 8 dereferenceable(8) [[BINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[CINSTANCE]]) #[[ATTR7]] +// CHECK-BOTHAUTH-NEXT: [[CALL2:%.*]] = call ptr @_ZN5test11CC1Ev(ptr nonnull align 8 dereferenceable(16) [[CINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[DINSTANCE]]) #[[ATTR7]] +// CHECK-BOTHAUTH-NEXT: [[CALL3:%.*]] = call ptr @_ZN5test11DC1Ev(ptr nonnull align 8 dereferenceable(8) [[DINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[EINSTANCE]]) #[[ATTR7]] +// CHECK-BOTHAUTH-NEXT: [[CALL4:%.*]] = call ptr @_ZN5test11EC1Ev(ptr nonnull align 8 dereferenceable(16) [[EINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: call void @llvm.lifetime.start.p0(ptr [[EARRAY]]) #[[ATTR7]] +// CHECK-BOTHAUTH-NEXT: call void @llvm.memset.p0.i64(ptr align 8 [[EARRAY]], i8 0, i64 16, i1 false) +// CHECK-BOTHAUTH-NEXT: [[CALL5:%.*]] = call ptr @_ZN5test11EC1Ev(ptr nonnull align 8 dereferenceable(16) [[EARRAY]]) +// CHECK-BOTHAUTH-NEXT: [[CALL6:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[AINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: [[CALL7:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[BINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: [[TMP0:%.*]] = icmp eq ptr [[CINSTANCE]], null +// CHECK-BOTHAUTH-NEXT: br i1 [[TMP0]], label %[[CAST_END:.*]], label %[[CAST_NOTNULL:.*]] +// CHECK-BOTHAUTH: [[CAST_NOTNULL]]: +// CHECK-BOTHAUTH-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[CINSTANCE]], i64 8 +// CHECK-BOTHAUTH-NEXT: br label %[[CAST_END]] +// CHECK-BOTHAUTH: [[CAST_END]]: +// CHECK-BOTHAUTH-NEXT: [[CAST_RESULT:%.*]] = phi ptr [ [[ADD_PTR]], %[[CAST_NOTNULL]] ], [ null, %[[ENTRY]] ] +// CHECK-BOTHAUTH-NEXT: [[CALL8:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT]]) +// CHECK-BOTHAUTH-NEXT: [[TMP1:%.*]] = icmp eq ptr [[DINSTANCE]], null +// CHECK-BOTHAUTH-NEXT: br i1 [[TMP1]], label %[[CAST_END11:.*]], label %[[CAST_NOTNULL9:.*]] +// CHECK-BOTHAUTH: [[CAST_NOTNULL9]]: +// CHECK-BOTHAUTH-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[DINSTANCE]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[DINSTANCE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP3:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP2]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP4:%.*]] = ptrtoint ptr [[VTABLE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP5:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP4]], i32 2, i64 [[TMP3]]) +// CHECK-BOTHAUTH-NEXT: [[TMP6:%.*]] = inttoptr i64 [[TMP5]] to ptr +// CHECK-BOTHAUTH-NEXT: [[VBASE_OFFSET_PTR:%.*]] = getelementptr i8, ptr [[TMP6]], i64 -32 +// CHECK-BOTHAUTH-NEXT: [[VBASE_OFFSET:%.*]] = load i64, ptr [[VBASE_OFFSET_PTR]], align 8 +// CHECK-BOTHAUTH-NEXT: [[ADD_PTR10:%.*]] = getelementptr inbounds i8, ptr [[DINSTANCE]], i64 [[VBASE_OFFSET]] +// CHECK-BOTHAUTH-NEXT: br label %[[CAST_END11]] +// CHECK-BOTHAUTH: [[CAST_END11]]: +// CHECK-BOTHAUTH-NEXT: [[CAST_RESULT12:%.*]] = phi ptr [ [[ADD_PTR10]], %[[CAST_NOTNULL9]] ], [ null, %[[CAST_END]] ] +// CHECK-BOTHAUTH-NEXT: [[CALL13:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT12]]) +// CHECK-BOTHAUTH-NEXT: [[TMP7:%.*]] = icmp eq ptr [[EINSTANCE]], null +// CHECK-BOTHAUTH-NEXT: br i1 [[TMP7]], label %[[CAST_END19:.*]], label %[[CAST_NOTNULL14:.*]] +// CHECK-BOTHAUTH: [[CAST_NOTNULL14]]: +// CHECK-BOTHAUTH-NEXT: [[VTABLE15:%.*]] = load ptr, ptr [[EINSTANCE]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP8:%.*]] = ptrtoint ptr [[EINSTANCE]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP9:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP8]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP10:%.*]] = ptrtoint ptr [[VTABLE15]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP11:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP10]], i32 2, i64 [[TMP9]]) +// CHECK-BOTHAUTH-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP11]] to ptr +// CHECK-BOTHAUTH-NEXT: [[VBASE_OFFSET_PTR16:%.*]] = getelementptr i8, ptr [[TMP12]], i64 -32 +// CHECK-BOTHAUTH-NEXT: [[VBASE_OFFSET17:%.*]] = load i64, ptr [[VBASE_OFFSET_PTR16]], align 8 +// CHECK-BOTHAUTH-NEXT: [[ADD_PTR18:%.*]] = getelementptr inbounds i8, ptr [[EINSTANCE]], i64 [[VBASE_OFFSET17]] +// CHECK-BOTHAUTH-NEXT: br label %[[CAST_END19]] +// CHECK-BOTHAUTH: [[CAST_END19]]: +// CHECK-BOTHAUTH-NEXT: [[CAST_RESULT20:%.*]] = phi ptr [ [[ADD_PTR18]], %[[CAST_NOTNULL14]] ], [ null, %[[CAST_END11]] ] +// CHECK-BOTHAUTH-NEXT: [[CALL21:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT20]]) +// CHECK-BOTHAUTH-NEXT: [[TMP13:%.*]] = icmp eq ptr [[EINSTANCE]], null +// CHECK-BOTHAUTH-NEXT: br i1 [[TMP13]], label %[[CAST_END24:.*]], label %[[CAST_NOTNULL22:.*]] +// CHECK-BOTHAUTH: [[CAST_NOTNULL22]]: +// CHECK-BOTHAUTH-NEXT: [[ADD_PTR23:%.*]] = getelementptr inbounds i8, ptr [[EINSTANCE]], i64 8 +// CHECK-BOTHAUTH-NEXT: br label %[[CAST_END24]] +// CHECK-BOTHAUTH: [[CAST_END24]]: +// CHECK-BOTHAUTH-NEXT: [[CAST_RESULT25:%.*]] = phi ptr [ [[ADD_PTR23]], %[[CAST_NOTNULL22]] ], [ null, %[[CAST_END19]] ] +// CHECK-BOTHAUTH-NEXT: [[CALL26:%.*]] = call ptr @_ZN5test11aEPNS_1AE(ptr [[CAST_RESULT25]]) +// CHECK-BOTHAUTH-NEXT: [[CALL27:%.*]] = call ptr @_ZN5test11bEPNS_1BE(ptr [[BINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: [[TMP14:%.*]] = icmp eq ptr [[CINSTANCE]], null +// CHECK-BOTHAUTH-NEXT: br i1 [[TMP14]], label %[[CAST_END30:.*]], label %[[CAST_NOTNULL28:.*]] +// CHECK-BOTHAUTH: [[CAST_NOTNULL28]]: +// CHECK-BOTHAUTH-NEXT: [[ADD_PTR29:%.*]] = getelementptr inbounds i8, ptr [[CINSTANCE]], i64 8 +// CHECK-BOTHAUTH-NEXT: br label %[[CAST_END30]] +// CHECK-BOTHAUTH: [[CAST_END30]]: +// CHECK-BOTHAUTH-NEXT: [[CAST_RESULT31:%.*]] = phi ptr [ [[ADD_PTR29]], %[[CAST_NOTNULL28]] ], [ null, %[[CAST_END24]] ] +// CHECK-BOTHAUTH-NEXT: [[CALL32:%.*]] = call ptr @_ZN5test11bEPNS_1BE(ptr [[CAST_RESULT31]]) +// CHECK-BOTHAUTH-NEXT: [[TMP15:%.*]] = icmp eq ptr [[EINSTANCE]], null +// CHECK-BOTHAUTH-NEXT: br i1 [[TMP15]], label %[[CAST_END35:.*]], label %[[CAST_NOTNULL33:.*]] +// CHECK-BOTHAUTH: [[CAST_NOTNULL33]]: +// CHECK-BOTHAUTH-NEXT: [[ADD_PTR34:%.*]] = getelementptr inbounds i8, ptr [[EINSTANCE]], i64 8 +// CHECK-BOTHAUTH-NEXT: br label %[[CAST_END35]] +// CHECK-BOTHAUTH: [[CAST_END35]]: +// CHECK-BOTHAUTH-NEXT: [[CAST_RESULT36:%.*]] = phi ptr [ [[ADD_PTR34]], %[[CAST_NOTNULL33]] ], [ null, %[[CAST_END30]] ] +// CHECK-BOTHAUTH-NEXT: [[CALL37:%.*]] = call ptr @_ZN5test11bEPNS_1BE(ptr [[CAST_RESULT36]]) +// CHECK-BOTHAUTH-NEXT: [[CALL38:%.*]] = call ptr @_ZN5test16b_as_AEPNS_1BE(ptr [[BINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: [[CALL39:%.*]] = call ptr @_ZN5test11cEPNS_1CE(ptr [[CINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: [[CALL40:%.*]] = call ptr @_ZN5test16c_as_ZEPNS_1CE(ptr [[CINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: [[CALL41:%.*]] = call ptr @_ZN5test16c_as_BEPNS_1CE(ptr [[CINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: [[CALL42:%.*]] = call ptr @_ZN5test11dEPNS_1DE(ptr [[DINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: [[CALL43:%.*]] = call ptr @_ZN5test11dEPNS_1DE(ptr [[EINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: [[CALL44:%.*]] = call ptr @_ZN5test16d_as_AEPNS_1DE(ptr [[DINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: [[CALL45:%.*]] = call ptr @_ZN5test16d_as_AEPNS_1DE(ptr [[EINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: [[CALL46:%.*]] = call ptr @_ZN5test11eEPNS_1EE(ptr [[EINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: [[CALL47:%.*]] = call ptr @_ZN5test16e_as_BEPNS_1EE(ptr [[EINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: [[CALL48:%.*]] = call ptr @_ZN5test16e_as_DEPNS_1EE(ptr [[EINSTANCE]]) +// CHECK-BOTHAUTH-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [1 x %"struct.test1::E"], ptr [[EARRAY]], i64 0, i64 0 +// CHECK-BOTHAUTH-NEXT: [[VTABLE49:%.*]] = load ptr, ptr [[ARRAYDECAY]], align 8, !tbaa [[TBAA9]] +// CHECK-BOTHAUTH-NEXT: [[TMP16:%.*]] = ptrtoint ptr [[ARRAYDECAY]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP17:%.*]] = call i64 @llvm.ptrauth.blend(i64 [[TMP16]], i64 48388) +// CHECK-BOTHAUTH-NEXT: [[TMP18:%.*]] = ptrtoint ptr [[VTABLE49]] to i64 +// CHECK-BOTHAUTH-NEXT: [[TMP19:%.*]] = call i64 @llvm.ptrauth.auth(i64 [[TMP18]], i32 2, i64 [[TMP17]]) +// CHECK-BOTHAUTH-NEXT: [[TMP20:%.*]] = inttoptr i64 [[TMP19]] to ptr +// CHECK-BOTHAUTH-NEXT: [[TMP21:%.*]] = load volatile i8, ptr [[TMP20]], align 8 +// CHECK-BOTHAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[EARRAY]]) #[[ATTR7]] +// CHECK-BOTHAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[EINSTANCE]]) #[[ATTR7]] +// CHECK-BOTHAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[DINSTANCE]]) #[[ATTR7]] +// CHECK-BOTHAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[CINSTANCE]]) #[[ATTR7]] +// CHECK-BOTHAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[BINSTANCE]]) #[[ATTR7]] +// CHECK-BOTHAUTH-NEXT: call void @llvm.lifetime.end.p0(ptr [[AINSTANCE]]) #[[ATTR7]] +// CHECK-BOTHAUTH-NEXT: ret void +// void test() { A aInstance; B bInstance; @@ -368,3 +1350,68 @@ void test() { (void)__builtin_get_vtable_pointer(eArray); } } // namespace test1 +//. +// CHECK-NOAUTH: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK-NOAUTH: [[META5]] = !{!"Simple C++ TBAA"} +// CHECK-NOAUTH: [[TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK-NOAUTH: [[META7]] = !{!"p1 _ZTSN5test11AE", [[META8:![0-9]+]], i64 0} +// CHECK-NOAUTH: [[META8]] = !{!"any pointer", [[META4]], i64 0} +// CHECK-NOAUTH: [[TBAA9]] = !{[[META10:![0-9]+]], [[META10]], i64 0} +// CHECK-NOAUTH: [[META10]] = !{!"vtable pointer", [[META5]], i64 0} +// CHECK-NOAUTH: [[TBAA11]] = !{[[META12:![0-9]+]], [[META12]], i64 0} +// CHECK-NOAUTH: [[META12]] = !{!"p1 _ZTSN5test11BE", [[META8]], i64 0} +// CHECK-NOAUTH: [[TBAA13]] = !{[[META14:![0-9]+]], [[META14]], i64 0} +// CHECK-NOAUTH: [[META14]] = !{!"p1 _ZTSN5test11CE", [[META8]], i64 0} +// CHECK-NOAUTH: [[TBAA15]] = !{[[META16:![0-9]+]], [[META16]], i64 0} +// CHECK-NOAUTH: [[META16]] = !{!"p1 _ZTSN5test11DE", [[META8]], i64 0} +// CHECK-NOAUTH: [[TBAA17]] = !{[[META18:![0-9]+]], [[META18]], i64 0} +// CHECK-NOAUTH: [[META18]] = !{!"p1 _ZTSN5test11EE", [[META8]], i64 0} +//. +// CHECK-TYPEAUTH: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK-TYPEAUTH: [[META5]] = !{!"Simple C++ TBAA"} +// CHECK-TYPEAUTH: [[TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK-TYPEAUTH: [[META7]] = !{!"p1 _ZTSN5test11AE", [[META8:![0-9]+]], i64 0} +// CHECK-TYPEAUTH: [[META8]] = !{!"any pointer", [[META4]], i64 0} +// CHECK-TYPEAUTH: [[TBAA9]] = !{[[META10:![0-9]+]], [[META10]], i64 0} +// CHECK-TYPEAUTH: [[META10]] = !{!"vtable pointer", [[META5]], i64 0} +// CHECK-TYPEAUTH: [[TBAA11]] = !{[[META12:![0-9]+]], [[META12]], i64 0} +// CHECK-TYPEAUTH: [[META12]] = !{!"p1 _ZTSN5test11BE", [[META8]], i64 0} +// CHECK-TYPEAUTH: [[TBAA13]] = !{[[META14:![0-9]+]], [[META14]], i64 0} +// CHECK-TYPEAUTH: [[META14]] = !{!"p1 _ZTSN5test11CE", [[META8]], i64 0} +// CHECK-TYPEAUTH: [[TBAA15]] = !{[[META16:![0-9]+]], [[META16]], i64 0} +// CHECK-TYPEAUTH: [[META16]] = !{!"p1 _ZTSN5test11DE", [[META8]], i64 0} +// CHECK-TYPEAUTH: [[TBAA17]] = !{[[META18:![0-9]+]], [[META18]], i64 0} +// CHECK-TYPEAUTH: [[META18]] = !{!"p1 _ZTSN5test11EE", [[META8]], i64 0} +//. +// CHECK-ADDRESSAUTH: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK-ADDRESSAUTH: [[META5]] = !{!"Simple C++ TBAA"} +// CHECK-ADDRESSAUTH: [[TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK-ADDRESSAUTH: [[META7]] = !{!"p1 _ZTSN5test11AE", [[META8:![0-9]+]], i64 0} +// CHECK-ADDRESSAUTH: [[META8]] = !{!"any pointer", [[META4]], i64 0} +// CHECK-ADDRESSAUTH: [[TBAA9]] = !{[[META10:![0-9]+]], [[META10]], i64 0} +// CHECK-ADDRESSAUTH: [[META10]] = !{!"vtable pointer", [[META5]], i64 0} +// CHECK-ADDRESSAUTH: [[TBAA11]] = !{[[META12:![0-9]+]], [[META12]], i64 0} +// CHECK-ADDRESSAUTH: [[META12]] = !{!"p1 _ZTSN5test11BE", [[META8]], i64 0} +// CHECK-ADDRESSAUTH: [[TBAA13]] = !{[[META14:![0-9]+]], [[META14]], i64 0} +// CHECK-ADDRESSAUTH: [[META14]] = !{!"p1 _ZTSN5test11CE", [[META8]], i64 0} +// CHECK-ADDRESSAUTH: [[TBAA15]] = !{[[META16:![0-9]+]], [[META16]], i64 0} +// CHECK-ADDRESSAUTH: [[META16]] = !{!"p1 _ZTSN5test11DE", [[META8]], i64 0} +// CHECK-ADDRESSAUTH: [[TBAA17]] = !{[[META18:![0-9]+]], [[META18]], i64 0} +// CHECK-ADDRESSAUTH: [[META18]] = !{!"p1 _ZTSN5test11EE", [[META8]], i64 0} +//. +// CHECK-BOTHAUTH: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK-BOTHAUTH: [[META5]] = !{!"Simple C++ TBAA"} +// CHECK-BOTHAUTH: [[TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK-BOTHAUTH: [[META7]] = !{!"p1 _ZTSN5test11AE", [[META8:![0-9]+]], i64 0} +// CHECK-BOTHAUTH: [[META8]] = !{!"any pointer", [[META4]], i64 0} +// CHECK-BOTHAUTH: [[TBAA9]] = !{[[META10:![0-9]+]], [[META10]], i64 0} +// CHECK-BOTHAUTH: [[META10]] = !{!"vtable pointer", [[META5]], i64 0} +// CHECK-BOTHAUTH: [[TBAA11]] = !{[[META12:![0-9]+]], [[META12]], i64 0} +// CHECK-BOTHAUTH: [[META12]] = !{!"p1 _ZTSN5test11BE", [[META8]], i64 0} +// CHECK-BOTHAUTH: [[TBAA13]] = !{[[META14:![0-9]+]], [[META14]], i64 0} +// CHECK-BOTHAUTH: [[META14]] = !{!"p1 _ZTSN5test11CE", [[META8]], i64 0} +// CHECK-BOTHAUTH: [[TBAA15]] = !{[[META16:![0-9]+]], [[META16]], i64 0} +// CHECK-BOTHAUTH: [[META16]] = !{!"p1 _ZTSN5test11DE", [[META8]], i64 0} +// CHECK-BOTHAUTH: [[TBAA17]] = !{[[META18:![0-9]+]], [[META18]], i64 0} +// CHECK-BOTHAUTH: [[META18]] = !{!"p1 _ZTSN5test11EE", [[META8]], i64 0} +//. diff --git a/clang/test/CodeGenCXX/cfi-mfcall-nomerge.cpp b/clang/test/CodeGenCXX/cfi-mfcall-nomerge.cpp index d4b4f30..5c3bd17 100644 --- a/clang/test/CodeGenCXX/cfi-mfcall-nomerge.cpp +++ b/clang/test/CodeGenCXX/cfi-mfcall-nomerge.cpp @@ -29,43 +29,43 @@ void f(S *s, void (S::*p)()) { // NO-MERGE-NEXT: [[MEMPTR_ISVIRTUAL_NOT:%.*]] = icmp eq i64 [[TMP1]], 0 // NO-MERGE-NEXT: br i1 [[MEMPTR_ISVIRTUAL_NOT]], label %[[MEMPTR_NONVIRTUAL:.*]], label %[[MEMPTR_VIRTUAL:.*]] // NO-MERGE: [[MEMPTR_VIRTUAL]]: -// NO-MERGE-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[VTABLE_POINTER_TBAA2:![0-9]+]] +// NO-MERGE-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[VTABLE_POINTER_TBAA6:![0-9]+]] // NO-MERGE-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[VTABLE]], i64 [[P_COERCE0]] // NO-MERGE-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[TMP2]], i64 -1 -// NO-MERGE-NEXT: [[TMP4:%.*]] = tail call i1 @llvm.type.test(ptr [[TMP3]], metadata !"_ZTSM1SFvvE.virtual"), !nosanitize [[META5:![0-9]+]] -// NO-MERGE-NEXT: br i1 [[TMP4]], label %[[MEMPTR_VIRTUAL7:.*]], label %[[TRAP:.*]], !prof [[PROF6:![0-9]+]], !nosanitize [[META5]] +// NO-MERGE-NEXT: [[TMP4:%.*]] = tail call i1 @llvm.type.test(ptr [[TMP3]], metadata !"_ZTSM1SFvvE.virtual"), !nosanitize [[META8:![0-9]+]] +// NO-MERGE-NEXT: br i1 [[TMP4]], label %[[MEMPTR_VIRTUAL7:.*]], label %[[TRAP:.*]], !prof [[PROF9:![0-9]+]], !nosanitize [[META8]] // NO-MERGE: [[TRAP]]: -// NO-MERGE-NEXT: tail call void @llvm.ubsantrap(i8 2) #[[ATTR3:[0-9]+]], !nosanitize [[META5]] -// NO-MERGE-NEXT: unreachable, !nosanitize [[META5]] +// NO-MERGE-NEXT: tail call void @llvm.ubsantrap(i8 2) #[[ATTR3:[0-9]+]], !nosanitize [[META8]] +// NO-MERGE-NEXT: unreachable, !nosanitize [[META8]] // NO-MERGE: [[MEMPTR_NONVIRTUAL]]: // NO-MERGE-NEXT: [[MEMPTR_NONVIRTUALFN:%.*]] = inttoptr i64 [[P_COERCE0]] to ptr -// NO-MERGE-NEXT: [[TMP5:%.*]] = tail call i1 @llvm.type.test(ptr [[MEMPTR_NONVIRTUALFN]], metadata !"_ZTSM2B1FvvE") -// NO-MERGE-NEXT: [[TMP6:%.*]] = tail call i1 @llvm.type.test(ptr [[MEMPTR_NONVIRTUALFN]], metadata !"_ZTSM2B2FvvE") -// NO-MERGE-NEXT: [[TMP7:%.*]] = or i1 [[TMP5]], [[TMP6]], !nosanitize [[META5]] -// NO-MERGE-NEXT: br i1 [[TMP7]], label %[[MEMPTR_NONVIRTUAL23:.*]], label %[[TRAP2:.*]], !prof [[PROF6]], !nosanitize [[META5]] +// NO-MERGE-NEXT: [[TMP5:%.*]] = tail call i1 @llvm.type.test(ptr [[MEMPTR_NONVIRTUALFN]], metadata !"_ZTSM2B1FvvE"), !nosanitize [[META8]] +// NO-MERGE-NEXT: [[TMP6:%.*]] = tail call i1 @llvm.type.test(ptr [[MEMPTR_NONVIRTUALFN]], metadata !"_ZTSM2B2FvvE"), !nosanitize [[META8]] +// NO-MERGE-NEXT: [[TMP7:%.*]] = or i1 [[TMP5]], [[TMP6]], !nosanitize [[META8]] +// NO-MERGE-NEXT: br i1 [[TMP7]], label %[[MEMPTR_NONVIRTUAL23:.*]], label %[[TRAP2:.*]], !prof [[PROF9]], !nosanitize [[META8]] // NO-MERGE: [[TRAP2]]: -// NO-MERGE-NEXT: tail call void @llvm.ubsantrap(i8 2) #[[ATTR4:[0-9]+]], !nosanitize [[META5]] -// NO-MERGE-NEXT: unreachable, !nosanitize [[META5]] +// NO-MERGE-NEXT: tail call void @llvm.ubsantrap(i8 2) #[[ATTR4:[0-9]+]], !nosanitize [[META8]] +// NO-MERGE-NEXT: unreachable, !nosanitize [[META8]] // NO-MERGE: [[MEMPTR_VIRTUAL7]]: -// NO-MERGE-NEXT: [[MEMPTR_VIRTUALFN:%.*]] = load ptr, ptr [[TMP3]], align 8, !nosanitize [[META5]] +// NO-MERGE-NEXT: [[MEMPTR_VIRTUALFN:%.*]] = load ptr, ptr [[TMP3]], align 8, !nosanitize [[META8]] // NO-MERGE-NEXT: tail call void [[MEMPTR_VIRTUALFN]](ptr noundef nonnull align 1 dereferenceable(1) [[TMP0]]) #[[ATTR5:[0-9]+]] -// NO-MERGE-NEXT: [[VTABLE8:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[VTABLE_POINTER_TBAA2]] +// NO-MERGE-NEXT: [[VTABLE8:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[VTABLE_POINTER_TBAA6]] // NO-MERGE-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[VTABLE8]], i64 [[P_COERCE0]] // NO-MERGE-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP8]], i64 -1 -// NO-MERGE-NEXT: [[TMP10:%.*]] = tail call i1 @llvm.type.test(ptr [[TMP9]], metadata !"_ZTSM1SFvvE.virtual"), !nosanitize [[META5]] -// NO-MERGE-NEXT: br i1 [[TMP10]], label %[[MEMPTR_VIRTUAL19:.*]], label %[[TRAP2]], !prof [[PROF6]], !nosanitize [[META5]] +// NO-MERGE-NEXT: [[TMP10:%.*]] = tail call i1 @llvm.type.test(ptr [[TMP9]], metadata !"_ZTSM1SFvvE.virtual"), !nosanitize [[META8]] +// NO-MERGE-NEXT: br i1 [[TMP10]], label %[[MEMPTR_VIRTUAL19:.*]], label %[[TRAP2]], !prof [[PROF9]], !nosanitize [[META8]] // NO-MERGE: [[TRAP13:.*]]: -// NO-MERGE-NEXT: tail call void @llvm.ubsantrap(i8 2) #[[ATTR4]], !nosanitize [[META5]] -// NO-MERGE-NEXT: unreachable, !nosanitize [[META5]] +// NO-MERGE-NEXT: tail call void @llvm.ubsantrap(i8 2) #[[ATTR4]], !nosanitize [[META8]] +// NO-MERGE-NEXT: unreachable, !nosanitize [[META8]] // NO-MERGE: [[MEMPTR_VIRTUAL19]]: -// NO-MERGE-NEXT: [[MEMPTR_VIRTUALFN9:%.*]] = load ptr, ptr [[TMP9]], align 8, !nosanitize [[META5]] +// NO-MERGE-NEXT: [[MEMPTR_VIRTUALFN9:%.*]] = load ptr, ptr [[TMP9]], align 8, !nosanitize [[META8]] // NO-MERGE-NEXT: tail call void [[MEMPTR_VIRTUALFN9]](ptr noundef nonnull align 1 dereferenceable(1) [[TMP0]]) #[[ATTR5]] -// NO-MERGE-NEXT: [[VTABLE20:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[VTABLE_POINTER_TBAA2]] +// NO-MERGE-NEXT: [[VTABLE20:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[VTABLE_POINTER_TBAA6]] // NO-MERGE-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[VTABLE20]], i64 [[P_COERCE0]] // NO-MERGE-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[TMP11]], i64 -1 -// NO-MERGE-NEXT: [[TMP13:%.*]] = tail call i1 @llvm.type.test(ptr [[TMP12]], metadata !"_ZTSM1SFvvE.virtual"), !nosanitize [[META5]] -// NO-MERGE-NEXT: [[MEMPTR_VIRTUALFN21:%.*]] = load ptr, ptr [[TMP12]], align 8, !nosanitize [[META5]] -// NO-MERGE-NEXT: br i1 [[TMP13]], label %[[MEMPTR_END27:.*]], label %[[TRAP13]], !prof [[PROF6]], !nosanitize [[META5]] +// NO-MERGE-NEXT: [[TMP13:%.*]] = tail call i1 @llvm.type.test(ptr [[TMP12]], metadata !"_ZTSM1SFvvE.virtual"), !nosanitize [[META8]] +// NO-MERGE-NEXT: [[MEMPTR_VIRTUALFN21:%.*]] = load ptr, ptr [[TMP12]], align 8, !nosanitize [[META8]] +// NO-MERGE-NEXT: br i1 [[TMP13]], label %[[MEMPTR_END27:.*]], label %[[TRAP13]], !prof [[PROF9]], !nosanitize [[META8]] // NO-MERGE: [[MEMPTR_NONVIRTUAL23]]: // NO-MERGE-NEXT: tail call void [[MEMPTR_NONVIRTUALFN]](ptr noundef nonnull align 1 dereferenceable(1) [[TMP0]]) #[[ATTR5]] // NO-MERGE-NEXT: tail call void [[MEMPTR_NONVIRTUALFN]](ptr noundef nonnull align 1 dereferenceable(1) [[TMP0]]) #[[ATTR5]] @@ -84,37 +84,37 @@ void f(S *s, void (S::*p)()) { // MERGE-NEXT: [[MEMPTR_ISVIRTUAL_NOT:%.*]] = icmp eq i64 [[TMP1]], 0 // MERGE-NEXT: br i1 [[MEMPTR_ISVIRTUAL_NOT]], label %[[MEMPTR_NONVIRTUAL:.*]], label %[[MEMPTR_VIRTUAL:.*]] // MERGE: [[MEMPTR_VIRTUAL]]: -// MERGE-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[VTABLE_POINTER_TBAA2:![0-9]+]] +// MERGE-NEXT: [[VTABLE:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[VTABLE_POINTER_TBAA6:![0-9]+]] // MERGE-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[VTABLE]], i64 [[P_COERCE0]] // MERGE-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[TMP2]], i64 -1 -// MERGE-NEXT: [[TMP4:%.*]] = tail call i1 @llvm.type.test(ptr [[TMP3]], metadata !"_ZTSM1SFvvE.virtual"), !nosanitize [[META5:![0-9]+]] -// MERGE-NEXT: br i1 [[TMP4]], label %[[MEMPTR_VIRTUAL6:.*]], label %[[TRAP:.*]], !prof [[PROF6:![0-9]+]], !nosanitize [[META5]] +// MERGE-NEXT: [[TMP4:%.*]] = tail call i1 @llvm.type.test(ptr [[TMP3]], metadata !"_ZTSM1SFvvE.virtual"), !nosanitize [[META8:![0-9]+]] +// MERGE-NEXT: br i1 [[TMP4]], label %[[MEMPTR_VIRTUAL6:.*]], label %[[TRAP:.*]], !prof [[PROF9:![0-9]+]], !nosanitize [[META8]] // MERGE: [[TRAP]]: -// MERGE-NEXT: tail call void @llvm.ubsantrap(i8 2) #[[ATTR3:[0-9]+]], !nosanitize [[META5]] -// MERGE-NEXT: unreachable, !nosanitize [[META5]] +// MERGE-NEXT: tail call void @llvm.ubsantrap(i8 2) #[[ATTR3:[0-9]+]], !nosanitize [[META8]] +// MERGE-NEXT: unreachable, !nosanitize [[META8]] // MERGE: [[MEMPTR_NONVIRTUAL]]: // MERGE-NEXT: [[MEMPTR_NONVIRTUALFN:%.*]] = inttoptr i64 [[P_COERCE0]] to ptr -// MERGE-NEXT: [[TMP5:%.*]] = tail call i1 @llvm.type.test(ptr [[MEMPTR_NONVIRTUALFN]], metadata !"_ZTSM2B1FvvE") -// MERGE-NEXT: [[TMP6:%.*]] = tail call i1 @llvm.type.test(ptr [[MEMPTR_NONVIRTUALFN]], metadata !"_ZTSM2B2FvvE") -// MERGE-NEXT: [[TMP7:%.*]] = or i1 [[TMP5]], [[TMP6]], !nosanitize [[META5]] -// MERGE-NEXT: br i1 [[TMP7]], label %[[MEMPTR_NONVIRTUAL21:.*]], label %[[TRAP]], !prof [[PROF6]], !nosanitize [[META5]] +// MERGE-NEXT: [[TMP5:%.*]] = tail call i1 @llvm.type.test(ptr [[MEMPTR_NONVIRTUALFN]], metadata !"_ZTSM2B1FvvE"), !nosanitize [[META8]] +// MERGE-NEXT: [[TMP6:%.*]] = tail call i1 @llvm.type.test(ptr [[MEMPTR_NONVIRTUALFN]], metadata !"_ZTSM2B2FvvE"), !nosanitize [[META8]] +// MERGE-NEXT: [[TMP7:%.*]] = or i1 [[TMP5]], [[TMP6]], !nosanitize [[META8]] +// MERGE-NEXT: br i1 [[TMP7]], label %[[MEMPTR_NONVIRTUAL21:.*]], label %[[TRAP]], !prof [[PROF9]], !nosanitize [[META8]] // MERGE: [[MEMPTR_VIRTUAL6]]: -// MERGE-NEXT: [[MEMPTR_VIRTUALFN:%.*]] = load ptr, ptr [[TMP3]], align 8, !nosanitize [[META5]] +// MERGE-NEXT: [[MEMPTR_VIRTUALFN:%.*]] = load ptr, ptr [[TMP3]], align 8, !nosanitize [[META8]] // MERGE-NEXT: tail call void [[MEMPTR_VIRTUALFN]](ptr noundef nonnull align 1 dereferenceable(1) [[TMP0]]) #[[ATTR4:[0-9]+]] -// MERGE-NEXT: [[VTABLE7:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[VTABLE_POINTER_TBAA2]] +// MERGE-NEXT: [[VTABLE7:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[VTABLE_POINTER_TBAA6]] // MERGE-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[VTABLE7]], i64 [[P_COERCE0]] // MERGE-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP8]], i64 -1 -// MERGE-NEXT: [[TMP10:%.*]] = tail call i1 @llvm.type.test(ptr [[TMP9]], metadata !"_ZTSM1SFvvE.virtual"), !nosanitize [[META5]] -// MERGE-NEXT: br i1 [[TMP10]], label %[[MEMPTR_VIRTUAL17:.*]], label %[[TRAP]], !prof [[PROF6]], !nosanitize [[META5]] +// MERGE-NEXT: [[TMP10:%.*]] = tail call i1 @llvm.type.test(ptr [[TMP9]], metadata !"_ZTSM1SFvvE.virtual"), !nosanitize [[META8]] +// MERGE-NEXT: br i1 [[TMP10]], label %[[MEMPTR_VIRTUAL17:.*]], label %[[TRAP]], !prof [[PROF9]], !nosanitize [[META8]] // MERGE: [[MEMPTR_VIRTUAL17]]: -// MERGE-NEXT: [[MEMPTR_VIRTUALFN8:%.*]] = load ptr, ptr [[TMP9]], align 8, !nosanitize [[META5]] +// MERGE-NEXT: [[MEMPTR_VIRTUALFN8:%.*]] = load ptr, ptr [[TMP9]], align 8, !nosanitize [[META8]] // MERGE-NEXT: tail call void [[MEMPTR_VIRTUALFN8]](ptr noundef nonnull align 1 dereferenceable(1) [[TMP0]]) #[[ATTR4]] -// MERGE-NEXT: [[VTABLE18:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[VTABLE_POINTER_TBAA2]] +// MERGE-NEXT: [[VTABLE18:%.*]] = load ptr, ptr [[TMP0]], align 8, !tbaa [[VTABLE_POINTER_TBAA6]] // MERGE-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[VTABLE18]], i64 [[P_COERCE0]] // MERGE-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[TMP11]], i64 -1 -// MERGE-NEXT: [[TMP13:%.*]] = tail call i1 @llvm.type.test(ptr [[TMP12]], metadata !"_ZTSM1SFvvE.virtual"), !nosanitize [[META5]] -// MERGE-NEXT: [[MEMPTR_VIRTUALFN19:%.*]] = load ptr, ptr [[TMP12]], align 8, !nosanitize [[META5]] -// MERGE-NEXT: br i1 [[TMP13]], label %[[MEMPTR_END24:.*]], label %[[TRAP]], !prof [[PROF6]], !nosanitize [[META5]] +// MERGE-NEXT: [[TMP13:%.*]] = tail call i1 @llvm.type.test(ptr [[TMP12]], metadata !"_ZTSM1SFvvE.virtual"), !nosanitize [[META8]] +// MERGE-NEXT: [[MEMPTR_VIRTUALFN19:%.*]] = load ptr, ptr [[TMP12]], align 8, !nosanitize [[META8]] +// MERGE-NEXT: br i1 [[TMP13]], label %[[MEMPTR_END24:.*]], label %[[TRAP]], !prof [[PROF9]], !nosanitize [[META8]] // MERGE: [[MEMPTR_NONVIRTUAL21]]: // MERGE-NEXT: tail call void [[MEMPTR_NONVIRTUALFN]](ptr noundef nonnull align 1 dereferenceable(1) [[TMP0]]) #[[ATTR4]] // MERGE-NEXT: tail call void [[MEMPTR_NONVIRTUALFN]](ptr noundef nonnull align 1 dereferenceable(1) [[TMP0]]) #[[ATTR4]] @@ -125,15 +125,15 @@ void f(S *s, void (S::*p)()) { // MERGE-NEXT: ret void // //. -// NO-MERGE: [[VTABLE_POINTER_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// NO-MERGE: [[META3]] = !{!"vtable pointer", [[META4:![0-9]+]], i64 0} -// NO-MERGE: [[META4]] = !{!"Simple C++ TBAA"} -// NO-MERGE: [[META5]] = !{} -// NO-MERGE: [[PROF6]] = !{!"branch_weights", i32 1048575, i32 1} +// NO-MERGE: [[META5:![0-9]+]] = !{!"Simple C++ TBAA"} +// NO-MERGE: [[VTABLE_POINTER_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// NO-MERGE: [[META7]] = !{!"vtable pointer", [[META5]], i64 0} +// NO-MERGE: [[META8]] = !{} +// NO-MERGE: [[PROF9]] = !{!"branch_weights", i32 1048575, i32 1} //. -// MERGE: [[VTABLE_POINTER_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// MERGE: [[META3]] = !{!"vtable pointer", [[META4:![0-9]+]], i64 0} -// MERGE: [[META4]] = !{!"Simple C++ TBAA"} -// MERGE: [[META5]] = !{} -// MERGE: [[PROF6]] = !{!"branch_weights", i32 1048575, i32 1} +// MERGE: [[META5:![0-9]+]] = !{!"Simple C++ TBAA"} +// MERGE: [[VTABLE_POINTER_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// MERGE: [[META7]] = !{!"vtable pointer", [[META5]], i64 0} +// MERGE: [[META8]] = !{} +// MERGE: [[PROF9]] = !{!"branch_weights", i32 1048575, i32 1} //. diff --git a/clang/test/CodeGenCXX/inline-then-fold-variadics.cpp b/clang/test/CodeGenCXX/inline-then-fold-variadics.cpp index ab3695a..3094165 100644 --- a/clang/test/CodeGenCXX/inline-then-fold-variadics.cpp +++ b/clang/test/CodeGenCXX/inline-then-fold-variadics.cpp @@ -110,8 +110,8 @@ int first_i32_ulong2(int x, ulong2 *y) { return first<int, ulong2>(x, *y); } // CHECK-LABEL: define void @second_i32_ulong2( // CHECK-SAME: i32 noundef [[X:%.*]], ptr noundef readonly captures(none) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 16)) [[R:%.*]]) local_unnamed_addr #[[ATTR1:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr [[Y]], align 16, !tbaa [[INT_TBAA2:![0-9]+]] -// CHECK-NEXT: store <2 x i64> [[TMP0]], ptr [[R]], align 16, !tbaa [[INT_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr [[Y]], align 16, !tbaa [[DOUBLE_TBAA6:![0-9]+]] +// CHECK-NEXT: store <2 x i64> [[TMP0]], ptr [[R]], align 16, !tbaa [[DOUBLE_TBAA6]] // CHECK-NEXT: ret void // void second_i32_ulong2(int x, ulong2 *y, ulong2 *r) { @@ -121,8 +121,8 @@ void second_i32_ulong2(int x, ulong2 *y, ulong2 *r) { // CHECK-LABEL: define void @first_ulong2_i32( // CHECK-SAME: ptr noundef readonly captures(none) [[X:%.*]], i32 noundef [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 16)) [[R:%.*]]) local_unnamed_addr #[[ATTR1]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr [[X]], align 16, !tbaa [[INT_TBAA2]] -// CHECK-NEXT: store <2 x i64> [[TMP0]], ptr [[R]], align 16, !tbaa [[INT_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr [[X]], align 16, !tbaa [[DOUBLE_TBAA6]] +// CHECK-NEXT: store <2 x i64> [[TMP0]], ptr [[R]], align 16, !tbaa [[DOUBLE_TBAA6]] // CHECK-NEXT: ret void // void first_ulong2_i32(ulong2 *x, int y, ulong2 *r) { @@ -180,7 +180,7 @@ void first_asc_i32(asc *x, int y, asc *r) { *r = first<asc, int>(*x, y); } int second_asc_i32(asc *x, int y) { return second<asc, int>(*x, y); } } //. -// CHECK: [[INT_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"Simple C++ TBAA"} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C++ TBAA"} +// CHECK: [[DOUBLE_TBAA6]] = !{[[META4]], [[META4]], i64 0} //. diff --git a/clang/test/CodeGenCXX/load-reference-metadata.cpp b/clang/test/CodeGenCXX/load-reference-metadata.cpp index abfdd05..b9777f4 100644 --- a/clang/test/CodeGenCXX/load-reference-metadata.cpp +++ b/clang/test/CodeGenCXX/load-reference-metadata.cpp @@ -13,18 +13,18 @@ struct S { // CHECK-SAME: ptr noundef nonnull align 8 dereferenceable(24) [[S:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[S_ADDR:%.*]] = alloca ptr, align 8 -// CHECK-NEXT: store ptr [[S]], ptr [[S_ADDR]], align 8, !tbaa [[_ZTS1SPTR_TBAA2:![0-9]+]] -// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[S_ADDR]], align 8, !tbaa [[_ZTS1SPTR_TBAA2]], !nonnull [[META7:![0-9]+]], !align [[META8:![0-9]+]] +// CHECK-NEXT: store ptr [[S]], ptr [[S_ADDR]], align 8, !tbaa [[_ZTS1SPTR_TBAA6:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[S_ADDR]], align 8, !tbaa [[_ZTS1SPTR_TBAA6]], !nonnull [[META9:![0-9]+]], !align [[META10:![0-9]+]] // CHECK-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[TMP0]], i32 0, i32 0 -// CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A]], align 8, !tbaa [[CHARPTR_TBAA9:![0-9]+]], !nonnull [[META7]] -// CHECK-NEXT: store i8 0, ptr [[TMP1]], align 1, !tbaa [[CHAR_TBAA14:![0-9]+]] -// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[S_ADDR]], align 8, !tbaa [[_ZTS1SPTR_TBAA2]], !nonnull [[META7]], !align [[META8]] +// CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A]], align 8, !tbaa [[CHARPTR_TBAA11:![0-9]+]], !nonnull [[META9]] +// CHECK-NEXT: store i8 0, ptr [[TMP1]], align 1, !tbaa [[CHAR_TBAA16:![0-9]+]] +// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[S_ADDR]], align 8, !tbaa [[_ZTS1SPTR_TBAA6]], !nonnull [[META9]], !align [[META10]] // CHECK-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_S]], ptr [[TMP2]], i32 0, i32 1 -// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[B]], align 8, !tbaa [[INTPTR_TBAA15:![0-9]+]], !nonnull [[META7]], !align [[META16:![0-9]+]] -// CHECK-NEXT: store i32 0, ptr [[TMP3]], align 4, !tbaa [[INT_TBAA17:![0-9]+]] -// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[S_ADDR]], align 8, !tbaa [[_ZTS1SPTR_TBAA2]], !nonnull [[META7]], !align [[META8]] +// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[B]], align 8, !tbaa [[INTPTR_TBAA17:![0-9]+]], !nonnull [[META9]], !align [[META18:![0-9]+]] +// CHECK-NEXT: store i32 0, ptr [[TMP3]], align 4, !tbaa [[INT_TBAA2:![0-9]+]] +// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[S_ADDR]], align 8, !tbaa [[_ZTS1SPTR_TBAA6]], !nonnull [[META9]], !align [[META10]] // CHECK-NEXT: [[C:%.*]] = getelementptr inbounds nuw [[STRUCT_S]], ptr [[TMP4]], i32 0, i32 2 -// CHECK-NEXT: [[TMP5:%.*]] = load ptr, ptr [[C]], align 8, !tbaa [[_ZTS1FPTR_TBAA19:![0-9]+]], !nonnull [[META7]], !align [[META20:![0-9]+]] +// CHECK-NEXT: [[TMP5:%.*]] = load ptr, ptr [[C]], align 8, !tbaa [[_ZTS1FPTR_TBAA19:![0-9]+]], !nonnull [[META9]], !align [[META20:![0-9]+]] // CHECK-NEXT: [[X:%.*]] = getelementptr inbounds nuw [[STRUCT_F:%.*]], ptr [[TMP5]], i32 0, i32 0 // CHECK-NEXT: store i32 0, ptr [[X]], align 32, !tbaa [[INT_TBAA21:![0-9]+]] // CHECK-NEXT: ret void @@ -42,10 +42,10 @@ extern B (&bb)[2]; // CHECK-LABEL: define dso_local void @_Z13test_externalv( // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr @b, align 8, !tbaa [[_ZTS1BPTR_TBAA23:![0-9]+]], !nonnull [[META7]], !align [[META8]] +// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr @b, align 8, !tbaa [[_ZTS1BPTR_TBAA23:![0-9]+]], !nonnull [[META9]], !align [[META10]] // CHECK-NEXT: [[C:%.*]] = getelementptr inbounds nuw [[STRUCT_B:%.*]], ptr [[TMP0]], i32 0, i32 2 // CHECK-NEXT: store i8 0, ptr [[C]], align 8, !tbaa [[CHAR_TBAA25:![0-9]+]] -// CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr @bb, align 8, !tbaa [[_ZTS1BPTR_TBAA23]], !nonnull [[META7]], !align [[META20]] +// CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr @bb, align 8, !tbaa [[_ZTS1BPTR_TBAA23]], !nonnull [[META9]], !align [[META20]] // CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x %struct.B], ptr [[TMP1]], i64 0, i64 0 // CHECK-NEXT: [[C1:%.*]] = getelementptr inbounds nuw [[STRUCT_B]], ptr [[ARRAYIDX]], i32 0, i32 2 // CHECK-NEXT: store i8 0, ptr [[C1]], align 16, !tbaa [[CHAR_TBAA25]] @@ -61,7 +61,7 @@ void test_external() { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[S_ADDR:%.*]] = alloca ptr, align 8 // CHECK-NEXT: store ptr [[S]], ptr [[S_ADDR]], align 8, !tbaa [[_ZTS1BPTR_TBAA23]] -// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[S_ADDR]], align 8, !tbaa [[_ZTS1BPTR_TBAA23]], !nonnull [[META7]], !align [[META8]] +// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[S_ADDR]], align 8, !tbaa [[_ZTS1BPTR_TBAA23]], !nonnull [[META9]], !align [[META10]] // CHECK-NEXT: [[C:%.*]] = getelementptr inbounds nuw [[STRUCT_B:%.*]], ptr [[TMP0]], i32 0, i32 2 // CHECK-NEXT: ret ptr [[C]] // @@ -69,30 +69,30 @@ char* test_deref_only(B &s) { return &s.c; } //. -// CHECK: [[_ZTS1SPTR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"p1 _ZTS1S", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"any pointer", [[META5:![0-9]+]], i64 0} -// CHECK: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// CHECK: [[META6]] = !{!"Simple C++ TBAA"} -// CHECK: [[META7]] = !{} -// CHECK: [[META8]] = !{i64 8} -// CHECK: [[CHARPTR_TBAA9]] = !{[[META10:![0-9]+]], [[META11:![0-9]+]], i64 0} -// CHECK: [[META10]] = !{!"_ZTS1S", [[META11]], i64 0, [[META12:![0-9]+]], i64 8, [[META13:![0-9]+]], i64 16} -// CHECK: [[META11]] = !{!"p1 omnipotent char", [[META4]], i64 0} -// CHECK: [[META12]] = !{!"p1 int", [[META4]], i64 0} -// CHECK: [[META13]] = !{!"p1 _ZTS1F", [[META4]], i64 0} -// CHECK: [[CHAR_TBAA14]] = !{[[META5]], [[META5]], i64 0} -// CHECK: [[INTPTR_TBAA15]] = !{[[META10]], [[META12]], i64 8} -// CHECK: [[META16]] = !{i64 4} -// CHECK: [[INT_TBAA17]] = !{[[META18:![0-9]+]], [[META18]], i64 0} -// CHECK: [[META18]] = !{!"int", [[META5]], i64 0} -// CHECK: [[_ZTS1FPTR_TBAA19]] = !{[[META10]], [[META13]], i64 16} +// CHECK: [[INT_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} +// CHECK: [[META3]] = !{!"int", [[META4:![0-9]+]], i64 0} +// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C++ TBAA"} +// CHECK: [[_ZTS1SPTR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK: [[META7]] = !{!"p1 _ZTS1S", [[META8:![0-9]+]], i64 0} +// CHECK: [[META8]] = !{!"any pointer", [[META4]], i64 0} +// CHECK: [[META9]] = !{} +// CHECK: [[META10]] = !{i64 8} +// CHECK: [[CHARPTR_TBAA11]] = !{[[META12:![0-9]+]], [[META13:![0-9]+]], i64 0} +// CHECK: [[META12]] = !{!"_ZTS1S", [[META13]], i64 0, [[META14:![0-9]+]], i64 8, [[META15:![0-9]+]], i64 16} +// CHECK: [[META13]] = !{!"p1 omnipotent char", [[META8]], i64 0} +// CHECK: [[META14]] = !{!"p1 int", [[META8]], i64 0} +// CHECK: [[META15]] = !{!"p1 _ZTS1F", [[META8]], i64 0} +// CHECK: [[CHAR_TBAA16]] = !{[[META4]], [[META4]], i64 0} +// CHECK: [[INTPTR_TBAA17]] = !{[[META12]], [[META14]], i64 8} +// CHECK: [[META18]] = !{i64 4} +// CHECK: [[_ZTS1FPTR_TBAA19]] = !{[[META12]], [[META15]], i64 16} // CHECK: [[META20]] = !{i64 32} -// CHECK: [[INT_TBAA21]] = !{[[META22:![0-9]+]], [[META18]], i64 0} -// CHECK: [[META22]] = !{!"_ZTS1F", [[META18]], i64 0} +// CHECK: [[INT_TBAA21]] = !{[[META22:![0-9]+]], [[META3]], i64 0} +// CHECK: [[META22]] = !{!"_ZTS1F", [[META3]], i64 0} // CHECK: [[_ZTS1BPTR_TBAA23]] = !{[[META24:![0-9]+]], [[META24]], i64 0} -// CHECK: [[META24]] = !{!"p1 _ZTS1B", [[META4]], i64 0} -// CHECK: [[CHAR_TBAA25]] = !{[[META26:![0-9]+]], [[META5]], i64 16} -// CHECK: [[META26]] = !{!"_ZTS1B", [[META27:![0-9]+]], i64 8, [[META5]], i64 16} -// CHECK: [[META27]] = !{!"long long", [[META5]], i64 0} +// CHECK: [[META24]] = !{!"p1 _ZTS1B", [[META8]], i64 0} +// CHECK: [[CHAR_TBAA25]] = !{[[META26:![0-9]+]], [[META4]], i64 16} +// CHECK: [[META26]] = !{!"_ZTS1B", [[META27:![0-9]+]], i64 8, [[META4]], i64 16} +// CHECK: [[META27]] = !{!"long long", [[META4]], i64 0} //. diff --git a/clang/test/CodeGenCXX/std-byte.cpp b/clang/test/CodeGenCXX/std-byte.cpp index 63f41e6..00191b2 100644 --- a/clang/test/CodeGenCXX/std-byte.cpp +++ b/clang/test/CodeGenCXX/std-byte.cpp @@ -1,3 +1,4 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5 // RUN: %clang_cc1 -std=c++1z -Werror -triple i386-unknown-unknown -emit-llvm -O1 -disable-llvm-passes -o - %s | FileCheck %s // std::byte should be considered equivalent to char for aliasing. @@ -8,10 +9,10 @@ enum byte : unsigned char {}; // CHECK-LABEL: define{{.*}} void @test0( extern "C" void test0(std::byte *sb, int *i) { - // CHECK: store i8 0, ptr %{{.*}} !tbaa [[TAG_CHAR:!.*]] + // CHECK: store i8 0, ptr %{{.*}}, align 1, !tbaa [[TBAA11:![0-9]+]] *sb = std::byte{0}; - // CHECK: store i32 1, ptr %{{.*}} !tbaa [[TAG_INT:!.*]] + // CHECK: store i32 1, ptr %{{.*}}, align 4, !tbaa [[TBAA3:![0-9]+]] *i = 1; } @@ -27,15 +28,24 @@ enum byte : unsigned char {}; // CHECK-LABEL: define{{.*}} void @test1( extern "C" void test1(::byte *b, ::my::byte *mb, ::my::std::byte *msb) { + // CHECK: store i8 0, ptr %{{.*}}, align 1, !tbaa [[TBAA12:![0-9]+]] *b = ::byte{0}; + // CHECK: store i8 0, ptr %{{.*}}, align 1, !tbaa [[TBAA14:![0-9]+]] *mb = ::my::byte{0}; + // CHECK: store i8 0, ptr %{{.*}}, align 1, !tbaa [[TBAA16:![0-9]+]] *msb = ::my::std::byte{0}; - // CHECK-NOT: store i8 0, ptr %{{.*}} !tbaa [[TAG_CHAR]] } -// CHECK: !"any pointer", [[TYPE_CHAR:!.*]], -// CHECK: [[TYPE_CHAR]] = !{!"omnipotent char", [[TAG_CXX_TBAA:!.*]], -// CHECK: [[TAG_CXX_TBAA]] = !{!"Simple C++ TBAA"} -// CHECK: [[TAG_CHAR]] = !{[[TYPE_CHAR:!.*]], [[TYPE_CHAR]], i64 0} -// CHECK: [[TAG_INT]] = !{[[TYPE_INT:!.*]], [[TYPE_INT]], i64 0} -// CHECK: [[TYPE_INT]] = !{!"int", [[TYPE_CHAR]] +//. +// CHECK: [[TBAA3]] = !{[[META4:![0-9]+]], [[META4]], i64 0} +// CHECK: [[META4]] = !{!"int", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} +// CHECK: [[META6]] = !{!"Simple C++ TBAA"} +// CHECK: [[TBAA11]] = !{[[META5]], [[META5]], i64 0} +// CHECK: [[TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} +// CHECK: [[META13]] = !{!"_ZTS4byte", [[META5]], i64 0} +// CHECK: [[TBAA14]] = !{[[META15:![0-9]+]], [[META15]], i64 0} +// CHECK: [[META15]] = !{!"_ZTSN2my4byteE", [[META5]], i64 0} +// CHECK: [[TBAA16]] = !{[[META17:![0-9]+]], [[META17]], i64 0} +// CHECK: [[META17]] = !{!"_ZTSN2my3std4byteE", [[META5]], i64 0} +//. diff --git a/clang/test/CodeGenOpenCL/amdgcn-buffer-rsrc-type.cl b/clang/test/CodeGenOpenCL/amdgcn-buffer-rsrc-type.cl index b55f663..05b7a9b 100644 --- a/clang/test/CodeGenOpenCL/amdgcn-buffer-rsrc-type.cl +++ b/clang/test/CodeGenOpenCL/amdgcn-buffer-rsrc-type.cl @@ -27,7 +27,7 @@ __amdgpu_buffer_rsrc_t getBuffer(void *p) { // CHECK-NEXT: [[TOBOOL_NOT:%.*]] = icmp eq ptr addrspace(5) [[P]], addrspacecast (ptr null to ptr addrspace(5)) // CHECK-NEXT: br i1 [[TOBOOL_NOT]], label %[[IF_END:.*]], label %[[IF_THEN:.*]] // CHECK: [[IF_THEN]]: -// CHECK-NEXT: [[TMP0:%.*]] = load ptr addrspace(8), ptr addrspace(5) [[P]], align 16, !tbaa [[__AMDGPU_BUFFER_RSRC_T_TBAA4:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load ptr addrspace(8), ptr addrspace(5) [[P]], align 16, !tbaa [[__AMDGPU_BUFFER_RSRC_T_TBAA8:![0-9]+]] // CHECK-NEXT: tail call void @consumeBuffer(ptr addrspace(8) [[TMP0]]) #[[ATTR2]] // CHECK-NEXT: br label %[[IF_END]] // CHECK: [[IF_END]]: @@ -41,14 +41,14 @@ void consumeBufferPtr(__amdgpu_buffer_rsrc_t *p) { // CHECK-LABEL: define dso_local void @test( // CHECK-SAME: ptr addrspace(5) noundef readonly captures(address) [[A:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[A]], align 16, !tbaa [[INT_TBAA8:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[A]], align 16, !tbaa [[INT_TBAA10:![0-9]+]] // CHECK-NEXT: [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0 // CHECK-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq ptr addrspace(5) [[A]], addrspacecast (ptr null to ptr addrspace(5)) // CHECK-NEXT: [[OR_COND:%.*]] = select i1 [[TOBOOL_NOT]], i1 true, i1 [[TOBOOL_NOT_I]] // CHECK-NEXT: br i1 [[OR_COND]], label %[[IF_END:.*]], label %[[IF_THEN_I:.*]] // CHECK: [[IF_THEN_I]]: // CHECK-NEXT: [[R:%.*]] = getelementptr inbounds nuw i8, ptr addrspace(5) [[A]], i32 16 -// CHECK-NEXT: [[TMP1:%.*]] = load ptr addrspace(8), ptr addrspace(5) [[R]], align 16, !tbaa [[__AMDGPU_BUFFER_RSRC_T_TBAA4]] +// CHECK-NEXT: [[TMP1:%.*]] = load ptr addrspace(8), ptr addrspace(5) [[R]], align 16, !tbaa [[__AMDGPU_BUFFER_RSRC_T_TBAA8]] // CHECK-NEXT: tail call void @consumeBuffer(ptr addrspace(8) [[TMP1]]) #[[ATTR2]] // CHECK-NEXT: br label %[[IF_END]] // CHECK: [[IF_END]]: @@ -81,11 +81,11 @@ AA bar(void *p) { return a; } //. -// CHECK: [[__AMDGPU_BUFFER_RSRC_T_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} -// CHECK: [[META5]] = !{!"__amdgpu_buffer_rsrc_t", [[META6:![0-9]+]], i64 0} +// CHECK: [[META5:![0-9]+]] = !{!"int", [[META6:![0-9]+]], i64 0} // CHECK: [[META6]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} // CHECK: [[META7]] = !{!"Simple C/C++ TBAA"} -// CHECK: [[INT_TBAA8]] = !{[[META9:![0-9]+]], [[META10:![0-9]+]], i64 0} -// CHECK: [[META9]] = !{!"AA_ty", [[META10]], i64 0, [[META5]], i64 16} -// CHECK: [[META10]] = !{!"int", [[META6]], i64 0} +// CHECK: [[__AMDGPU_BUFFER_RSRC_T_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} +// CHECK: [[META9]] = !{!"__amdgpu_buffer_rsrc_t", [[META6]], i64 0} +// CHECK: [[INT_TBAA10]] = !{[[META11:![0-9]+]], [[META5]], i64 0} +// CHECK: [[META11]] = !{!"AA_ty", [[META5]], i64 0, [[META9]], i64 16} //. diff --git a/clang/test/CodeGenOpenCL/amdgpu-cluster-dims.cl b/clang/test/CodeGenOpenCL/amdgpu-cluster-dims.cl index be822a6..8c3e5b7 100644 --- a/clang/test/CodeGenOpenCL/amdgpu-cluster-dims.cl +++ b/clang/test/CodeGenOpenCL/amdgpu-cluster-dims.cl @@ -1,28 +1,28 @@ -// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --check-attributes --check-globals all --include-generated-funcs --prefix-filecheck-ir-name VAR --version 5 +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --check-attributes --check-globals all --include-generated-funcs --prefix-filecheck-ir-name VAR --version 6 // RUN: %clang_cc1 -cl-std=CL2.0 -triple amdgcn-amd-amdhsa -target-cpu gfx1250 -disable-llvm-passes -fno-ident -emit-llvm %s -o - | FileCheck %s kernel void foo(global int *p) { *p = 1; } // CHECK: Function Attrs: convergent norecurse nounwind // CHECK-LABEL: define dso_local amdgpu_kernel void @foo( -// CHECK-SAME: ptr addrspace(1) noundef align 4 [[P:%.*]]) #[[ATTR0:[0-9]+]] !kernel_arg_addr_space [[META3:![0-9]+]] !kernel_arg_access_qual [[META4:![0-9]+]] !kernel_arg_type [[META5:![0-9]+]] !kernel_arg_base_type [[META5]] !kernel_arg_type_qual [[META6:![0-9]+]] { +// CHECK-SAME: ptr addrspace(1) noundef align 4 [[P:%.*]]) #[[ATTR0:[0-9]+]] !kernel_arg_addr_space [[META7:![0-9]+]] !kernel_arg_access_qual [[META8:![0-9]+]] !kernel_arg_type [[META9:![0-9]+]] !kernel_arg_base_type [[META9]] !kernel_arg_type_qual [[META10:![0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[P_ADDR:%.*]] = alloca ptr addrspace(1), align 8, addrspace(5) // CHECK-NEXT: [[P_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[P_ADDR]] to ptr -// CHECK-NEXT: store ptr addrspace(1) [[P]], ptr [[P_ADDR_ASCAST]], align 8, !tbaa [[TBAA7:![0-9]+]] -// CHECK-NEXT: [[TMP0:%.*]] = load ptr addrspace(1), ptr [[P_ADDR_ASCAST]], align 8, !tbaa [[TBAA7]] +// CHECK-NEXT: store ptr addrspace(1) [[P]], ptr [[P_ADDR_ASCAST]], align 8, !tbaa [[INTPTR_TBAA11:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load ptr addrspace(1), ptr [[P_ADDR_ASCAST]], align 8, !tbaa [[INTPTR_TBAA11]] // CHECK-NEXT: call void @__clang_ocl_kern_imp_foo(ptr addrspace(1) noundef align 4 [[TMP0]]) #[[ATTR2:[0-9]+]] // CHECK-NEXT: ret void // // // CHECK: Function Attrs: alwaysinline convergent norecurse nounwind // CHECK-LABEL: define dso_local void @__clang_ocl_kern_imp_foo( -// CHECK-SAME: ptr addrspace(1) noundef align 4 [[P:%.*]]) #[[ATTR1:[0-9]+]] !kernel_arg_addr_space [[META3]] !kernel_arg_access_qual [[META4]] !kernel_arg_type [[META5]] !kernel_arg_base_type [[META5]] !kernel_arg_type_qual [[META6]] { +// CHECK-SAME: ptr addrspace(1) noundef align 4 [[P:%.*]]) #[[ATTR1:[0-9]+]] !kernel_arg_addr_space [[META7]] !kernel_arg_access_qual [[META8]] !kernel_arg_type [[META9]] !kernel_arg_base_type [[META9]] !kernel_arg_type_qual [[META10]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[P_ADDR:%.*]] = alloca ptr addrspace(1), align 8, addrspace(5) // CHECK-NEXT: [[P_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[P_ADDR]] to ptr -// CHECK-NEXT: store ptr addrspace(1) [[P]], ptr [[P_ADDR_ASCAST]], align 8, !tbaa [[TBAA7]] -// CHECK-NEXT: [[TMP0:%.*]] = load ptr addrspace(1), ptr [[P_ADDR_ASCAST]], align 8, !tbaa [[TBAA7]] -// CHECK-NEXT: store i32 1, ptr addrspace(1) [[TMP0]], align 4, !tbaa [[TBAA12:![0-9]+]] +// CHECK-NEXT: store ptr addrspace(1) [[P]], ptr [[P_ADDR_ASCAST]], align 8, !tbaa [[INTPTR_TBAA11]] +// CHECK-NEXT: [[TMP0:%.*]] = load ptr addrspace(1), ptr [[P_ADDR_ASCAST]], align 8, !tbaa [[INTPTR_TBAA11]] +// CHECK-NEXT: store i32 1, ptr addrspace(1) [[TMP0]], align 4, !tbaa [[INT_TBAA3:![0-9]+]] // CHECK-NEXT: ret void // //. @@ -33,15 +33,15 @@ kernel void foo(global int *p) { *p = 1; } // CHECK: [[META0:![0-9]+]] = !{i32 1, !"amdhsa_code_object_version", i32 600} // CHECK: [[META1:![0-9]+]] = !{i32 1, !"wchar_size", i32 4} // CHECK: [[META2:![0-9]+]] = !{i32 2, i32 0} -// CHECK: [[META3]] = !{i32 1} -// CHECK: [[META4]] = !{!"none"} -// CHECK: [[META5]] = !{!"int*"} -// CHECK: [[META6]] = !{!""} -// CHECK: [[TBAA7]] = !{[[META8:![0-9]+]], [[META8]], i64 0} -// CHECK: [[META8]] = !{!"p1 int", [[META9:![0-9]+]], i64 0} -// CHECK: [[META9]] = !{!"any pointer", [[META10:![0-9]+]], i64 0} -// CHECK: [[META10]] = !{!"omnipotent char", [[META11:![0-9]+]], i64 0} -// CHECK: [[META11]] = !{!"Simple C/C++ TBAA"} -// CHECK: [[TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} -// CHECK: [[META13]] = !{!"int", [[META10]], i64 0} +// CHECK: [[INT_TBAA3]] = !{[[META4:![0-9]+]], [[META4]], i64 0} +// CHECK: [[META4]] = !{!"int", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} +// CHECK: [[META6]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META7]] = !{i32 1} +// CHECK: [[META8]] = !{!"none"} +// CHECK: [[META9]] = !{!"int*"} +// CHECK: [[META10]] = !{!""} +// CHECK: [[INTPTR_TBAA11]] = !{[[META12:![0-9]+]], [[META12]], i64 0} +// CHECK: [[META12]] = !{!"p1 int", [[META13:![0-9]+]], i64 0} +// CHECK: [[META13]] = !{!"any pointer", [[META5]], i64 0} //. diff --git a/clang/test/CodeGenOpenCL/amdgpu-enqueue-kernel.cl b/clang/test/CodeGenOpenCL/amdgpu-enqueue-kernel.cl index 6d57323..e9adac2 100644 --- a/clang/test/CodeGenOpenCL/amdgpu-enqueue-kernel.cl +++ b/clang/test/CodeGenOpenCL/amdgpu-enqueue-kernel.cl @@ -451,19 +451,19 @@ kernel void test_target_features_kernel(global int *i) { // GFX900-NEXT: [[OUT_ADDR:%.*]] = alloca ptr addrspace(1), align 8, addrspace(5) // GFX900-NEXT: [[ID_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ID_ADDR]] to ptr // GFX900-NEXT: [[OUT_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[OUT_ADDR]] to ptr -// GFX900-NEXT: store i64 [[ID]], ptr [[ID_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA3:![0-9]+]] -// GFX900-NEXT: store ptr addrspace(1) [[OUT]], ptr [[OUT_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA7:![0-9]+]] -// GFX900-NEXT: [[TMP0:%.*]] = load i64, ptr [[ID_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA3]] -// GFX900-NEXT: [[TMP1:%.*]] = load ptr addrspace(1), ptr [[OUT_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA7]] -// GFX900-NEXT: [[TMP2:%.*]] = load i64, ptr [[ID_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA3]] +// GFX900-NEXT: store i64 [[ID]], ptr [[ID_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA7:![0-9]+]] +// GFX900-NEXT: store ptr addrspace(1) [[OUT]], ptr [[OUT_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA9:![0-9]+]] +// GFX900-NEXT: [[TMP0:%.*]] = load i64, ptr [[ID_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA7]] +// GFX900-NEXT: [[TMP1:%.*]] = load ptr addrspace(1), ptr [[OUT_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA9]] +// GFX900-NEXT: [[TMP2:%.*]] = load i64, ptr [[ID_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA7]] // GFX900-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr addrspace(1) [[TMP1]], i64 [[TMP2]] -// GFX900-NEXT: store i64 [[TMP0]], ptr addrspace(1) [[ARRAYIDX]], align 8, !tbaa [[LONG_TBAA3]] +// GFX900-NEXT: store i64 [[TMP0]], ptr addrspace(1) [[ARRAYIDX]], align 8, !tbaa [[LONG_TBAA7]] // GFX900-NEXT: ret void // // // GFX900: Function Attrs: convergent norecurse nounwind // GFX900-LABEL: define dso_local amdgpu_kernel void @test( -// GFX900-SAME: ptr addrspace(1) noundef align 1 [[A:%.*]], i8 noundef [[B:%.*]], ptr addrspace(1) noundef align 8 [[C:%.*]], i64 noundef [[D:%.*]]) #[[ATTR2:[0-9]+]] !kernel_arg_addr_space [[META10:![0-9]+]] !kernel_arg_access_qual [[META11:![0-9]+]] !kernel_arg_type [[META12:![0-9]+]] !kernel_arg_base_type [[META12]] !kernel_arg_type_qual [[META13:![0-9]+]] { +// GFX900-SAME: ptr addrspace(1) noundef align 1 [[A:%.*]], i8 noundef [[B:%.*]], ptr addrspace(1) noundef align 8 [[C:%.*]], i64 noundef [[D:%.*]]) #[[ATTR2:[0-9]+]] !kernel_arg_addr_space [[META12:![0-9]+]] !kernel_arg_access_qual [[META13:![0-9]+]] !kernel_arg_type [[META14:![0-9]+]] !kernel_arg_base_type [[META14]] !kernel_arg_type_qual [[META15:![0-9]+]] { // GFX900-NEXT: [[ENTRY:.*:]] // GFX900-NEXT: [[A_ADDR:%.*]] = alloca ptr addrspace(1), align 8, addrspace(5) // GFX900-NEXT: [[B_ADDR:%.*]] = alloca i8, align 1, addrspace(5) @@ -473,21 +473,21 @@ kernel void test_target_features_kernel(global int *i) { // GFX900-NEXT: [[B_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[B_ADDR]] to ptr // GFX900-NEXT: [[C_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[C_ADDR]] to ptr // GFX900-NEXT: [[D_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[D_ADDR]] to ptr -// GFX900-NEXT: store ptr addrspace(1) [[A]], ptr [[A_ADDR_ASCAST]], align 8, !tbaa [[CHARPTR_TBAA14:![0-9]+]] -// GFX900-NEXT: store i8 [[B]], ptr [[B_ADDR_ASCAST]], align 1, !tbaa [[CHAR_TBAA16:![0-9]+]] -// GFX900-NEXT: store ptr addrspace(1) [[C]], ptr [[C_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA7]] -// GFX900-NEXT: store i64 [[D]], ptr [[D_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA3]] -// GFX900-NEXT: [[TMP0:%.*]] = load ptr addrspace(1), ptr [[A_ADDR_ASCAST]], align 8, !tbaa [[CHARPTR_TBAA14]] -// GFX900-NEXT: [[TMP1:%.*]] = load i8, ptr [[B_ADDR_ASCAST]], align 1, !tbaa [[CHAR_TBAA16]] -// GFX900-NEXT: [[TMP2:%.*]] = load ptr addrspace(1), ptr [[C_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA7]] -// GFX900-NEXT: [[TMP3:%.*]] = load i64, ptr [[D_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA3]] +// GFX900-NEXT: store ptr addrspace(1) [[A]], ptr [[A_ADDR_ASCAST]], align 8, !tbaa [[CHARPTR_TBAA16:![0-9]+]] +// GFX900-NEXT: store i8 [[B]], ptr [[B_ADDR_ASCAST]], align 1, !tbaa [[CHAR_TBAA18:![0-9]+]] +// GFX900-NEXT: store ptr addrspace(1) [[C]], ptr [[C_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA9]] +// GFX900-NEXT: store i64 [[D]], ptr [[D_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA7]] +// GFX900-NEXT: [[TMP0:%.*]] = load ptr addrspace(1), ptr [[A_ADDR_ASCAST]], align 8, !tbaa [[CHARPTR_TBAA16]] +// GFX900-NEXT: [[TMP1:%.*]] = load i8, ptr [[B_ADDR_ASCAST]], align 1, !tbaa [[CHAR_TBAA18]] +// GFX900-NEXT: [[TMP2:%.*]] = load ptr addrspace(1), ptr [[C_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA9]] +// GFX900-NEXT: [[TMP3:%.*]] = load i64, ptr [[D_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA7]] // GFX900-NEXT: call void @__clang_ocl_kern_imp_test(ptr addrspace(1) noundef align 1 [[TMP0]], i8 noundef signext [[TMP1]], ptr addrspace(1) noundef align 8 [[TMP2]], i64 noundef [[TMP3]]) #[[ATTR8:[0-9]+]] // GFX900-NEXT: ret void // // // GFX900: Function Attrs: alwaysinline convergent norecurse nounwind // GFX900-LABEL: define dso_local void @__clang_ocl_kern_imp_test( -// GFX900-SAME: ptr addrspace(1) noundef align 1 [[A:%.*]], i8 noundef signext [[B:%.*]], ptr addrspace(1) noundef align 8 [[C:%.*]], i64 noundef [[D:%.*]]) #[[ATTR3:[0-9]+]] !kernel_arg_addr_space [[META10]] !kernel_arg_access_qual [[META11]] !kernel_arg_type [[META12]] !kernel_arg_base_type [[META12]] !kernel_arg_type_qual [[META13]] { +// GFX900-SAME: ptr addrspace(1) noundef align 1 [[A:%.*]], i8 noundef signext [[B:%.*]], ptr addrspace(1) noundef align 8 [[C:%.*]], i64 noundef [[D:%.*]]) #[[ATTR3:[0-9]+]] !kernel_arg_addr_space [[META12]] !kernel_arg_access_qual [[META13]] !kernel_arg_type [[META14]] !kernel_arg_base_type [[META14]] !kernel_arg_type_qual [[META15]] { // GFX900-NEXT: [[ENTRY:.*:]] // GFX900-NEXT: [[A_ADDR:%.*]] = alloca ptr addrspace(1), align 8, addrspace(5) // GFX900-NEXT: [[B_ADDR:%.*]] = alloca i8, align 1, addrspace(5) @@ -519,16 +519,16 @@ kernel void test_target_features_kernel(global int *i) { // GFX900-NEXT: [[BLOCK_SIZES_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[BLOCK_SIZES]] to ptr // GFX900-NEXT: [[BLOCK21_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[BLOCK21]] to ptr // GFX900-NEXT: [[TMP27_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VARTMP27]] to ptr -// GFX900-NEXT: store ptr addrspace(1) [[A]], ptr [[A_ADDR_ASCAST]], align 8, !tbaa [[CHARPTR_TBAA14]] -// GFX900-NEXT: store i8 [[B]], ptr [[B_ADDR_ASCAST]], align 1, !tbaa [[CHAR_TBAA16]] -// GFX900-NEXT: store ptr addrspace(1) [[C]], ptr [[C_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA7]] -// GFX900-NEXT: store i64 [[D]], ptr [[D_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA3]] +// GFX900-NEXT: store ptr addrspace(1) [[A]], ptr [[A_ADDR_ASCAST]], align 8, !tbaa [[CHARPTR_TBAA16]] +// GFX900-NEXT: store i8 [[B]], ptr [[B_ADDR_ASCAST]], align 1, !tbaa [[CHAR_TBAA18]] +// GFX900-NEXT: store ptr addrspace(1) [[C]], ptr [[C_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA9]] +// GFX900-NEXT: store i64 [[D]], ptr [[D_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA7]] // GFX900-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[DEFAULT_QUEUE]]) #[[ATTR9:[0-9]+]] // GFX900-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[FLAGS]]) #[[ATTR9]] -// GFX900-NEXT: store i32 0, ptr addrspace(5) [[FLAGS]], align 4, !tbaa [[INT_TBAA17:![0-9]+]] +// GFX900-NEXT: store i32 0, ptr addrspace(5) [[FLAGS]], align 4, !tbaa [[INT_TBAA3:![0-9]+]] // GFX900-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[NDRANGE]]) #[[ATTR9]] // GFX900-NEXT: [[TMP0:%.*]] = load ptr addrspace(1), ptr addrspace(5) [[DEFAULT_QUEUE]], align 8, !tbaa [[QUEUE_T_TBAA19:![0-9]+]] -// GFX900-NEXT: [[TMP1:%.*]] = load i32, ptr addrspace(5) [[FLAGS]], align 4, !tbaa [[INT_TBAA17]] +// GFX900-NEXT: [[TMP1:%.*]] = load i32, ptr addrspace(5) [[FLAGS]], align 4, !tbaa [[INT_TBAA3]] // GFX900-NEXT: call void @llvm.memcpy.p0.p5.i64(ptr align 4 [[TMP_ASCAST]], ptr addrspace(5) align 4 [[NDRANGE]], i64 4, i1 false), !tbaa.struct [[TBAA_STRUCT21:![0-9]+]] // GFX900-NEXT: [[BLOCK_SIZE:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), i8 }>, ptr [[BLOCK_ASCAST]], i32 0, i32 0 // GFX900-NEXT: store i32 25, ptr [[BLOCK_SIZE]], align 8 @@ -537,14 +537,14 @@ kernel void test_target_features_kernel(global int *i) { // GFX900-NEXT: [[BLOCK_INVOKE:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), i8 }>, ptr [[BLOCK_ASCAST]], i32 0, i32 2 // GFX900-NEXT: store ptr @__test_block_invoke, ptr [[BLOCK_INVOKE]], align 8 // GFX900-NEXT: [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), i8 }>, ptr [[BLOCK_ASCAST]], i32 0, i32 3 -// GFX900-NEXT: [[TMP2:%.*]] = load ptr addrspace(1), ptr [[A_ADDR_ASCAST]], align 8, !tbaa [[CHARPTR_TBAA14]] -// GFX900-NEXT: store ptr addrspace(1) [[TMP2]], ptr [[BLOCK_CAPTURED]], align 8, !tbaa [[CHARPTR_TBAA14]] +// GFX900-NEXT: [[TMP2:%.*]] = load ptr addrspace(1), ptr [[A_ADDR_ASCAST]], align 8, !tbaa [[CHARPTR_TBAA16]] +// GFX900-NEXT: store ptr addrspace(1) [[TMP2]], ptr [[BLOCK_CAPTURED]], align 8, !tbaa [[CHARPTR_TBAA16]] // GFX900-NEXT: [[BLOCK_CAPTURED1:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), i8 }>, ptr [[BLOCK_ASCAST]], i32 0, i32 4 -// GFX900-NEXT: [[TMP3:%.*]] = load i8, ptr [[B_ADDR_ASCAST]], align 1, !tbaa [[CHAR_TBAA16]] -// GFX900-NEXT: store i8 [[TMP3]], ptr [[BLOCK_CAPTURED1]], align 8, !tbaa [[CHAR_TBAA16]] +// GFX900-NEXT: [[TMP3:%.*]] = load i8, ptr [[B_ADDR_ASCAST]], align 1, !tbaa [[CHAR_TBAA18]] +// GFX900-NEXT: store i8 [[TMP3]], ptr [[BLOCK_CAPTURED1]], align 8, !tbaa [[CHAR_TBAA18]] // GFX900-NEXT: [[TMP4:%.*]] = call i32 @__enqueue_kernel_basic(ptr addrspace(1) [[TMP0]], i32 [[TMP1]], ptr addrspace(5) [[TMP]], ptr addrspacecast (ptr addrspace(1) @__test_block_invoke_kernel.runtime.handle to ptr), ptr [[BLOCK_ASCAST]]) // GFX900-NEXT: [[TMP5:%.*]] = load ptr addrspace(1), ptr addrspace(5) [[DEFAULT_QUEUE]], align 8, !tbaa [[QUEUE_T_TBAA19]] -// GFX900-NEXT: [[TMP6:%.*]] = load i32, ptr addrspace(5) [[FLAGS]], align 4, !tbaa [[INT_TBAA17]] +// GFX900-NEXT: [[TMP6:%.*]] = load i32, ptr addrspace(5) [[FLAGS]], align 4, !tbaa [[INT_TBAA3]] // GFX900-NEXT: call void @llvm.memcpy.p0.p5.i64(ptr align 4 [[TMP2_ASCAST]], ptr addrspace(5) align 4 [[NDRANGE]], i64 4, i1 false), !tbaa.struct [[TBAA_STRUCT21]] // GFX900-NEXT: [[BLOCK_SIZE4:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[BLOCK3_ASCAST]], i32 0, i32 0 // GFX900-NEXT: store i32 41, ptr [[BLOCK_SIZE4]], align 8 @@ -553,20 +553,20 @@ kernel void test_target_features_kernel(global int *i) { // GFX900-NEXT: [[BLOCK_INVOKE6:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[BLOCK3_ASCAST]], i32 0, i32 2 // GFX900-NEXT: store ptr @__test_block_invoke_2, ptr [[BLOCK_INVOKE6]], align 8 // GFX900-NEXT: [[BLOCK_CAPTURED7:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[BLOCK3_ASCAST]], i32 0, i32 3 -// GFX900-NEXT: [[TMP7:%.*]] = load ptr addrspace(1), ptr [[A_ADDR_ASCAST]], align 8, !tbaa [[CHARPTR_TBAA14]] -// GFX900-NEXT: store ptr addrspace(1) [[TMP7]], ptr [[BLOCK_CAPTURED7]], align 8, !tbaa [[CHARPTR_TBAA14]] +// GFX900-NEXT: [[TMP7:%.*]] = load ptr addrspace(1), ptr [[A_ADDR_ASCAST]], align 8, !tbaa [[CHARPTR_TBAA16]] +// GFX900-NEXT: store ptr addrspace(1) [[TMP7]], ptr [[BLOCK_CAPTURED7]], align 8, !tbaa [[CHARPTR_TBAA16]] // GFX900-NEXT: [[BLOCK_CAPTURED8:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[BLOCK3_ASCAST]], i32 0, i32 6 -// GFX900-NEXT: [[TMP8:%.*]] = load i8, ptr [[B_ADDR_ASCAST]], align 1, !tbaa [[CHAR_TBAA16]] -// GFX900-NEXT: store i8 [[TMP8]], ptr [[BLOCK_CAPTURED8]], align 8, !tbaa [[CHAR_TBAA16]] +// GFX900-NEXT: [[TMP8:%.*]] = load i8, ptr [[B_ADDR_ASCAST]], align 1, !tbaa [[CHAR_TBAA18]] +// GFX900-NEXT: store i8 [[TMP8]], ptr [[BLOCK_CAPTURED8]], align 8, !tbaa [[CHAR_TBAA18]] // GFX900-NEXT: [[BLOCK_CAPTURED9:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[BLOCK3_ASCAST]], i32 0, i32 4 -// GFX900-NEXT: [[TMP9:%.*]] = load ptr addrspace(1), ptr [[C_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA7]] -// GFX900-NEXT: store ptr addrspace(1) [[TMP9]], ptr [[BLOCK_CAPTURED9]], align 8, !tbaa [[LONGPTR_TBAA7]] +// GFX900-NEXT: [[TMP9:%.*]] = load ptr addrspace(1), ptr [[C_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA9]] +// GFX900-NEXT: store ptr addrspace(1) [[TMP9]], ptr [[BLOCK_CAPTURED9]], align 8, !tbaa [[LONGPTR_TBAA9]] // GFX900-NEXT: [[BLOCK_CAPTURED10:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[BLOCK3_ASCAST]], i32 0, i32 5 -// GFX900-NEXT: [[TMP10:%.*]] = load i64, ptr [[D_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA3]] -// GFX900-NEXT: store i64 [[TMP10]], ptr [[BLOCK_CAPTURED10]], align 8, !tbaa [[LONG_TBAA3]] +// GFX900-NEXT: [[TMP10:%.*]] = load i64, ptr [[D_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA7]] +// GFX900-NEXT: store i64 [[TMP10]], ptr [[BLOCK_CAPTURED10]], align 8, !tbaa [[LONG_TBAA7]] // GFX900-NEXT: [[TMP11:%.*]] = call i32 @__enqueue_kernel_basic(ptr addrspace(1) [[TMP5]], i32 [[TMP6]], ptr addrspace(5) [[VARTMP2]], ptr addrspacecast (ptr addrspace(1) @__test_block_invoke_2_kernel.runtime.handle to ptr), ptr [[BLOCK3_ASCAST]]) // GFX900-NEXT: [[TMP12:%.*]] = load ptr addrspace(1), ptr addrspace(5) [[DEFAULT_QUEUE]], align 8, !tbaa [[QUEUE_T_TBAA19]] -// GFX900-NEXT: [[TMP13:%.*]] = load i32, ptr addrspace(5) [[FLAGS]], align 4, !tbaa [[INT_TBAA17]] +// GFX900-NEXT: [[TMP13:%.*]] = load i32, ptr addrspace(5) [[FLAGS]], align 4, !tbaa [[INT_TBAA3]] // GFX900-NEXT: call void @llvm.memcpy.p0.p5.i64(ptr align 4 [[TMP11_ASCAST]], ptr addrspace(5) align 4 [[NDRANGE]], i64 4, i1 false), !tbaa.struct [[TBAA_STRUCT21]] // GFX900-NEXT: [[BLOCK_SIZE13:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[BLOCK12_ASCAST]], i32 0, i32 0 // GFX900-NEXT: store i32 41, ptr [[BLOCK_SIZE13]], align 8 @@ -575,17 +575,17 @@ kernel void test_target_features_kernel(global int *i) { // GFX900-NEXT: [[BLOCK_INVOKE15:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[BLOCK12_ASCAST]], i32 0, i32 2 // GFX900-NEXT: store ptr @__test_block_invoke_3, ptr [[BLOCK_INVOKE15]], align 8 // GFX900-NEXT: [[BLOCK_CAPTURED16:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[BLOCK12_ASCAST]], i32 0, i32 3 -// GFX900-NEXT: [[TMP14:%.*]] = load ptr addrspace(1), ptr [[A_ADDR_ASCAST]], align 8, !tbaa [[CHARPTR_TBAA14]] -// GFX900-NEXT: store ptr addrspace(1) [[TMP14]], ptr [[BLOCK_CAPTURED16]], align 8, !tbaa [[CHARPTR_TBAA14]] +// GFX900-NEXT: [[TMP14:%.*]] = load ptr addrspace(1), ptr [[A_ADDR_ASCAST]], align 8, !tbaa [[CHARPTR_TBAA16]] +// GFX900-NEXT: store ptr addrspace(1) [[TMP14]], ptr [[BLOCK_CAPTURED16]], align 8, !tbaa [[CHARPTR_TBAA16]] // GFX900-NEXT: [[BLOCK_CAPTURED17:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[BLOCK12_ASCAST]], i32 0, i32 6 -// GFX900-NEXT: [[TMP15:%.*]] = load i8, ptr [[B_ADDR_ASCAST]], align 1, !tbaa [[CHAR_TBAA16]] -// GFX900-NEXT: store i8 [[TMP15]], ptr [[BLOCK_CAPTURED17]], align 8, !tbaa [[CHAR_TBAA16]] +// GFX900-NEXT: [[TMP15:%.*]] = load i8, ptr [[B_ADDR_ASCAST]], align 1, !tbaa [[CHAR_TBAA18]] +// GFX900-NEXT: store i8 [[TMP15]], ptr [[BLOCK_CAPTURED17]], align 8, !tbaa [[CHAR_TBAA18]] // GFX900-NEXT: [[BLOCK_CAPTURED18:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[BLOCK12_ASCAST]], i32 0, i32 4 -// GFX900-NEXT: [[TMP16:%.*]] = load ptr addrspace(1), ptr [[C_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA7]] -// GFX900-NEXT: store ptr addrspace(1) [[TMP16]], ptr [[BLOCK_CAPTURED18]], align 8, !tbaa [[LONGPTR_TBAA7]] +// GFX900-NEXT: [[TMP16:%.*]] = load ptr addrspace(1), ptr [[C_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA9]] +// GFX900-NEXT: store ptr addrspace(1) [[TMP16]], ptr [[BLOCK_CAPTURED18]], align 8, !tbaa [[LONGPTR_TBAA9]] // GFX900-NEXT: [[BLOCK_CAPTURED19:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[BLOCK12_ASCAST]], i32 0, i32 5 -// GFX900-NEXT: [[TMP17:%.*]] = load i64, ptr [[D_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA3]] -// GFX900-NEXT: store i64 [[TMP17]], ptr [[BLOCK_CAPTURED19]], align 8, !tbaa [[LONG_TBAA3]] +// GFX900-NEXT: [[TMP17:%.*]] = load i64, ptr [[D_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA7]] +// GFX900-NEXT: store i64 [[TMP17]], ptr [[BLOCK_CAPTURED19]], align 8, !tbaa [[LONG_TBAA7]] // GFX900-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[BLOCK_SIZES]]) #[[ATTR9]] // GFX900-NEXT: [[TMP18:%.*]] = getelementptr [1 x i64], ptr addrspace(5) [[BLOCK_SIZES]], i32 0, i32 0 // GFX900-NEXT: store i64 100, ptr addrspace(5) [[TMP18]], align 8 @@ -599,16 +599,16 @@ kernel void test_target_features_kernel(global int *i) { // GFX900-NEXT: [[BLOCK_INVOKE24:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, i64, ptr addrspace(1) }>, ptr [[BLOCK21_ASCAST]], i32 0, i32 2 // GFX900-NEXT: store ptr @__test_block_invoke_4, ptr [[BLOCK_INVOKE24]], align 8 // GFX900-NEXT: [[BLOCK_CAPTURED25:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, i64, ptr addrspace(1) }>, ptr [[BLOCK21_ASCAST]], i32 0, i32 3 -// GFX900-NEXT: [[TMP20:%.*]] = load i64, ptr [[D_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA3]] -// GFX900-NEXT: store i64 [[TMP20]], ptr [[BLOCK_CAPTURED25]], align 8, !tbaa [[LONG_TBAA3]] +// GFX900-NEXT: [[TMP20:%.*]] = load i64, ptr [[D_ADDR_ASCAST]], align 8, !tbaa [[LONG_TBAA7]] +// GFX900-NEXT: store i64 [[TMP20]], ptr [[BLOCK_CAPTURED25]], align 8, !tbaa [[LONG_TBAA7]] // GFX900-NEXT: [[BLOCK_CAPTURED26:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, i64, ptr addrspace(1) }>, ptr [[BLOCK21_ASCAST]], i32 0, i32 4 -// GFX900-NEXT: [[TMP21:%.*]] = load ptr addrspace(1), ptr [[C_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA7]] -// GFX900-NEXT: store ptr addrspace(1) [[TMP21]], ptr [[BLOCK_CAPTURED26]], align 8, !tbaa [[LONGPTR_TBAA7]] -// GFX900-NEXT: store ptr [[BLOCK21_ASCAST]], ptr addrspace(5) [[BLOCK20]], align 8, !tbaa [[CHAR_TBAA16]] +// GFX900-NEXT: [[TMP21:%.*]] = load ptr addrspace(1), ptr [[C_ADDR_ASCAST]], align 8, !tbaa [[LONGPTR_TBAA9]] +// GFX900-NEXT: store ptr addrspace(1) [[TMP21]], ptr [[BLOCK_CAPTURED26]], align 8, !tbaa [[LONGPTR_TBAA9]] +// GFX900-NEXT: store ptr [[BLOCK21_ASCAST]], ptr addrspace(5) [[BLOCK20]], align 8, !tbaa [[CHAR_TBAA18]] // GFX900-NEXT: [[TMP22:%.*]] = load ptr addrspace(1), ptr addrspace(5) [[DEFAULT_QUEUE]], align 8, !tbaa [[QUEUE_T_TBAA19]] -// GFX900-NEXT: [[TMP23:%.*]] = load i32, ptr addrspace(5) [[FLAGS]], align 4, !tbaa [[INT_TBAA17]] +// GFX900-NEXT: [[TMP23:%.*]] = load i32, ptr addrspace(5) [[FLAGS]], align 4, !tbaa [[INT_TBAA3]] // GFX900-NEXT: call void @llvm.memcpy.p0.p5.i64(ptr align 4 [[TMP27_ASCAST]], ptr addrspace(5) align 4 [[NDRANGE]], i64 4, i1 false), !tbaa.struct [[TBAA_STRUCT21]] -// GFX900-NEXT: [[TMP24:%.*]] = load ptr, ptr addrspace(5) [[BLOCK20]], align 8, !tbaa [[CHAR_TBAA16]] +// GFX900-NEXT: [[TMP24:%.*]] = load ptr, ptr addrspace(5) [[BLOCK20]], align 8, !tbaa [[CHAR_TBAA18]] // GFX900-NEXT: [[TMP25:%.*]] = call i32 @__enqueue_kernel_basic(ptr addrspace(1) [[TMP22]], i32 [[TMP23]], ptr addrspace(5) [[VARTMP27]], ptr addrspacecast (ptr addrspace(1) @__test_block_invoke_4_kernel.runtime.handle to ptr), ptr [[BLOCK21_ASCAST]]) // GFX900-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[BLOCK20]]) #[[ATTR9]] // GFX900-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[NDRANGE]]) #[[ATTR9]] @@ -643,11 +643,11 @@ kernel void test_target_features_kernel(global int *i) { // GFX900-NEXT: store ptr addrspace(1) [[I]], ptr [[I_ADDR_ASCAST]], align 8, !tbaa [[INTPTR_TBAA26]] // GFX900-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[DEFAULT_QUEUE]]) #[[ATTR9]] // GFX900-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[FLAGS]]) #[[ATTR9]] -// GFX900-NEXT: store i32 0, ptr addrspace(5) [[FLAGS]], align 4, !tbaa [[INT_TBAA17]] +// GFX900-NEXT: store i32 0, ptr addrspace(5) [[FLAGS]], align 4, !tbaa [[INT_TBAA3]] // GFX900-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[NDRANGE]]) #[[ATTR9]] // GFX900-NEXT: [[TMP0:%.*]] = call i64 @llvm.amdgcn.s.memtime() // GFX900-NEXT: [[TMP1:%.*]] = load ptr addrspace(1), ptr addrspace(5) [[DEFAULT_QUEUE]], align 8, !tbaa [[QUEUE_T_TBAA19]] -// GFX900-NEXT: [[TMP2:%.*]] = load i32, ptr addrspace(5) [[FLAGS]], align 4, !tbaa [[INT_TBAA17]] +// GFX900-NEXT: [[TMP2:%.*]] = load i32, ptr addrspace(5) [[FLAGS]], align 4, !tbaa [[INT_TBAA3]] // GFX900-NEXT: call void @llvm.memcpy.p0.p5.i64(ptr align 4 [[TMP_ASCAST]], ptr addrspace(5) align 4 [[NDRANGE]], i64 4, i1 false), !tbaa.struct [[TBAA_STRUCT21]] // GFX900-NEXT: [[TMP3:%.*]] = call i32 @__enqueue_kernel_basic(ptr addrspace(1) [[TMP1]], i32 [[TMP2]], ptr addrspace(5) [[TMP]], ptr addrspacecast (ptr addrspace(1) @__test_target_features_kernel_block_invoke_kernel.runtime.handle to ptr), ptr addrspacecast (ptr addrspace(1) @__block_literal_global to ptr)) // GFX900-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[NDRANGE]]) #[[ATTR9]] @@ -664,11 +664,11 @@ kernel void test_target_features_kernel(global int *i) { // GFX900-NEXT: [[DOTBLOCK_DESCRIPTOR_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTBLOCK_DESCRIPTOR_ADDR]] to ptr // GFX900-NEXT: store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[DOTBLOCK_DESCRIPTOR_ADDR_ASCAST]], align 8 // GFX900-NEXT: [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), i8 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 4 -// GFX900-NEXT: [[TMP0:%.*]] = load i8, ptr [[BLOCK_CAPTURE_ADDR]], align 8, !tbaa [[CHAR_TBAA16]] +// GFX900-NEXT: [[TMP0:%.*]] = load i8, ptr [[BLOCK_CAPTURE_ADDR]], align 8, !tbaa [[CHAR_TBAA18]] // GFX900-NEXT: [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), i8 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 3 -// GFX900-NEXT: [[TMP1:%.*]] = load ptr addrspace(1), ptr [[BLOCK_CAPTURE_ADDR1]], align 8, !tbaa [[CHARPTR_TBAA14]] +// GFX900-NEXT: [[TMP1:%.*]] = load ptr addrspace(1), ptr [[BLOCK_CAPTURE_ADDR1]], align 8, !tbaa [[CHARPTR_TBAA16]] // GFX900-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[TMP1]], i64 0 -// GFX900-NEXT: store i8 [[TMP0]], ptr addrspace(1) [[ARRAYIDX]], align 1, !tbaa [[CHAR_TBAA16]] +// GFX900-NEXT: store i8 [[TMP0]], ptr addrspace(1) [[ARRAYIDX]], align 1, !tbaa [[CHAR_TBAA18]] // GFX900-NEXT: ret void // // @@ -691,17 +691,17 @@ kernel void test_target_features_kernel(global int *i) { // GFX900-NEXT: [[DOTBLOCK_DESCRIPTOR_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTBLOCK_DESCRIPTOR_ADDR]] to ptr // GFX900-NEXT: store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[DOTBLOCK_DESCRIPTOR_ADDR_ASCAST]], align 8 // GFX900-NEXT: [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 6 -// GFX900-NEXT: [[TMP0:%.*]] = load i8, ptr [[BLOCK_CAPTURE_ADDR]], align 8, !tbaa [[CHAR_TBAA16]] +// GFX900-NEXT: [[TMP0:%.*]] = load i8, ptr [[BLOCK_CAPTURE_ADDR]], align 8, !tbaa [[CHAR_TBAA18]] // GFX900-NEXT: [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 3 -// GFX900-NEXT: [[TMP1:%.*]] = load ptr addrspace(1), ptr [[BLOCK_CAPTURE_ADDR1]], align 8, !tbaa [[CHARPTR_TBAA14]] +// GFX900-NEXT: [[TMP1:%.*]] = load ptr addrspace(1), ptr [[BLOCK_CAPTURE_ADDR1]], align 8, !tbaa [[CHARPTR_TBAA16]] // GFX900-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[TMP1]], i64 0 -// GFX900-NEXT: store i8 [[TMP0]], ptr addrspace(1) [[ARRAYIDX]], align 1, !tbaa [[CHAR_TBAA16]] +// GFX900-NEXT: store i8 [[TMP0]], ptr addrspace(1) [[ARRAYIDX]], align 1, !tbaa [[CHAR_TBAA18]] // GFX900-NEXT: [[BLOCK_CAPTURE_ADDR2:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 5 -// GFX900-NEXT: [[TMP2:%.*]] = load i64, ptr [[BLOCK_CAPTURE_ADDR2]], align 8, !tbaa [[LONG_TBAA3]] +// GFX900-NEXT: [[TMP2:%.*]] = load i64, ptr [[BLOCK_CAPTURE_ADDR2]], align 8, !tbaa [[LONG_TBAA7]] // GFX900-NEXT: [[BLOCK_CAPTURE_ADDR3:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 4 -// GFX900-NEXT: [[TMP3:%.*]] = load ptr addrspace(1), ptr [[BLOCK_CAPTURE_ADDR3]], align 8, !tbaa [[LONGPTR_TBAA7]] +// GFX900-NEXT: [[TMP3:%.*]] = load ptr addrspace(1), ptr [[BLOCK_CAPTURE_ADDR3]], align 8, !tbaa [[LONGPTR_TBAA9]] // GFX900-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i64, ptr addrspace(1) [[TMP3]], i64 0 -// GFX900-NEXT: store i64 [[TMP2]], ptr addrspace(1) [[ARRAYIDX4]], align 8, !tbaa [[LONG_TBAA3]] +// GFX900-NEXT: store i64 [[TMP2]], ptr addrspace(1) [[ARRAYIDX4]], align 8, !tbaa [[LONG_TBAA7]] // GFX900-NEXT: ret void // // @@ -727,20 +727,20 @@ kernel void test_target_features_kernel(global int *i) { // GFX900-NEXT: store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[DOTBLOCK_DESCRIPTOR_ADDR_ASCAST]], align 8 // GFX900-NEXT: store ptr addrspace(3) [[LP]], ptr [[LP_ADDR_ASCAST]], align 4, !tbaa [[ANYPTR_TBAA32:![0-9]+]] // GFX900-NEXT: [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 6 -// GFX900-NEXT: [[TMP0:%.*]] = load i8, ptr [[BLOCK_CAPTURE_ADDR]], align 8, !tbaa [[CHAR_TBAA16]] +// GFX900-NEXT: [[TMP0:%.*]] = load i8, ptr [[BLOCK_CAPTURE_ADDR]], align 8, !tbaa [[CHAR_TBAA18]] // GFX900-NEXT: [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 3 -// GFX900-NEXT: [[TMP1:%.*]] = load ptr addrspace(1), ptr [[BLOCK_CAPTURE_ADDR1]], align 8, !tbaa [[CHARPTR_TBAA14]] +// GFX900-NEXT: [[TMP1:%.*]] = load ptr addrspace(1), ptr [[BLOCK_CAPTURE_ADDR1]], align 8, !tbaa [[CHARPTR_TBAA16]] // GFX900-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[TMP1]], i64 0 -// GFX900-NEXT: store i8 [[TMP0]], ptr addrspace(1) [[ARRAYIDX]], align 1, !tbaa [[CHAR_TBAA16]] +// GFX900-NEXT: store i8 [[TMP0]], ptr addrspace(1) [[ARRAYIDX]], align 1, !tbaa [[CHAR_TBAA18]] // GFX900-NEXT: [[BLOCK_CAPTURE_ADDR2:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 5 -// GFX900-NEXT: [[TMP2:%.*]] = load i64, ptr [[BLOCK_CAPTURE_ADDR2]], align 8, !tbaa [[LONG_TBAA3]] +// GFX900-NEXT: [[TMP2:%.*]] = load i64, ptr [[BLOCK_CAPTURE_ADDR2]], align 8, !tbaa [[LONG_TBAA7]] // GFX900-NEXT: [[BLOCK_CAPTURE_ADDR3:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, ptr addrspace(1), ptr addrspace(1), i64, i8 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 4 -// GFX900-NEXT: [[TMP3:%.*]] = load ptr addrspace(1), ptr [[BLOCK_CAPTURE_ADDR3]], align 8, !tbaa [[LONGPTR_TBAA7]] +// GFX900-NEXT: [[TMP3:%.*]] = load ptr addrspace(1), ptr [[BLOCK_CAPTURE_ADDR3]], align 8, !tbaa [[LONGPTR_TBAA9]] // GFX900-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i64, ptr addrspace(1) [[TMP3]], i64 0 -// GFX900-NEXT: store i64 [[TMP2]], ptr addrspace(1) [[ARRAYIDX4]], align 8, !tbaa [[LONG_TBAA3]] +// GFX900-NEXT: store i64 [[TMP2]], ptr addrspace(1) [[ARRAYIDX4]], align 8, !tbaa [[LONG_TBAA7]] // GFX900-NEXT: [[TMP4:%.*]] = load ptr addrspace(3), ptr [[LP_ADDR_ASCAST]], align 4, !tbaa [[ANYPTR_TBAA32]] // GFX900-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, ptr addrspace(3) [[TMP4]], i64 0 -// GFX900-NEXT: store i32 1, ptr addrspace(3) [[ARRAYIDX5]], align 4, !tbaa [[INT_TBAA17]] +// GFX900-NEXT: store i32 1, ptr addrspace(3) [[ARRAYIDX5]], align 4, !tbaa [[INT_TBAA3]] // GFX900-NEXT: ret void // // @@ -763,9 +763,9 @@ kernel void test_target_features_kernel(global int *i) { // GFX900-NEXT: [[DOTBLOCK_DESCRIPTOR_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[DOTBLOCK_DESCRIPTOR_ADDR]] to ptr // GFX900-NEXT: store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[DOTBLOCK_DESCRIPTOR_ADDR_ASCAST]], align 8 // GFX900-NEXT: [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, i64, ptr addrspace(1) }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 3 -// GFX900-NEXT: [[TMP0:%.*]] = load i64, ptr [[BLOCK_CAPTURE_ADDR]], align 8, !tbaa [[LONG_TBAA3]] +// GFX900-NEXT: [[TMP0:%.*]] = load i64, ptr [[BLOCK_CAPTURE_ADDR]], align 8, !tbaa [[LONG_TBAA7]] // GFX900-NEXT: [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds nuw <{ i32, i32, ptr, i64, ptr addrspace(1) }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 4 -// GFX900-NEXT: [[TMP1:%.*]] = load ptr addrspace(1), ptr [[BLOCK_CAPTURE_ADDR1]], align 8, !tbaa [[LONGPTR_TBAA7]] +// GFX900-NEXT: [[TMP1:%.*]] = load ptr addrspace(1), ptr [[BLOCK_CAPTURE_ADDR1]], align 8, !tbaa [[LONGPTR_TBAA9]] // GFX900-NEXT: call void @callee(i64 noundef [[TMP0]], ptr addrspace(1) noundef [[TMP1]]) #[[ATTR8]] // GFX900-NEXT: ret void // @@ -852,36 +852,36 @@ kernel void test_target_features_kernel(global int *i) { // GFX900: [[META0:![0-9]+]] = !{i32 1, !"amdhsa_code_object_version", i32 600} // GFX900: [[META1:![0-9]+]] = !{i32 1, !"wchar_size", i32 4} // GFX900: [[META2:![0-9]+]] = !{i32 2, i32 0} -// GFX900: [[LONG_TBAA3]] = !{[[META4:![0-9]+]], [[META4]], i64 0} -// GFX900: [[META4]] = !{!"long", [[META5:![0-9]+]], i64 0} +// GFX900: [[INT_TBAA3]] = !{[[META4:![0-9]+]], [[META4]], i64 0} +// GFX900: [[META4]] = !{!"int", [[META5:![0-9]+]], i64 0} // GFX900: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} // GFX900: [[META6]] = !{!"Simple C/C++ TBAA"} -// GFX900: [[LONGPTR_TBAA7]] = !{[[META8:![0-9]+]], [[META8]], i64 0} -// GFX900: [[META8]] = !{!"p1 long", [[META9:![0-9]+]], i64 0} -// GFX900: [[META9]] = !{!"any pointer", [[META5]], i64 0} -// GFX900: [[META10]] = !{i32 1, i32 0, i32 1, i32 0} -// GFX900: [[META11]] = !{!"none", !"none", !"none", !"none"} -// GFX900: [[META12]] = !{!"char*", !"char", !"long*", !"long"} -// GFX900: [[META13]] = !{!"", !"", !"", !""} -// GFX900: [[CHARPTR_TBAA14]] = !{[[META15:![0-9]+]], [[META15]], i64 0} -// GFX900: [[META15]] = !{!"p1 omnipotent char", [[META9]], i64 0} -// GFX900: [[CHAR_TBAA16]] = !{[[META5]], [[META5]], i64 0} -// GFX900: [[INT_TBAA17]] = !{[[META18:![0-9]+]], [[META18]], i64 0} -// GFX900: [[META18]] = !{!"int", [[META5]], i64 0} +// GFX900: [[LONG_TBAA7]] = !{[[META8:![0-9]+]], [[META8]], i64 0} +// GFX900: [[META8]] = !{!"long", [[META5]], i64 0} +// GFX900: [[LONGPTR_TBAA9]] = !{[[META10:![0-9]+]], [[META10]], i64 0} +// GFX900: [[META10]] = !{!"p1 long", [[META11:![0-9]+]], i64 0} +// GFX900: [[META11]] = !{!"any pointer", [[META5]], i64 0} +// GFX900: [[META12]] = !{i32 1, i32 0, i32 1, i32 0} +// GFX900: [[META13]] = !{!"none", !"none", !"none", !"none"} +// GFX900: [[META14]] = !{!"char*", !"char", !"long*", !"long"} +// GFX900: [[META15]] = !{!"", !"", !"", !""} +// GFX900: [[CHARPTR_TBAA16]] = !{[[META17:![0-9]+]], [[META17]], i64 0} +// GFX900: [[META17]] = !{!"p1 omnipotent char", [[META11]], i64 0} +// GFX900: [[CHAR_TBAA18]] = !{[[META5]], [[META5]], i64 0} // GFX900: [[QUEUE_T_TBAA19]] = !{[[META20:![0-9]+]], [[META20]], i64 0} // GFX900: [[META20]] = !{!"queue_t", [[META5]], i64 0} -// GFX900: [[TBAA_STRUCT21]] = !{i64 0, i64 4, [[INT_TBAA17]]} +// GFX900: [[TBAA_STRUCT21]] = !{i64 0, i64 4, [[INT_TBAA3]]} // GFX900: [[META22]] = !{i32 1} // GFX900: [[META23]] = !{!"none"} // GFX900: [[META24]] = !{!"int*"} // GFX900: [[META25]] = !{!""} // GFX900: [[INTPTR_TBAA26]] = !{[[META27:![0-9]+]], [[META27]], i64 0} -// GFX900: [[META27]] = !{!"p1 int", [[META9]], i64 0} +// GFX900: [[META27]] = !{!"p1 int", [[META11]], i64 0} // GFX900: [[META28]] = !{ptr addrspace(1) @__test_block_invoke_kernel.runtime.handle} // GFX900: [[META29]] = !{i32 0} // GFX900: [[META30]] = !{!"__block_literal"} // GFX900: [[META31]] = !{ptr addrspace(1) @__test_block_invoke_2_kernel.runtime.handle} -// GFX900: [[ANYPTR_TBAA32]] = !{[[META9]], [[META9]], i64 0} +// GFX900: [[ANYPTR_TBAA32]] = !{[[META11]], [[META11]], i64 0} // GFX900: [[META33]] = !{ptr addrspace(1) @__test_block_invoke_3_kernel.runtime.handle} // GFX900: [[META34]] = !{i32 0, i32 3} // GFX900: [[META35]] = !{!"none", !"none"} diff --git a/clang/test/CodeGenOpenCL/amdgpu-printf.cl b/clang/test/CodeGenOpenCL/amdgpu-printf.cl index cea7ee5..829f672 100644 --- a/clang/test/CodeGenOpenCL/amdgpu-printf.cl +++ b/clang/test/CodeGenOpenCL/amdgpu-printf.cl @@ -16,71 +16,71 @@ __kernel void test_printf_str_int(int i) { printf("%s:%d", s, i); } // CHECK-LABEL: define dso_local amdgpu_kernel void @test_printf_noargs( -// CHECK-SAME: ) #[[ATTR0:[0-9]+]] !kernel_arg_addr_space [[META4:![0-9]+]] !kernel_arg_access_qual [[META4]] !kernel_arg_type [[META4]] !kernel_arg_base_type [[META4]] !kernel_arg_type_qual [[META4]] { +// CHECK-SAME: ) #[[ATTR0:[0-9]+]] !kernel_arg_addr_space [[META8:![0-9]+]] !kernel_arg_access_qual [[META8]] !kernel_arg_type [[META8]] !kernel_arg_base_type [[META8]] !kernel_arg_type_qual [[META8]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: call void @__clang_ocl_kern_imp_test_printf_noargs() #[[ATTR5:[0-9]+]] // CHECK-NEXT: ret void // // // CHECK-LABEL: define dso_local void @__clang_ocl_kern_imp_test_printf_noargs( -// CHECK-SAME: ) #[[ATTR1:[0-9]+]] !kernel_arg_addr_space [[META4]] !kernel_arg_access_qual [[META4]] !kernel_arg_type [[META4]] !kernel_arg_base_type [[META4]] !kernel_arg_type_qual [[META4]] { +// CHECK-SAME: ) #[[ATTR1:[0-9]+]] !kernel_arg_addr_space [[META8]] !kernel_arg_access_qual [[META8]] !kernel_arg_type [[META8]] !kernel_arg_base_type [[META8]] !kernel_arg_type_qual [[META8]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[CALL:%.*]] = call i32 (ptr addrspace(4), ...) @printf(ptr addrspace(4) noundef @.str) #[[ATTR5]] // CHECK-NEXT: ret void // // // CHECK-LABEL: define dso_local amdgpu_kernel void @test_printf_int( -// CHECK-SAME: i32 noundef [[I:%.*]]) #[[ATTR0]] !kernel_arg_addr_space [[META5:![0-9]+]] !kernel_arg_access_qual [[META6:![0-9]+]] !kernel_arg_type [[META7:![0-9]+]] !kernel_arg_base_type [[META7]] !kernel_arg_type_qual [[META8:![0-9]+]] { +// CHECK-SAME: i32 noundef [[I:%.*]]) #[[ATTR0]] !kernel_arg_addr_space [[META9:![0-9]+]] !kernel_arg_access_qual [[META10:![0-9]+]] !kernel_arg_type [[META11:![0-9]+]] !kernel_arg_base_type [[META11]] !kernel_arg_type_qual [[META12:![0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[I_ADDR:%.*]] = alloca i32, align 4, addrspace(5) -// CHECK-NEXT: store i32 [[I]], ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA9:![0-9]+]] -// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA9]] +// CHECK-NEXT: store i32 [[I]], ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA4:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA4]] // CHECK-NEXT: call void @__clang_ocl_kern_imp_test_printf_int(i32 noundef [[TMP0]]) #[[ATTR5]] // CHECK-NEXT: ret void // // // CHECK-LABEL: define dso_local void @__clang_ocl_kern_imp_test_printf_int( -// CHECK-SAME: i32 noundef [[I:%.*]]) #[[ATTR1]] !kernel_arg_addr_space [[META5]] !kernel_arg_access_qual [[META6]] !kernel_arg_type [[META7]] !kernel_arg_base_type [[META7]] !kernel_arg_type_qual [[META8]] { +// CHECK-SAME: i32 noundef [[I:%.*]]) #[[ATTR1]] !kernel_arg_addr_space [[META9]] !kernel_arg_access_qual [[META10]] !kernel_arg_type [[META11]] !kernel_arg_base_type [[META11]] !kernel_arg_type_qual [[META12]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[I_ADDR:%.*]] = alloca i32, align 4, addrspace(5) -// CHECK-NEXT: store i32 [[I]], ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA9]] -// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA9]] +// CHECK-NEXT: store i32 [[I]], ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA4]] +// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA4]] // CHECK-NEXT: [[CALL:%.*]] = call i32 (ptr addrspace(4), ...) @printf(ptr addrspace(4) noundef @.str.1, i32 noundef [[TMP0]]) #[[ATTR5]] // CHECK-NEXT: ret void // // // CHECK-LABEL: define dso_local amdgpu_kernel void @test_printf_str_int( -// CHECK-SAME: i32 noundef [[I:%.*]]) #[[ATTR0]] !kernel_arg_addr_space [[META5]] !kernel_arg_access_qual [[META6]] !kernel_arg_type [[META7]] !kernel_arg_base_type [[META7]] !kernel_arg_type_qual [[META8]] { +// CHECK-SAME: i32 noundef [[I:%.*]]) #[[ATTR0]] !kernel_arg_addr_space [[META9]] !kernel_arg_access_qual [[META10]] !kernel_arg_type [[META11]] !kernel_arg_base_type [[META11]] !kernel_arg_type_qual [[META12]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[I_ADDR:%.*]] = alloca i32, align 4, addrspace(5) -// CHECK-NEXT: store i32 [[I]], ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA9]] -// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA9]] +// CHECK-NEXT: store i32 [[I]], ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA4]] +// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA4]] // CHECK-NEXT: call void @__clang_ocl_kern_imp_test_printf_str_int(i32 noundef [[TMP0]]) #[[ATTR5]] // CHECK-NEXT: ret void // // // CHECK-LABEL: define dso_local void @__clang_ocl_kern_imp_test_printf_str_int( -// CHECK-SAME: i32 noundef [[I:%.*]]) #[[ATTR1]] !kernel_arg_addr_space [[META5]] !kernel_arg_access_qual [[META6]] !kernel_arg_type [[META7]] !kernel_arg_base_type [[META7]] !kernel_arg_type_qual [[META8]] { +// CHECK-SAME: i32 noundef [[I:%.*]]) #[[ATTR1]] !kernel_arg_addr_space [[META9]] !kernel_arg_access_qual [[META10]] !kernel_arg_type [[META11]] !kernel_arg_base_type [[META11]] !kernel_arg_type_qual [[META12]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[I_ADDR:%.*]] = alloca i32, align 4, addrspace(5) // CHECK-NEXT: [[S:%.*]] = alloca [4 x i8], align 1, addrspace(5) -// CHECK-NEXT: store i32 [[I]], ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA9]] +// CHECK-NEXT: store i32 [[I]], ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA4]] // CHECK-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[S]]) #[[ATTR6:[0-9]+]] // CHECK-NEXT: call void @llvm.memcpy.p5.p4.i64(ptr addrspace(5) align 1 [[S]], ptr addrspace(4) align 1 @__const.test_printf_str_int.s, i64 4, i1 false) // CHECK-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [4 x i8], ptr addrspace(5) [[S]], i64 0, i64 0 -// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA9]] +// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[I_ADDR]], align 4, !tbaa [[INT_TBAA4]] // CHECK-NEXT: [[CALL:%.*]] = call i32 (ptr addrspace(4), ...) @printf(ptr addrspace(4) noundef @.str.2, ptr addrspace(5) noundef [[ARRAYDECAY]], i32 noundef [[TMP0]]) #[[ATTR5]] // CHECK-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[S]]) #[[ATTR6]] // CHECK-NEXT: ret void // //. -// CHECK: [[META4]] = !{} -// CHECK: [[META5]] = !{i32 0} -// CHECK: [[META6]] = !{!"none"} -// CHECK: [[META7]] = !{!"int"} -// CHECK: [[META8]] = !{!""} -// CHECK: [[INT_TBAA9]] = !{[[META10:![0-9]+]], [[META10]], i64 0} -// CHECK: [[META10]] = !{!"int", [[META11:![0-9]+]], i64 0} -// CHECK: [[META11]] = !{!"omnipotent char", [[META12:![0-9]+]], i64 0} -// CHECK: [[META12]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[INT_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} +// CHECK: [[META5]] = !{!"int", [[META6:![0-9]+]], i64 0} +// CHECK: [[META6]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} +// CHECK: [[META7]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META8]] = !{} +// CHECK: [[META9]] = !{i32 0} +// CHECK: [[META10]] = !{!"none"} +// CHECK: [[META11]] = !{!"int"} +// CHECK: [[META12]] = !{!""} //. diff --git a/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx12-wmma-w32.cl b/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx12-wmma-w32.cl index 321835c..6326866 100644 --- a/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx12-wmma-w32.cl +++ b/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx12-wmma-w32.cl @@ -18,7 +18,7 @@ typedef int v8i __attribute__((ext_vector_type(8))); // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <8 x half> noundef [[A:%.*]], <8 x half> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x16.f16.v8f32.v8f16(<8 x half> [[A]], <8 x half> [[B]], <8 x float> [[C]]) -// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4:![0-9]+]] +// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8:![0-9]+]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_f16_w32(global v8f* out, v8h a, v8h b, v8f c) @@ -34,7 +34,7 @@ void test_amdgcn_wmma_f32_16x16x16_f16_w32(global v8f* out, v8h a, v8h b, v8f c) // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <8 x i16> noundef [[A:%.*]], <8 x i16> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf16.v8f32.v8i16(<8 x i16> [[A]], <8 x i16> [[B]], <8 x float> [[C]]) -// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_bf16_w32(global v8f* out, v8s a, v8s b, v8f c) @@ -50,7 +50,7 @@ void test_amdgcn_wmma_f32_16x16x16_bf16_w32(global v8f* out, v8s a, v8s b, v8f c // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <8 x half> noundef [[A:%.*]], <8 x half> noundef [[B:%.*]], <8 x half> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.wmma.f16.16x16x16.f16.v8f16.v8f16(<8 x half> [[A]], <8 x half> [[B]], <8 x half> [[C]], i1 false) -// CHECK-GFX1200-NEXT: store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_f16_16x16x16_f16_w32(global v8h* out, v8h a, v8h b, v8h c) @@ -66,7 +66,7 @@ void test_amdgcn_wmma_f16_16x16x16_f16_w32(global v8h* out, v8h a, v8h b, v8h c) // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <8 x i16> noundef [[A:%.*]], <8 x i16> noundef [[B:%.*]], <8 x i16> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x i16> @llvm.amdgcn.wmma.bf16.16x16x16.bf16.v8i16.v8i16(<8 x i16> [[A]], <8 x i16> [[B]], <8 x i16> [[C]], i1 false) -// CHECK-GFX1200-NEXT: store <8 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_bf16_16x16x16_bf16_w32(global v8s* out, v8s a, v8s b, v8s c) @@ -82,7 +82,7 @@ void test_amdgcn_wmma_bf16_16x16x16_bf16_w32(global v8s* out, v8s a, v8s b, v8s // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <8 x i32> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.wmma.i32.16x16x16.iu8.v8i32.v2i32(i1 true, <2 x i32> [[A]], i1 true, <2 x i32> [[B]], <8 x i32> [[C]], i1 false) -// CHECK-GFX1200-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_i32_16x16x16_iu8_w32(global v8i* out, v2i a, v2i b, v8i c) @@ -98,7 +98,7 @@ void test_amdgcn_wmma_i32_16x16x16_iu8_w32(global v8i* out, v2i a, v2i b, v8i c) // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], i32 noundef [[A:%.*]], i32 noundef [[B:%.*]], <8 x i32> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.wmma.i32.16x16x16.iu4.v8i32.i32(i1 true, i32 [[A]], i1 true, i32 [[B]], <8 x i32> [[C]], i1 false) -// CHECK-GFX1200-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_i32_16x16x16_iu4_w32(global v8i* out, int a, int b, v8i c) @@ -110,7 +110,7 @@ void test_amdgcn_wmma_i32_16x16x16_iu4_w32(global v8i* out, int a, int b, v8i c) // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x16.fp8.fp8.v8f32.v2i32(<2 x i32> [[A]], <2 x i32> [[B]], <8 x float> [[C]]) -// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_fp8_fp8_w32(global v8f* out, v2i a, v2i b, v8f c) @@ -122,7 +122,7 @@ void test_amdgcn_wmma_f32_16x16x16_fp8_fp8_w32(global v8f* out, v2i a, v2i b, v8 // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x16.fp8.bf8.v8f32.v2i32(<2 x i32> [[A]], <2 x i32> [[B]], <8 x float> [[C]]) -// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_fp8_bf8_w32(global v8f* out, v2i a, v2i b, v8f c) @@ -134,7 +134,7 @@ void test_amdgcn_wmma_f32_16x16x16_fp8_bf8_w32(global v8f* out, v2i a, v2i b, v8 // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf8.fp8.v8f32.v2i32(<2 x i32> [[A]], <2 x i32> [[B]], <8 x float> [[C]]) -// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_bf8_fp8_w32(global v8f* out, v2i a, v2i b, v8f c) @@ -146,7 +146,7 @@ void test_amdgcn_wmma_f32_16x16x16_bf8_fp8_w32(global v8f* out, v2i a, v2i b, v8 // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf8.bf8.v8f32.v2i32(<2 x i32> [[A]], <2 x i32> [[B]], <8 x float> [[C]]) -// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_bf8_bf8_w32(global v8f* out, v2i a, v2i b, v8f c) @@ -158,7 +158,7 @@ void test_amdgcn_wmma_f32_16x16x16_bf8_bf8_w32(global v8f* out, v2i a, v2i b, v8 // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <8 x i32> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.wmma.i32.16x16x32.iu4.v8i32.v2i32(i1 true, <2 x i32> [[A]], i1 true, <2 x i32> [[B]], <8 x i32> [[C]], i1 false) -// CHECK-GFX1200-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_i32_16x16x32_iu4_w32(global v8i* out, v2i a, v2i b, v8i c) @@ -166,7 +166,7 @@ void test_amdgcn_wmma_i32_16x16x32_iu4_w32(global v8i* out, v2i a, v2i b, v8i c) *out = __builtin_amdgcn_wmma_i32_16x16x32_iu4_w32_gfx12(true, a, true, b, c, false); } //. -// CHECK-GFX1200: [[CHAR_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} -// CHECK-GFX1200: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// CHECK-GFX1200: [[META6]] = !{!"Simple C/C++ TBAA"} +// CHECK-GFX1200: [[META6:![0-9]+]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} +// CHECK-GFX1200: [[META7]] = !{!"Simple C/C++ TBAA"} +// CHECK-GFX1200: [[CHAR_TBAA8]] = !{[[META6]], [[META6]], i64 0} //. diff --git a/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx12-wmma-w64.cl b/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx12-wmma-w64.cl index 8b5b315..a79c3d4 100644 --- a/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx12-wmma-w64.cl +++ b/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx12-wmma-w64.cl @@ -17,7 +17,7 @@ typedef int v4i __attribute__((ext_vector_type(4))); // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <4 x half> noundef [[A:%.*]], <4 x half> noundef [[B:%.*]], <4 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.f16.v4f32.v4f16(<4 x half> [[A]], <4 x half> [[B]], <4 x float> [[C]]) -// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4:![0-9]+]] +// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8:![0-9]+]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_f16_w64(global v4f* out, v4h a, v4h b, v4f c) @@ -33,7 +33,7 @@ void test_amdgcn_wmma_f32_16x16x16_f16_w64(global v4f* out, v4h a, v4h b, v4f c) // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <4 x i16> noundef [[A:%.*]], <4 x i16> noundef [[B:%.*]], <4 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf16.v4f32.v4i16(<4 x i16> [[A]], <4 x i16> [[B]], <4 x float> [[C]]) -// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_bf16_w64(global v4f* out, v4s a, v4s b, v4f c) @@ -49,7 +49,7 @@ void test_amdgcn_wmma_f32_16x16x16_bf16_w64(global v4f* out, v4s a, v4s b, v4f c // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 8)) [[OUT:%.*]], <4 x half> noundef [[A:%.*]], <4 x half> noundef [[B:%.*]], <4 x half> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x half> @llvm.amdgcn.wmma.f16.16x16x16.f16.v4f16.v4f16(<4 x half> [[A]], <4 x half> [[B]], <4 x half> [[C]], i1 false) -// CHECK-GFX1200-NEXT: store <4 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 8, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 8, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_f16_16x16x16_f16_w64(global v4h* out, v4h a, v4h b, v4h c) @@ -65,7 +65,7 @@ void test_amdgcn_wmma_f16_16x16x16_f16_w64(global v4h* out, v4h a, v4h b, v4h c) // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 8)) [[OUT:%.*]], <4 x i16> noundef [[A:%.*]], <4 x i16> noundef [[B:%.*]], <4 x i16> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x i16> @llvm.amdgcn.wmma.bf16.16x16x16.bf16.v4i16.v4i16(<4 x i16> [[A]], <4 x i16> [[B]], <4 x i16> [[C]], i1 false) -// CHECK-GFX1200-NEXT: store <4 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 8, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 8, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_bf16_16x16x16_bf16_w64(global v4s* out, v4s a, v4s b, v4s c) @@ -81,7 +81,7 @@ void test_amdgcn_wmma_bf16_16x16x16_bf16_w64(global v4s* out, v4s a, v4s b, v4s // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], i32 noundef [[A:%.*]], i32 noundef [[B:%.*]], <4 x i32> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x i32> @llvm.amdgcn.wmma.i32.16x16x16.iu8.v4i32.i32(i1 true, i32 [[A]], i1 true, i32 [[B]], <4 x i32> [[C]], i1 false) -// CHECK-GFX1200-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_i32_16x16x16_iu8_w64(global v4i* out, int a, int b, v4i c) @@ -97,7 +97,7 @@ void test_amdgcn_wmma_i32_16x16x16_iu8_w64(global v4i* out, int a, int b, v4i c) // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], i32 noundef [[A:%.*]], i32 noundef [[B:%.*]], <4 x i32> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x i32> @llvm.amdgcn.wmma.i32.16x16x16.iu4.v4i32.i32(i1 true, i32 [[A]], i1 true, i32 [[B]], <4 x i32> [[C]], i1 false) -// CHECK-GFX1200-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_i32_16x16x16_iu4_w64(global v4i* out, int a, int b, v4i c) @@ -109,7 +109,7 @@ void test_amdgcn_wmma_i32_16x16x16_iu4_w64(global v4i* out, int a, int b, v4i c) // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], i32 noundef [[A:%.*]], i32 noundef [[B:%.*]], <4 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.fp8.fp8.v4f32.i32(i32 [[A]], i32 [[B]], <4 x float> [[C]]) -// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_fp8_fp8_w32(global v4f* out, int a, int b, v4f c) @@ -121,7 +121,7 @@ void test_amdgcn_wmma_f32_16x16x16_fp8_fp8_w32(global v4f* out, int a, int b, v4 // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], i32 noundef [[A:%.*]], i32 noundef [[B:%.*]], <4 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.fp8.bf8.v4f32.i32(i32 [[A]], i32 [[B]], <4 x float> [[C]]) -// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_fp8_bf8_w32(global v4f* out, int a, int b, v4f c) @@ -133,7 +133,7 @@ void test_amdgcn_wmma_f32_16x16x16_fp8_bf8_w32(global v4f* out, int a, int b, v4 // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], i32 noundef [[A:%.*]], i32 noundef [[B:%.*]], <4 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf8.fp8.v4f32.i32(i32 [[A]], i32 [[B]], <4 x float> [[C]]) -// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_bf8_fp8_w32(global v4f* out, int a, int b, v4f c) @@ -145,7 +145,7 @@ void test_amdgcn_wmma_f32_16x16x16_bf8_fp8_w32(global v4f* out, int a, int b, v4 // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], i32 noundef [[A:%.*]], i32 noundef [[B:%.*]], <4 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf8.bf8.v4f32.i32(i32 [[A]], i32 [[B]], <4 x float> [[C]]) -// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_bf8_bf8_w32(global v4f* out, int a, int b, v4f c) @@ -157,7 +157,7 @@ void test_amdgcn_wmma_f32_16x16x16_bf8_bf8_w32(global v4f* out, int a, int b, v4 // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], i32 noundef [[A:%.*]], i32 noundef [[B:%.*]], <4 x i32> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x i32> @llvm.amdgcn.wmma.i32.16x16x32.iu4.v4i32.i32(i1 true, i32 [[A]], i1 true, i32 [[B]], <4 x i32> [[C]], i1 false) -// CHECK-GFX1200-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_wmma_i32_16x16x32_iu4_w32(global v4i* out, int a, int b, v4i c) @@ -165,7 +165,7 @@ void test_amdgcn_wmma_i32_16x16x32_iu4_w32(global v4i* out, int a, int b, v4i c) *out = __builtin_amdgcn_wmma_i32_16x16x32_iu4_w64_gfx12(true, a, true, b, c, false); } //. -// CHECK-GFX1200: [[CHAR_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} -// CHECK-GFX1200: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// CHECK-GFX1200: [[META6]] = !{!"Simple C/C++ TBAA"} +// CHECK-GFX1200: [[META6:![0-9]+]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} +// CHECK-GFX1200: [[META7]] = !{!"Simple C/C++ TBAA"} +// CHECK-GFX1200: [[CHAR_TBAA8]] = !{[[META6]], [[META6]], i64 0} //. diff --git a/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx1250-async-load-store-lds.cl b/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx1250-async-load-store-lds.cl index e03ae66..2200462 100644 --- a/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx1250-async-load-store-lds.cl +++ b/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx1250-async-load-store-lds.cl @@ -152,7 +152,7 @@ void test_amdgcn_ds_atomic_async_barrier_arrive_b64(local long* addr) // CHECK-GFX1250-SAME: ptr addrspace(3) noundef captures(none) [[ADDR:%.*]], i64 noundef [[DATA:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[OUT:%.*]]) local_unnamed_addr #[[ATTR4:[0-9]+]] { // CHECK-GFX1250-NEXT: [[ENTRY:.*:]] // CHECK-GFX1250-NEXT: [[TMP0:%.*]] = tail call i64 @llvm.amdgcn.ds.atomic.barrier.arrive.rtn.b64(ptr addrspace(3) [[ADDR]], i64 [[DATA]]) -// CHECK-GFX1250-NEXT: store i64 [[TMP0]], ptr [[OUT]], align 8, !tbaa [[LONG_TBAA4:![0-9]+]] +// CHECK-GFX1250-NEXT: store i64 [[TMP0]], ptr [[OUT]], align 8, !tbaa [[LONG_TBAA8:![0-9]+]] // CHECK-GFX1250-NEXT: ret void // void test_amdgcn_ds_atomic_barrier_arrive_rtn_b64(local long* addr, long data, long *out) @@ -160,8 +160,8 @@ void test_amdgcn_ds_atomic_barrier_arrive_rtn_b64(local long* addr, long data, l *out = __builtin_amdgcn_ds_atomic_barrier_arrive_rtn_b64(addr, data); } //. -// CHECK-GFX1250: [[LONG_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} -// CHECK-GFX1250: [[META5]] = !{!"long", [[META6:![0-9]+]], i64 0} -// CHECK-GFX1250: [[META6]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} +// CHECK-GFX1250: [[META6:![0-9]+]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} // CHECK-GFX1250: [[META7]] = !{!"Simple C/C++ TBAA"} +// CHECK-GFX1250: [[LONG_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} +// CHECK-GFX1250: [[META9]] = !{!"long", [[META6]], i64 0} //. diff --git a/clang/test/CodeGenOpenCL/builtins-amdgcn-gws-insts.cl b/clang/test/CodeGenOpenCL/builtins-amdgcn-gws-insts.cl index b336720..a02c97b 100644 --- a/clang/test/CodeGenOpenCL/builtins-amdgcn-gws-insts.cl +++ b/clang/test/CodeGenOpenCL/builtins-amdgcn-gws-insts.cl @@ -13,7 +13,7 @@ typedef unsigned int uint; // CHECK-LABEL: define dso_local amdgpu_kernel void @test_builtins_amdgcn_gws_insts -// CHECK-SAME: (i32 noundef [[A:%.*]], i32 noundef [[B:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] !kernel_arg_addr_space !4 !kernel_arg_access_qual !5 !kernel_arg_type !6 !kernel_arg_base_type !6 !kernel_arg_type_qual !7 { +// CHECK-SAME: (i32 noundef [[A:%.*]], i32 noundef [[B:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] !kernel_arg_addr_space !8 !kernel_arg_access_qual !9 !kernel_arg_type !10 !kernel_arg_base_type !10 !kernel_arg_type_qual !11 { // CHECK-NEXT: entry: // CHECK-NEXT: tail call void @llvm.amdgcn.ds.gws.init(i32 [[A]], i32 [[B]]) // CHECK-NEXT: tail call void @llvm.amdgcn.ds.gws.barrier(i32 [[A]], i32 [[B]]) diff --git a/clang/test/CodeGenOpenCL/builtins-amdgcn-swmmac-w32.cl b/clang/test/CodeGenOpenCL/builtins-amdgcn-swmmac-w32.cl index 2143901..72ba191 100644 --- a/clang/test/CodeGenOpenCL/builtins-amdgcn-swmmac-w32.cl +++ b/clang/test/CodeGenOpenCL/builtins-amdgcn-swmmac-w32.cl @@ -17,7 +17,7 @@ typedef short v16s __attribute__((ext_vector_type(16))); // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <8 x half> noundef [[A:%.*]], <16 x half> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x32.f16.v8f32.v8f16.v16f16.i32(<8 x half> [[A]], <16 x half> [[B]], <8 x float> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4:![0-9]+]] +// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8:![0-9]+]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_f32_16x16x32_f16_w32(global v8f* out, v8h a, v16h b, v8f c, int index) @@ -29,7 +29,7 @@ void test_amdgcn_swmmac_f32_16x16x32_f16_w32(global v8f* out, v8h a, v16h b, v8f // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <8 x i16> noundef [[A:%.*]], <16 x i16> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf16.v8f32.v8i16.v16i16.i32(<8 x i16> [[A]], <16 x i16> [[B]], <8 x float> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_f32_16x16x32_bf16_w32(global v8f* out, v8s a, v16s b, v8f c, int index) @@ -41,7 +41,7 @@ void test_amdgcn_swmmac_f32_16x16x32_bf16_w32(global v8f* out, v8s a, v16s b, v8 // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <8 x half> noundef [[A:%.*]], <16 x half> noundef [[B:%.*]], <8 x half> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.swmmac.f16.16x16x32.f16.v8f16.v8f16.v16f16.i32(<8 x half> [[A]], <16 x half> [[B]], <8 x half> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_f16_16x16x32_f16_w32(global v8h* out, v8h a, v16h b, v8h c, int index) @@ -53,7 +53,7 @@ void test_amdgcn_swmmac_f16_16x16x32_f16_w32(global v8h* out, v8h a, v16h b, v8h // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <8 x i16> noundef [[A:%.*]], <16 x i16> noundef [[B:%.*]], <8 x i16> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x i16> @llvm.amdgcn.swmmac.bf16.16x16x32.bf16.v8i16.v8i16.v16i16.i32(<8 x i16> [[A]], <16 x i16> [[B]], <8 x i16> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <8 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_bf16_16x16x32_bf16_w32(global v8s* out, v8s a, v16s b, v8s c, int index) @@ -65,7 +65,7 @@ void test_amdgcn_swmmac_bf16_16x16x32_bf16_w32(global v8s* out, v8s a, v16s b, v // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <4 x i32> noundef [[B:%.*]], <8 x i32> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.swmmac.i32.16x16x32.iu8.v8i32.v2i32.v4i32.i32(i1 true, <2 x i32> [[A]], i1 true, <4 x i32> [[B]], <8 x i32> [[C]], i32 [[INDEX]], i1 true) -// CHECK-GFX1200-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_i32_16x16x32_iu8_w32(global v8i* out, v2i a, v4i b, v8i c, int index) @@ -77,7 +77,7 @@ void test_amdgcn_swmmac_i32_16x16x32_iu8_w32(global v8i* out, v2i a, v4i b, v8i // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], i32 noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <8 x i32> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.swmmac.i32.16x16x32.iu4.v8i32.i32.v2i32.i32(i1 true, i32 [[A]], i1 true, <2 x i32> [[B]], <8 x i32> [[C]], i32 [[INDEX]], i1 true) -// CHECK-GFX1200-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_i32_16x16x32_iu4_w32(global v8i* out, int a, v2i b, v8i c, int index) @@ -89,7 +89,7 @@ void test_amdgcn_swmmac_i32_16x16x32_iu4_w32(global v8i* out, int a, v2i b, v8i // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <4 x i32> noundef [[B:%.*]], <8 x i32> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.swmmac.i32.16x16x64.iu4.v8i32.v2i32.v4i32.i32(i1 true, <2 x i32> [[A]], i1 true, <4 x i32> [[B]], <8 x i32> [[C]], i32 [[INDEX]], i1 true) -// CHECK-GFX1200-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_i32_16x16x64_iu4_w32(global v8i* out, v2i a, v4i b, v8i c, int index) @@ -101,7 +101,7 @@ void test_amdgcn_swmmac_i32_16x16x64_iu4_w32(global v8i* out, v2i a, v4i b, v8i // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <4 x i32> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x32.fp8.fp8.v8f32.v2i32.v4i32.i32(<2 x i32> [[A]], <4 x i32> [[B]], <8 x float> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_f32_16x16x32_fp8_fp8_w32(global v8f* out, v2i a, v4i b, v8f c, int index) @@ -113,7 +113,7 @@ void test_amdgcn_swmmac_f32_16x16x32_fp8_fp8_w32(global v8f* out, v2i a, v4i b, // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <4 x i32> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x32.fp8.bf8.v8f32.v2i32.v4i32.i32(<2 x i32> [[A]], <4 x i32> [[B]], <8 x float> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_f32_16x16x32_fp8_bf8_w32(global v8f* out, v2i a, v4i b, v8f c, int index) @@ -125,7 +125,7 @@ void test_amdgcn_swmmac_f32_16x16x32_fp8_bf8_w32(global v8f* out, v2i a, v4i b, // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <4 x i32> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf8.fp8.v8f32.v2i32.v4i32.i32(<2 x i32> [[A]], <4 x i32> [[B]], <8 x float> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_f32_16x16x32_bf8_fp8_w32(global v8f* out, v2i a, v4i b, v8f c, int index) @@ -137,7 +137,7 @@ void test_amdgcn_swmmac_f32_16x16x32_bf8_fp8_w32(global v8f* out, v2i a, v4i b, // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <4 x i32> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf8.bf8.v8f32.v2i32.v4i32.i32(<2 x i32> [[A]], <4 x i32> [[B]], <8 x float> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_f32_16x16x32_bf8_bf8_w32(global v8f* out, v2i a, v4i b, v8f c, int index) @@ -145,7 +145,7 @@ void test_amdgcn_swmmac_f32_16x16x32_bf8_bf8_w32(global v8f* out, v2i a, v4i b, *out = __builtin_amdgcn_swmmac_f32_16x16x32_bf8_bf8_w32(a, b, c, index); } //. -// CHECK-GFX1200: [[CHAR_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} -// CHECK-GFX1200: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// CHECK-GFX1200: [[META6]] = !{!"Simple C/C++ TBAA"} +// CHECK-GFX1200: [[META6:![0-9]+]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} +// CHECK-GFX1200: [[META7]] = !{!"Simple C/C++ TBAA"} +// CHECK-GFX1200: [[CHAR_TBAA8]] = !{[[META6]], [[META6]], i64 0} //. diff --git a/clang/test/CodeGenOpenCL/builtins-amdgcn-swmmac-w64.cl b/clang/test/CodeGenOpenCL/builtins-amdgcn-swmmac-w64.cl index 47753af..015c493 100644 --- a/clang/test/CodeGenOpenCL/builtins-amdgcn-swmmac-w64.cl +++ b/clang/test/CodeGenOpenCL/builtins-amdgcn-swmmac-w64.cl @@ -16,7 +16,7 @@ typedef short v8s __attribute__((ext_vector_type(8))); // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <4 x half> noundef [[A:%.*]], <8 x half> noundef [[B:%.*]], <4 x float> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.f16.v4f32.v4f16.v8f16.i32(<4 x half> [[A]], <8 x half> [[B]], <4 x float> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4:![0-9]+]] +// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8:![0-9]+]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_f32_16x16x32_f16_w64(global v4f* out, v4h a, v8h b, v4f c, int index) @@ -28,7 +28,7 @@ void test_amdgcn_swmmac_f32_16x16x32_f16_w64(global v4f* out, v4h a, v8h b, v4f // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <4 x i16> noundef [[A:%.*]], <8 x i16> noundef [[B:%.*]], <4 x float> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf16.v4f32.v4i16.v8i16.i32(<4 x i16> [[A]], <8 x i16> [[B]], <4 x float> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_f32_16x16x32_bf16_w64(global v4f* out, v4s a, v8s b, v4f c, int index) @@ -40,7 +40,7 @@ void test_amdgcn_swmmac_f32_16x16x32_bf16_w64(global v4f* out, v4s a, v8s b, v4f // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 8)) [[OUT:%.*]], <4 x half> noundef [[A:%.*]], <8 x half> noundef [[B:%.*]], <4 x half> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x half> @llvm.amdgcn.swmmac.f16.16x16x32.f16.v4f16.v4f16.v8f16.i32(<4 x half> [[A]], <8 x half> [[B]], <4 x half> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <4 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 8, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 8, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_f16_16x16x32_f16_w64(global v4h* out, v4h a, v8h b, v4h c, int index) @@ -52,7 +52,7 @@ void test_amdgcn_swmmac_f16_16x16x32_f16_w64(global v4h* out, v4h a, v8h b, v4h // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 8)) [[OUT:%.*]], <4 x i16> noundef [[A:%.*]], <8 x i16> noundef [[B:%.*]], <4 x i16> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x i16> @llvm.amdgcn.swmmac.bf16.16x16x32.bf16.v4i16.v4i16.v8i16.i32(<4 x i16> [[A]], <8 x i16> [[B]], <4 x i16> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <4 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 8, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 8, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_bf16_16x16x32_bf16_w64(global v4s* out, v4s a, v8s b, v4s c, int index) @@ -64,7 +64,7 @@ void test_amdgcn_swmmac_bf16_16x16x32_bf16_w64(global v4s* out, v4s a, v8s b, v4 // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], i32 noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <4 x i32> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x i32> @llvm.amdgcn.swmmac.i32.16x16x32.iu8.v4i32.i32.v2i32.i32(i1 true, i32 [[A]], i1 true, <2 x i32> [[B]], <4 x i32> [[C]], i32 [[INDEX]], i1 true) -// CHECK-GFX1200-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_i32_16x16x32_iu8_w64(global v4i* out, int a, v2i b, v4i c, int index) @@ -76,7 +76,7 @@ void test_amdgcn_swmmac_i32_16x16x32_iu8_w64(global v4i* out, int a, v2i b, v4i // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], i32 noundef [[A:%.*]], i32 noundef [[B:%.*]], <4 x i32> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x i32> @llvm.amdgcn.swmmac.i32.16x16x32.iu4.v4i32.i32.i32.i32(i1 true, i32 [[A]], i1 true, i32 [[B]], <4 x i32> [[C]], i32 [[INDEX]], i1 true) -// CHECK-GFX1200-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_i32_16x16x32_iu4_w64(global v4i* out, int a, int b, v4i c, int index) @@ -88,7 +88,7 @@ void test_amdgcn_swmmac_i32_16x16x32_iu4_w64(global v4i* out, int a, int b, v4i // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], i32 noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <4 x i32> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x i32> @llvm.amdgcn.swmmac.i32.16x16x64.iu4.v4i32.i32.v2i32.i32(i1 true, i32 [[A]], i1 true, <2 x i32> [[B]], <4 x i32> [[C]], i32 [[INDEX]], i1 true) -// CHECK-GFX1200-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_i32_16x16x64_iu4_w64(global v4i* out, int a, v2i b, v4i c, int index) @@ -100,7 +100,7 @@ void test_amdgcn_swmmac_i32_16x16x64_iu4_w64(global v4i* out, int a, v2i b, v4i // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], i32 noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <4 x float> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.fp8.fp8.v4f32.i32.v2i32.i32(i32 [[A]], <2 x i32> [[B]], <4 x float> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_f32_16x16x32_fp8_fp8_w64(global v4f* out, int a, v2i b, v4f c, int index) @@ -112,7 +112,7 @@ void test_amdgcn_swmmac_f32_16x16x32_fp8_fp8_w64(global v4f* out, int a, v2i b, // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], i32 noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <4 x float> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.fp8.bf8.v4f32.i32.v2i32.i32(i32 [[A]], <2 x i32> [[B]], <4 x float> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_f32_16x16x32_fp8_bf8_w64(global v4f* out, int a, v2i b, v4f c, int index) @@ -124,7 +124,7 @@ void test_amdgcn_swmmac_f32_16x16x32_fp8_bf8_w64(global v4f* out, int a, v2i b, // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], i32 noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <4 x float> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf8.fp8.v4f32.i32.v2i32.i32(i32 [[A]], <2 x i32> [[B]], <4 x float> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_f32_16x16x32_bf8_fp8_w64(global v4f* out, int a, v2i b, v4f c, int index) @@ -136,7 +136,7 @@ void test_amdgcn_swmmac_f32_16x16x32_bf8_fp8_w64(global v4f* out, int a, v2i b, // CHECK-GFX1200-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], i32 noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <4 x float> noundef [[C:%.*]], i32 noundef [[INDEX:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1200-NEXT: [[ENTRY:.*:]] // CHECK-GFX1200-NEXT: [[TMP0:%.*]] = tail call <4 x float> @llvm.amdgcn.swmmac.f32.16x16x32.bf8.bf8.v4f32.i32.v2i32.i32(i32 [[A]], <2 x i32> [[B]], <4 x float> [[C]], i32 [[INDEX]]) -// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1200-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1200-NEXT: ret void // void test_amdgcn_swmmac_f32_16x16x32_bf8_bf8_w64(global v4f* out, int a, v2i b, v4f c, int index) @@ -144,7 +144,7 @@ void test_amdgcn_swmmac_f32_16x16x32_bf8_bf8_w64(global v4f* out, int a, v2i b, *out = __builtin_amdgcn_swmmac_f32_16x16x32_bf8_bf8_w64(a, b, c, index); } //. -// CHECK-GFX1200: [[CHAR_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} -// CHECK-GFX1200: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// CHECK-GFX1200: [[META6]] = !{!"Simple C/C++ TBAA"} +// CHECK-GFX1200: [[META6:![0-9]+]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} +// CHECK-GFX1200: [[META7]] = !{!"Simple C/C++ TBAA"} +// CHECK-GFX1200: [[CHAR_TBAA8]] = !{[[META6]], [[META6]], i64 0} //. diff --git a/clang/test/CodeGenOpenCL/builtins-amdgcn-vi.cl b/clang/test/CodeGenOpenCL/builtins-amdgcn-vi.cl index 6bb20bf..faf6a7d4 100644 --- a/clang/test/CodeGenOpenCL/builtins-amdgcn-vi.cl +++ b/clang/test/CodeGenOpenCL/builtins-amdgcn-vi.cl @@ -5,6 +5,8 @@ // RUN: %clang_cc1 -triple amdgcn-unknown-unknown -target-cpu gfx1012 -emit-llvm -o - %s | FileCheck --check-prefixes=CHECK,GCN %s // RUN: %clang_cc1 -triple spirv64-amd-amdhsa -emit-llvm -o - %s | FileCheck --check-prefixes=CHECK,AMDGCNSPIRV %s +#define INVALID_MEMORY_SCOPE (__MEMORY_SCOPE_CLUSTR+1) + #pragma OPENCL EXTENSION cl_khr_fp16 : enable typedef unsigned long ulong; @@ -252,13 +254,19 @@ void test_update_dpp_const_int(global int* out, int arg1) // CHECK: atomicrmw fadd ptr addrspace(3) %out, float %src seq_cst, align 4{{$}} // CHECK: atomicrmw fadd ptr addrspace(3) %out, float %src seq_cst, align 4{{$}} -// GCN: atomicrmw fadd ptr addrspace(3) %out, float %src syncscope("agent") monotonic, align 4{{$}} +// GCN: atomicrmw fadd ptr addrspace(3) %out, float %src syncscope("agent") monotonic, align 4{{$}} // AMDGCNSPIRV: atomicrmw fadd ptr addrspace(3) %out, float %src syncscope("device") monotonic, align 4{{$}} -// CHECK: atomicrmw fadd ptr addrspace(3) %out, float %src syncscope("workgroup") monotonic, align 4{{$}} -// GCN: atomicrmw fadd ptr addrspace(3) %out, float %src syncscope("wavefront") monotonic, align 4{{$}} + +// CHECK: atomicrmw fadd ptr addrspace(3) %out, float %src syncscope("workgroup") monotonic, align 4{{$}} + +// GCN: atomicrmw fadd ptr addrspace(3) %out, float %src syncscope("cluster") monotonic, align 4{{$}} +// AMDGCNSPIRV: atomicrmw fadd ptr addrspace(3) %out, float %src syncscope("workgroup") monotonic, align 4{{$}} + +// GCN: atomicrmw fadd ptr addrspace(3) %out, float %src syncscope("wavefront") monotonic, align 4{{$}} // AMDGCNSPIRV: atomicrmw fadd ptr addrspace(3) %out, float %src syncscope("subgroup") monotonic, align 4{{$}} -// CHECK: atomicrmw fadd ptr addrspace(3) %out, float %src syncscope("singlethread") monotonic, align 4{{$}} -// CHECK: atomicrmw fadd ptr addrspace(3) %out, float %src monotonic, align 4{{$}} + +// CHECK: atomicrmw fadd ptr addrspace(3) %out, float %src syncscope("singlethread") monotonic, align 4{{$}} +// CHECK: atomicrmw fadd ptr addrspace(3) %out, float %src monotonic, align 4{{$}} #if !defined(__SPIRV__) void test_ds_faddf(local float *out, float src) { #else @@ -279,9 +287,10 @@ void test_ds_faddf(local float *out, float src) { // Test all syncscopes. *out = __builtin_amdgcn_ds_faddf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_DEVICE, false); *out = __builtin_amdgcn_ds_faddf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_WRKGRP, false); + *out = __builtin_amdgcn_ds_faddf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR, false); *out = __builtin_amdgcn_ds_faddf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_WVFRNT, false); *out = __builtin_amdgcn_ds_faddf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_SINGLE, false); - *out = __builtin_amdgcn_ds_faddf(out, src, __ATOMIC_RELAXED, 5, false); // invalid + *out = __builtin_amdgcn_ds_faddf(out, src, __ATOMIC_RELAXED, INVALID_MEMORY_SCOPE, false); // invalid } // CHECK-LABEL: @test_ds_fmin @@ -295,13 +304,19 @@ void test_ds_faddf(local float *out, float src) { // CHECK: atomicrmw fmin ptr addrspace(3) %out, float %src seq_cst, align 4{{$}} // CHECK: atomicrmw fmin ptr addrspace(3) %out, float %src seq_cst, align 4{{$}} -// GCN: atomicrmw fmin ptr addrspace(3) %out, float %src syncscope("agent") monotonic, align 4{{$}} +// GCN: atomicrmw fmin ptr addrspace(3) %out, float %src syncscope("agent") monotonic, align 4{{$}} // AMDGCNSPIRV: atomicrmw fmin ptr addrspace(3) %out, float %src syncscope("device") monotonic, align 4{{$}} -// CHECK: atomicrmw fmin ptr addrspace(3) %out, float %src syncscope("workgroup") monotonic, align 4{{$}} -// GCN: atomicrmw fmin ptr addrspace(3) %out, float %src syncscope("wavefront") monotonic, align 4{{$}} + +// CHECK: atomicrmw fmin ptr addrspace(3) %out, float %src syncscope("workgroup") monotonic, align 4{{$}} + +// GCN: atomicrmw fmin ptr addrspace(3) %out, float %src syncscope("cluster") monotonic, align 4{{$}} +// AMDGCNSPIRV: atomicrmw fmin ptr addrspace(3) %out, float %src syncscope("workgroup") monotonic, align 4{{$}} + +// GCN: atomicrmw fmin ptr addrspace(3) %out, float %src syncscope("wavefront") monotonic, align 4{{$}} // AMDGCNSPIRV: atomicrmw fmin ptr addrspace(3) %out, float %src syncscope("subgroup") monotonic, align 4{{$}} -// CHECK: atomicrmw fmin ptr addrspace(3) %out, float %src syncscope("singlethread") monotonic, align 4{{$}} -// CHECK: atomicrmw fmin ptr addrspace(3) %out, float %src monotonic, align 4{{$}} + +// CHECK: atomicrmw fmin ptr addrspace(3) %out, float %src syncscope("singlethread") monotonic, align 4{{$}} +// CHECK: atomicrmw fmin ptr addrspace(3) %out, float %src monotonic, align 4{{$}} #if !defined(__SPIRV__) void test_ds_fminf(local float *out, float src) { @@ -322,9 +337,10 @@ void test_ds_fminf(__attribute__((address_space(3))) float *out, float src) { // Test all syncscopes. *out = __builtin_amdgcn_ds_fminf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_DEVICE, false); *out = __builtin_amdgcn_ds_fminf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_WRKGRP, false); + *out = __builtin_amdgcn_ds_fminf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR, false); *out = __builtin_amdgcn_ds_fminf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_WVFRNT, false); *out = __builtin_amdgcn_ds_fminf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_SINGLE, false); - *out = __builtin_amdgcn_ds_fminf(out, src, __ATOMIC_RELAXED, 5, false); // invalid + *out = __builtin_amdgcn_ds_fminf(out, src, __ATOMIC_RELAXED, INVALID_MEMORY_SCOPE, false); // invalid } // CHECK-LABEL: @test_ds_fmax @@ -338,13 +354,19 @@ void test_ds_fminf(__attribute__((address_space(3))) float *out, float src) { // CHECK: atomicrmw fmax ptr addrspace(3) %out, float %src seq_cst, align 4{{$}} // CHECK: atomicrmw fmax ptr addrspace(3) %out, float %src seq_cst, align 4{{$}} -// GCN: atomicrmw fmax ptr addrspace(3) %out, float %src syncscope("agent") monotonic, align 4{{$}} +// GCN: atomicrmw fmax ptr addrspace(3) %out, float %src syncscope("agent") monotonic, align 4{{$}} // AMDGCNSPIRV: atomicrmw fmax ptr addrspace(3) %out, float %src syncscope("device") monotonic, align 4{{$}} -// CHECK: atomicrmw fmax ptr addrspace(3) %out, float %src syncscope("workgroup") monotonic, align 4{{$}} -// GCN: atomicrmw fmax ptr addrspace(3) %out, float %src syncscope("wavefront") monotonic, align 4{{$}} + +// CHECK: atomicrmw fmax ptr addrspace(3) %out, float %src syncscope("workgroup") monotonic, align 4{{$}} + +// GCN: atomicrmw fmax ptr addrspace(3) %out, float %src syncscope("cluster") monotonic, align 4{{$}} +// AMDGCNSPIRV: atomicrmw fmax ptr addrspace(3) %out, float %src syncscope("workgroup") monotonic, align 4{{$}} + +// GCN: atomicrmw fmax ptr addrspace(3) %out, float %src syncscope("wavefront") monotonic, align 4{{$}} // AMDGCNSPIRV: atomicrmw fmax ptr addrspace(3) %out, float %src syncscope("subgroup") monotonic, align 4{{$}} -// CHECK: atomicrmw fmax ptr addrspace(3) %out, float %src syncscope("singlethread") monotonic, align 4{{$}} -// CHECK: atomicrmw fmax ptr addrspace(3) %out, float %src monotonic, align 4{{$}} + +// CHECK: atomicrmw fmax ptr addrspace(3) %out, float %src syncscope("singlethread") monotonic, align 4{{$}} +// CHECK: atomicrmw fmax ptr addrspace(3) %out, float %src monotonic, align 4{{$}} #if !defined(__SPIRV__) void test_ds_fmaxf(local float *out, float src) { @@ -365,9 +387,10 @@ void test_ds_fmaxf(__attribute__((address_space(3))) float *out, float src) { // Test all syncscopes. *out = __builtin_amdgcn_ds_fmaxf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_DEVICE, false); *out = __builtin_amdgcn_ds_fmaxf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_WRKGRP, false); + *out = __builtin_amdgcn_ds_fmaxf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_CLUSTR, false); *out = __builtin_amdgcn_ds_fmaxf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_WVFRNT, false); *out = __builtin_amdgcn_ds_fmaxf(out, src, __ATOMIC_RELAXED, __MEMORY_SCOPE_SINGLE, false); - *out = __builtin_amdgcn_ds_fmaxf(out, src, __ATOMIC_RELAXED, 5, false); // invalid + *out = __builtin_amdgcn_ds_fmaxf(out, src, __ATOMIC_RELAXED, INVALID_MEMORY_SCOPE, false); // invalid } // CHECK-LABEL: @test_s_memtime diff --git a/clang/test/CodeGenOpenCL/builtins-amdgcn-wmma-w32.cl b/clang/test/CodeGenOpenCL/builtins-amdgcn-wmma-w32.cl index 853cd32..2f3ab11 100644 --- a/clang/test/CodeGenOpenCL/builtins-amdgcn-wmma-w32.cl +++ b/clang/test/CodeGenOpenCL/builtins-amdgcn-wmma-w32.cl @@ -21,7 +21,7 @@ typedef short v16s __attribute__((ext_vector_type(16))); // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <16 x half> noundef [[A:%.*]], <16 x half> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x16.f16.v8f32.v16f16(<16 x half> [[A]], <16 x half> [[B]], <8 x float> [[C]]) -// CHECK-GFX1100-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4:![0-9]+]] +// CHECK-GFX1100-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8:![0-9]+]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_f16_w32(global v8f* out, v16h a, v16h b, v8f c) @@ -37,7 +37,7 @@ void test_amdgcn_wmma_f32_16x16x16_f16_w32(global v8f* out, v16h a, v16h b, v8f // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <16 x i16> noundef [[A:%.*]], <16 x i16> noundef [[B:%.*]], <8 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf16.v8f32.v16i16(<16 x i16> [[A]], <16 x i16> [[B]], <8 x float> [[C]]) -// CHECK-GFX1100-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1100-NEXT: store <8 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_bf16_w32(global v8f* out, v16s a, v16s b, v8f c) @@ -53,7 +53,7 @@ void test_amdgcn_wmma_f32_16x16x16_bf16_w32(global v8f* out, v16s a, v16s b, v8f // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <16 x half> noundef [[A:%.*]], <16 x half> noundef [[B:%.*]], <16 x half> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <16 x half> @llvm.amdgcn.wmma.f16.16x16x16.f16.v16f16.v16f16(<16 x half> [[A]], <16 x half> [[B]], <16 x half> [[C]], i1 true) -// CHECK-GFX1100-NEXT: store <16 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1100-NEXT: store <16 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_f16_16x16x16_f16_w32(global v16h* out, v16h a, v16h b, v16h c) @@ -69,7 +69,7 @@ void test_amdgcn_wmma_f16_16x16x16_f16_w32(global v16h* out, v16h a, v16h b, v16 // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <16 x i16> noundef [[A:%.*]], <16 x i16> noundef [[B:%.*]], <16 x i16> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <16 x i16> @llvm.amdgcn.wmma.bf16.16x16x16.bf16.v16i16.v16i16(<16 x i16> [[A]], <16 x i16> [[B]], <16 x i16> [[C]], i1 true) -// CHECK-GFX1100-NEXT: store <16 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1100-NEXT: store <16 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_bf16_16x16x16_bf16_w32(global v16s* out, v16s a, v16s b, v16s c) @@ -85,7 +85,7 @@ void test_amdgcn_wmma_bf16_16x16x16_bf16_w32(global v16s* out, v16s a, v16s b, v // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <16 x half> noundef [[A:%.*]], <16 x half> noundef [[B:%.*]], <16 x half> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <16 x half> @llvm.amdgcn.wmma.f16.16x16x16.f16.tied.v16f16.v16f16(<16 x half> [[A]], <16 x half> [[B]], <16 x half> [[C]], i1 true) -// CHECK-GFX1100-NEXT: store <16 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1100-NEXT: store <16 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_f16_16x16x16_f16_tied_w32(global v16h* out, v16h a, v16h b, v16h c) @@ -101,7 +101,7 @@ void test_amdgcn_wmma_f16_16x16x16_f16_tied_w32(global v16h* out, v16h a, v16h b // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <16 x i16> noundef [[A:%.*]], <16 x i16> noundef [[B:%.*]], <16 x i16> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <16 x i16> @llvm.amdgcn.wmma.bf16.16x16x16.bf16.tied.v16i16.v16i16(<16 x i16> [[A]], <16 x i16> [[B]], <16 x i16> [[C]], i1 true) -// CHECK-GFX1100-NEXT: store <16 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1100-NEXT: store <16 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_bf16_16x16x16_bf16_tied_w32(global v16s* out, v16s a, v16s b, v16s c) @@ -117,7 +117,7 @@ void test_amdgcn_wmma_bf16_16x16x16_bf16_tied_w32(global v16s* out, v16s a, v16s // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <4 x i32> noundef [[A:%.*]], <4 x i32> noundef [[B:%.*]], <8 x i32> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.wmma.i32.16x16x16.iu8.v8i32.v4i32(i1 true, <4 x i32> [[A]], i1 true, <4 x i32> [[B]], <8 x i32> [[C]], i1 false) -// CHECK-GFX1100-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1100-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_i32_16x16x16_iu8_w32(global v8i* out, v4i a, v4i b, v8i c) @@ -133,7 +133,7 @@ void test_amdgcn_wmma_i32_16x16x16_iu8_w32(global v8i* out, v4i a, v4i b, v8i c) // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 32)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <8 x i32> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <8 x i32> @llvm.amdgcn.wmma.i32.16x16x16.iu4.v8i32.v2i32(i1 true, <2 x i32> [[A]], i1 true, <2 x i32> [[B]], <8 x i32> [[C]], i1 false) -// CHECK-GFX1100-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1100-NEXT: store <8 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 32, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_i32_16x16x16_iu4_w32(global v8i* out, v2i a, v2i b, v8i c) @@ -143,7 +143,7 @@ void test_amdgcn_wmma_i32_16x16x16_iu4_w32(global v8i* out, v2i a, v2i b, v8i c) #endif //. -// CHECK-GFX1100: [[CHAR_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} -// CHECK-GFX1100: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// CHECK-GFX1100: [[META6]] = !{!"Simple C/C++ TBAA"} +// CHECK-GFX1100: [[META6:![0-9]+]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} +// CHECK-GFX1100: [[META7]] = !{!"Simple C/C++ TBAA"} +// CHECK-GFX1100: [[CHAR_TBAA8]] = !{[[META6]], [[META6]], i64 0} //. diff --git a/clang/test/CodeGenOpenCL/builtins-amdgcn-wmma-w64.cl b/clang/test/CodeGenOpenCL/builtins-amdgcn-wmma-w64.cl index 9b6872f..024a928 100644 --- a/clang/test/CodeGenOpenCL/builtins-amdgcn-wmma-w64.cl +++ b/clang/test/CodeGenOpenCL/builtins-amdgcn-wmma-w64.cl @@ -22,7 +22,7 @@ typedef short v16s __attribute__((ext_vector_type(16))); // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <16 x half> noundef [[A:%.*]], <16 x half> noundef [[B:%.*]], <4 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.f16.v4f32.v16f16(<16 x half> [[A]], <16 x half> [[B]], <4 x float> [[C]]) -// CHECK-GFX1100-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4:![0-9]+]] +// CHECK-GFX1100-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8:![0-9]+]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_f16_w64(global v4f* out, v16h a, v16h b, v4f c) @@ -38,7 +38,7 @@ void test_amdgcn_wmma_f32_16x16x16_f16_w64(global v4f* out, v16h a, v16h b, v4f // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <16 x i16> noundef [[A:%.*]], <16 x i16> noundef [[B:%.*]], <4 x float> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.bf16.v4f32.v16i16(<16 x i16> [[A]], <16 x i16> [[B]], <4 x float> [[C]]) -// CHECK-GFX1100-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1100-NEXT: store <4 x float> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_f32_16x16x16_bf16_w64(global v4f* out, v16s a, v16s b, v4f c) @@ -54,7 +54,7 @@ void test_amdgcn_wmma_f32_16x16x16_bf16_w64(global v4f* out, v16s a, v16s b, v4f // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <16 x half> noundef [[A:%.*]], <16 x half> noundef [[B:%.*]], <8 x half> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.wmma.f16.16x16x16.f16.v8f16.v16f16(<16 x half> [[A]], <16 x half> [[B]], <8 x half> [[C]], i1 true) -// CHECK-GFX1100-NEXT: store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1100-NEXT: store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_f16_16x16x16_f16_w64(global v8h* out, v16h a, v16h b, v8h c) @@ -70,7 +70,7 @@ void test_amdgcn_wmma_f16_16x16x16_f16_w64(global v8h* out, v16h a, v16h b, v8h // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <16 x i16> noundef [[A:%.*]], <16 x i16> noundef [[B:%.*]], <8 x i16> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <8 x i16> @llvm.amdgcn.wmma.bf16.16x16x16.bf16.v8i16.v16i16(<16 x i16> [[A]], <16 x i16> [[B]], <8 x i16> [[C]], i1 true) -// CHECK-GFX1100-NEXT: store <8 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1100-NEXT: store <8 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_bf16_16x16x16_bf16_w64(global v8s* out, v16s a, v16s b, v8s c) @@ -86,7 +86,7 @@ void test_amdgcn_wmma_bf16_16x16x16_bf16_w64(global v8s* out, v16s a, v16s b, v8 // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <16 x half> noundef [[A:%.*]], <16 x half> noundef [[B:%.*]], <8 x half> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <8 x half> @llvm.amdgcn.wmma.f16.16x16x16.f16.tied.v8f16.v16f16(<16 x half> [[A]], <16 x half> [[B]], <8 x half> [[C]], i1 true) -// CHECK-GFX1100-NEXT: store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1100-NEXT: store <8 x half> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_f16_16x16x16_f16_tied_w64(global v8h* out, v16h a, v16h b, v8h c) @@ -102,7 +102,7 @@ void test_amdgcn_wmma_f16_16x16x16_f16_tied_w64(global v8h* out, v16h a, v16h b, // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <16 x i16> noundef [[A:%.*]], <16 x i16> noundef [[B:%.*]], <8 x i16> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <8 x i16> @llvm.amdgcn.wmma.bf16.16x16x16.bf16.tied.v8i16.v16i16(<16 x i16> [[A]], <16 x i16> [[B]], <8 x i16> [[C]], i1 true) -// CHECK-GFX1100-NEXT: store <8 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1100-NEXT: store <8 x i16> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_bf16_16x16x16_bf16_tied_w64(global v8s* out, v16s a, v16s b, v8s c) @@ -118,7 +118,7 @@ void test_amdgcn_wmma_bf16_16x16x16_bf16_tied_w64(global v8s* out, v16s a, v16s // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <4 x i32> noundef [[A:%.*]], <4 x i32> noundef [[B:%.*]], <4 x i32> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <4 x i32> @llvm.amdgcn.wmma.i32.16x16x16.iu8.v4i32.v4i32(i1 true, <4 x i32> [[A]], i1 true, <4 x i32> [[B]], <4 x i32> [[C]], i1 false) -// CHECK-GFX1100-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1100-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_i32_16x16x16_iu8_w64(global v4i* out, v4i a, v4i b, v4i c) @@ -134,7 +134,7 @@ void test_amdgcn_wmma_i32_16x16x16_iu8_w64(global v4i* out, v4i a, v4i b, v4i c) // CHECK-GFX1100-SAME: ptr addrspace(1) noundef writeonly captures(none) initializes((0, 16)) [[OUT:%.*]], <2 x i32> noundef [[A:%.*]], <2 x i32> noundef [[B:%.*]], <4 x i32> noundef [[C:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-GFX1100-NEXT: [[ENTRY:.*:]] // CHECK-GFX1100-NEXT: [[TMP0:%.*]] = tail call <4 x i32> @llvm.amdgcn.wmma.i32.16x16x16.iu4.v4i32.v2i32(i1 true, <2 x i32> [[A]], i1 true, <2 x i32> [[B]], <4 x i32> [[C]], i1 false) -// CHECK-GFX1100-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA4]] +// CHECK-GFX1100-NEXT: store <4 x i32> [[TMP0]], ptr addrspace(1) [[OUT]], align 16, !tbaa [[CHAR_TBAA8]] // CHECK-GFX1100-NEXT: ret void // void test_amdgcn_wmma_i32_16x16x16_iu4_w64(global v4i* out, v2i a, v2i b, v4i c) @@ -144,7 +144,7 @@ void test_amdgcn_wmma_i32_16x16x16_iu4_w64(global v4i* out, v2i a, v2i b, v4i c) #endif //. -// CHECK-GFX1100: [[CHAR_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} -// CHECK-GFX1100: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// CHECK-GFX1100: [[META6]] = !{!"Simple C/C++ TBAA"} +// CHECK-GFX1100: [[META6:![0-9]+]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} +// CHECK-GFX1100: [[META7]] = !{!"Simple C/C++ TBAA"} +// CHECK-GFX1100: [[CHAR_TBAA8]] = !{[[META6]], [[META6]], i64 0} //. diff --git a/clang/test/CodeGenOpenCL/preserve_vec3.cl b/clang/test/CodeGenOpenCL/preserve_vec3.cl index 6e5c1c4..e76aa81 100644 --- a/clang/test/CodeGenOpenCL/preserve_vec3.cl +++ b/clang/test/CodeGenOpenCL/preserve_vec3.cl @@ -9,11 +9,11 @@ typedef float float3 __attribute__((ext_vector_type(3))); typedef float float4 __attribute__((ext_vector_type(4))); // CHECK-LABEL: define dso_local spir_kernel void @foo( -// CHECK-SAME: ptr addrspace(1) noundef readonly align 16 captures(none) [[A:%.*]], ptr addrspace(1) noundef writeonly align 16 captures(none) initializes((0, 16)) [[B:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] !kernel_arg_addr_space [[META3:![0-9]+]] !kernel_arg_access_qual [[META4:![0-9]+]] !kernel_arg_type [[META5:![0-9]+]] !kernel_arg_base_type [[META6:![0-9]+]] !kernel_arg_type_qual [[META7:![0-9]+]] { +// CHECK-SAME: ptr addrspace(1) noundef readonly align 16 captures(none) [[A:%.*]], ptr addrspace(1) noundef writeonly align 16 captures(none) initializes((0, 16)) [[B:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] !kernel_arg_addr_space [[META7:![0-9]+]] !kernel_arg_access_qual [[META8:![0-9]+]] !kernel_arg_type [[META9:![0-9]+]] !kernel_arg_base_type [[META10:![0-9]+]] !kernel_arg_type_qual [[META11:![0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = load <3 x float>, ptr addrspace(1) [[A]], align 16 // CHECK-NEXT: [[EXTRACTVEC1_I:%.*]] = shufflevector <3 x float> [[TMP0]], <3 x float> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 poison> -// CHECK-NEXT: store <4 x float> [[EXTRACTVEC1_I]], ptr addrspace(1) [[B]], align 16, !tbaa [[CHAR_TBAA8:![0-9]+]] +// CHECK-NEXT: store <4 x float> [[EXTRACTVEC1_I]], ptr addrspace(1) [[B]], align 16, !tbaa [[CHAR_TBAA12:![0-9]+]] // CHECK-NEXT: ret void // void kernel foo(global float3 *a, global float3 *b) { @@ -21,11 +21,11 @@ void kernel foo(global float3 *a, global float3 *b) { } // CHECK-LABEL: define dso_local spir_kernel void @float4_to_float3( -// CHECK-SAME: ptr addrspace(1) noundef writeonly align 16 captures(none) initializes((0, 16)) [[A:%.*]], ptr addrspace(1) noundef readonly align 16 captures(none) [[B:%.*]]) local_unnamed_addr #[[ATTR0]] !kernel_arg_addr_space [[META3]] !kernel_arg_access_qual [[META4]] !kernel_arg_type [[META11:![0-9]+]] !kernel_arg_base_type [[META12:![0-9]+]] !kernel_arg_type_qual [[META7]] { +// CHECK-SAME: ptr addrspace(1) noundef writeonly align 16 captures(none) initializes((0, 16)) [[A:%.*]], ptr addrspace(1) noundef readonly align 16 captures(none) [[B:%.*]]) local_unnamed_addr #[[ATTR0]] !kernel_arg_addr_space [[META7]] !kernel_arg_access_qual [[META8]] !kernel_arg_type [[META13:![0-9]+]] !kernel_arg_base_type [[META14:![0-9]+]] !kernel_arg_type_qual [[META11]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <3 x float>, ptr addrspace(1) [[B]], align 16, !tbaa [[CHAR_TBAA8]] +// CHECK-NEXT: [[TMP0:%.*]] = load <3 x float>, ptr addrspace(1) [[B]], align 16, !tbaa [[CHAR_TBAA12]] // CHECK-NEXT: [[EXTRACTVEC_I:%.*]] = shufflevector <3 x float> [[TMP0]], <3 x float> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 poison> -// CHECK-NEXT: store <4 x float> [[EXTRACTVEC_I]], ptr addrspace(1) [[A]], align 16, !tbaa [[CHAR_TBAA8]] +// CHECK-NEXT: store <4 x float> [[EXTRACTVEC_I]], ptr addrspace(1) [[A]], align 16, !tbaa [[CHAR_TBAA12]] // CHECK-NEXT: ret void // void kernel float4_to_float3(global float3 *a, global float4 *b) { @@ -33,11 +33,11 @@ void kernel float4_to_float3(global float3 *a, global float4 *b) { } // CHECK-LABEL: define dso_local spir_kernel void @float3_to_float4( -// CHECK-SAME: ptr addrspace(1) noundef readonly align 16 captures(none) [[A:%.*]], ptr addrspace(1) noundef writeonly align 16 captures(none) initializes((0, 16)) [[B:%.*]]) local_unnamed_addr #[[ATTR0]] !kernel_arg_addr_space [[META3]] !kernel_arg_access_qual [[META4]] !kernel_arg_type [[META11]] !kernel_arg_base_type [[META12]] !kernel_arg_type_qual [[META7]] { +// CHECK-SAME: ptr addrspace(1) noundef readonly align 16 captures(none) [[A:%.*]], ptr addrspace(1) noundef writeonly align 16 captures(none) initializes((0, 16)) [[B:%.*]]) local_unnamed_addr #[[ATTR0]] !kernel_arg_addr_space [[META7]] !kernel_arg_access_qual [[META8]] !kernel_arg_type [[META13]] !kernel_arg_base_type [[META14]] !kernel_arg_type_qual [[META11]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = load <3 x float>, ptr addrspace(1) [[A]], align 16 // CHECK-NEXT: [[ASTYPE_I:%.*]] = shufflevector <3 x float> [[TMP0]], <3 x float> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 poison> -// CHECK-NEXT: store <4 x float> [[ASTYPE_I]], ptr addrspace(1) [[B]], align 16, !tbaa [[CHAR_TBAA8]] +// CHECK-NEXT: store <4 x float> [[ASTYPE_I]], ptr addrspace(1) [[B]], align 16, !tbaa [[CHAR_TBAA12]] // CHECK-NEXT: ret void // void kernel float3_to_float4(global float3 *a, global float4 *b) { @@ -45,11 +45,11 @@ void kernel float3_to_float4(global float3 *a, global float4 *b) { } // CHECK-LABEL: define dso_local spir_kernel void @float3_to_double2( -// CHECK-SAME: ptr addrspace(1) noundef readonly align 16 captures(none) [[A:%.*]], ptr addrspace(1) noundef writeonly align 16 captures(none) initializes((0, 16)) [[B:%.*]]) local_unnamed_addr #[[ATTR0]] !kernel_arg_addr_space [[META3]] !kernel_arg_access_qual [[META4]] !kernel_arg_type [[META13:![0-9]+]] !kernel_arg_base_type [[META14:![0-9]+]] !kernel_arg_type_qual [[META7]] { +// CHECK-SAME: ptr addrspace(1) noundef readonly align 16 captures(none) [[A:%.*]], ptr addrspace(1) noundef writeonly align 16 captures(none) initializes((0, 16)) [[B:%.*]]) local_unnamed_addr #[[ATTR0]] !kernel_arg_addr_space [[META7]] !kernel_arg_access_qual [[META8]] !kernel_arg_type [[META15:![0-9]+]] !kernel_arg_base_type [[META16:![0-9]+]] !kernel_arg_type_qual [[META11]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = load <3 x float>, ptr addrspace(1) [[A]], align 16 // CHECK-NEXT: [[TMP1:%.*]] = shufflevector <3 x float> [[TMP0]], <3 x float> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 poison> -// CHECK-NEXT: store <4 x float> [[TMP1]], ptr addrspace(1) [[B]], align 16, !tbaa [[CHAR_TBAA8]] +// CHECK-NEXT: store <4 x float> [[TMP1]], ptr addrspace(1) [[B]], align 16, !tbaa [[CHAR_TBAA12]] // CHECK-NEXT: ret void // void kernel float3_to_double2(global float3 *a, global double2 *b) { @@ -57,11 +57,11 @@ void kernel float3_to_double2(global float3 *a, global double2 *b) { } // CHECK-LABEL: define dso_local spir_kernel void @char8_to_short3( -// CHECK-SAME: ptr addrspace(1) noundef writeonly align 8 captures(none) initializes((0, 8)) [[A:%.*]], ptr addrspace(1) noundef readonly align 8 captures(none) [[B:%.*]]) local_unnamed_addr #[[ATTR0]] !kernel_arg_addr_space [[META3]] !kernel_arg_access_qual [[META4]] !kernel_arg_type [[META15:![0-9]+]] !kernel_arg_base_type [[META16:![0-9]+]] !kernel_arg_type_qual [[META7]] { +// CHECK-SAME: ptr addrspace(1) noundef writeonly align 8 captures(none) initializes((0, 8)) [[A:%.*]], ptr addrspace(1) noundef readonly align 8 captures(none) [[B:%.*]]) local_unnamed_addr #[[ATTR0]] !kernel_arg_addr_space [[META7]] !kernel_arg_access_qual [[META8]] !kernel_arg_type [[META17:![0-9]+]] !kernel_arg_base_type [[META18:![0-9]+]] !kernel_arg_type_qual [[META11]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <3 x i16>, ptr addrspace(1) [[B]], align 8, !tbaa [[CHAR_TBAA8]] +// CHECK-NEXT: [[TMP0:%.*]] = load <3 x i16>, ptr addrspace(1) [[B]], align 8, !tbaa [[CHAR_TBAA12]] // CHECK-NEXT: [[EXTRACTVEC_I:%.*]] = shufflevector <3 x i16> [[TMP0]], <3 x i16> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 poison> -// CHECK-NEXT: store <4 x i16> [[EXTRACTVEC_I]], ptr addrspace(1) [[A]], align 8, !tbaa [[CHAR_TBAA8]] +// CHECK-NEXT: store <4 x i16> [[EXTRACTVEC_I]], ptr addrspace(1) [[A]], align 8, !tbaa [[CHAR_TBAA12]] // CHECK-NEXT: ret void // void kernel char8_to_short3(global short3 *a, global char8 *b) { @@ -72,7 +72,7 @@ void kernel char8_to_short3(global short3 *a, global char8 *b) { // CHECK-SAME: <3 x i8> noundef [[A:%.*]], ptr addrspace(1) noundef writeonly captures(none) initializes((0, 4)) [[OUT:%.*]]) local_unnamed_addr #[[ATTR2:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[EXTRACTVEC:%.*]] = shufflevector <3 x i8> [[A]], <3 x i8> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 poison> -// CHECK-NEXT: store <4 x i8> [[EXTRACTVEC]], ptr addrspace(1) [[OUT]], align 4, !tbaa [[INT_TBAA17:![0-9]+]] +// CHECK-NEXT: store <4 x i8> [[EXTRACTVEC]], ptr addrspace(1) [[OUT]], align 4, !tbaa [[INT_TBAA3:![0-9]+]] // CHECK-NEXT: ret void // void from_char3(char3 a, global int *out) { @@ -95,7 +95,7 @@ void from_short3(short3 a, global long *out) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = bitcast i32 [[A]] to <4 x i8> // CHECK-NEXT: [[EXTRACTVEC:%.*]] = shufflevector <4 x i8> [[TMP0]], <4 x i8> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 poison> -// CHECK-NEXT: store <4 x i8> [[EXTRACTVEC]], ptr addrspace(1) [[OUT]], align 4, !tbaa [[CHAR_TBAA8]] +// CHECK-NEXT: store <4 x i8> [[EXTRACTVEC]], ptr addrspace(1) [[OUT]], align 4, !tbaa [[CHAR_TBAA12]] // CHECK-NEXT: ret void // void scalar_to_char3(int a, global char3 *out) { @@ -107,7 +107,7 @@ void scalar_to_char3(int a, global char3 *out) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = bitcast i64 [[A]] to <4 x i16> // CHECK-NEXT: [[EXTRACTVEC:%.*]] = shufflevector <4 x i16> [[TMP0]], <4 x i16> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 poison> -// CHECK-NEXT: store <4 x i16> [[EXTRACTVEC]], ptr addrspace(1) [[OUT]], align 8, !tbaa [[CHAR_TBAA8]] +// CHECK-NEXT: store <4 x i16> [[EXTRACTVEC]], ptr addrspace(1) [[OUT]], align 8, !tbaa [[CHAR_TBAA12]] // CHECK-NEXT: ret void // void scalar_to_short3(long a, global short3 *out) { @@ -115,22 +115,22 @@ void scalar_to_short3(long a, global short3 *out) { } //. -// CHECK: [[META3]] = !{i32 1, i32 1} -// CHECK: [[META4]] = !{!"none", !"none"} -// CHECK: [[META5]] = !{!"float3*", !"float3*"} -// CHECK: [[META6]] = !{!"float __attribute__((ext_vector_type(3)))*", !"float __attribute__((ext_vector_type(3)))*"} -// CHECK: [[META7]] = !{!"", !""} -// CHECK: [[CHAR_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} -// CHECK: [[META9]] = !{!"omnipotent char", [[META10:![0-9]+]], i64 0} -// CHECK: [[META10]] = !{!"Simple C/C++ TBAA"} -// CHECK: [[META11]] = !{!"float3*", !"float4*"} -// CHECK: [[META12]] = !{!"float __attribute__((ext_vector_type(3)))*", !"float __attribute__((ext_vector_type(4)))*"} -// CHECK: [[META13]] = !{!"float3*", !"double2*"} -// CHECK: [[META14]] = !{!"float __attribute__((ext_vector_type(3)))*", !"double __attribute__((ext_vector_type(2)))*"} -// CHECK: [[META15]] = !{!"short3*", !"char8*"} -// CHECK: [[META16]] = !{!"short __attribute__((ext_vector_type(3)))*", !"char __attribute__((ext_vector_type(8)))*"} -// CHECK: [[INT_TBAA17]] = !{[[META18:![0-9]+]], [[META18]], i64 0} -// CHECK: [[META18]] = !{!"int", [[META9]], i64 0} +// CHECK: [[INT_TBAA3]] = !{[[META4:![0-9]+]], [[META4]], i64 0} +// CHECK: [[META4]] = !{!"int", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} +// CHECK: [[META6]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META7]] = !{i32 1, i32 1} +// CHECK: [[META8]] = !{!"none", !"none"} +// CHECK: [[META9]] = !{!"float3*", !"float3*"} +// CHECK: [[META10]] = !{!"float __attribute__((ext_vector_type(3)))*", !"float __attribute__((ext_vector_type(3)))*"} +// CHECK: [[META11]] = !{!"", !""} +// CHECK: [[CHAR_TBAA12]] = !{[[META5]], [[META5]], i64 0} +// CHECK: [[META13]] = !{!"float3*", !"float4*"} +// CHECK: [[META14]] = !{!"float __attribute__((ext_vector_type(3)))*", !"float __attribute__((ext_vector_type(4)))*"} +// CHECK: [[META15]] = !{!"float3*", !"double2*"} +// CHECK: [[META16]] = !{!"float __attribute__((ext_vector_type(3)))*", !"double __attribute__((ext_vector_type(2)))*"} +// CHECK: [[META17]] = !{!"short3*", !"char8*"} +// CHECK: [[META18]] = !{!"short __attribute__((ext_vector_type(3)))*", !"char __attribute__((ext_vector_type(8)))*"} // CHECK: [[LONG_TBAA19]] = !{[[META20:![0-9]+]], [[META20]], i64 0} -// CHECK: [[META20]] = !{!"long", [[META9]], i64 0} +// CHECK: [[META20]] = !{!"long", [[META5]], i64 0} //. diff --git a/clang/test/CodeGenOpenCLCXX/array-type-infinite-loop.clcpp b/clang/test/CodeGenOpenCLCXX/array-type-infinite-loop.clcpp index e932e75..07f5929 100644 --- a/clang/test/CodeGenOpenCLCXX/array-type-infinite-loop.clcpp +++ b/clang/test/CodeGenOpenCLCXX/array-type-infinite-loop.clcpp @@ -2,11 +2,11 @@ //RUN: %clang_cc1 %s -triple spir -emit-llvm -O1 -o - | FileCheck %s // CHECK-LABEL: define dso_local spir_kernel void @test( -// CHECK-SAME: ptr addrspace(1) noundef readonly align 8 captures(none) [[IN:%.*]], ptr addrspace(1) noundef writeonly align 8 captures(none) initializes((0, 8)) [[OUT:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] !kernel_arg_addr_space [[META4:![0-9]+]] !kernel_arg_access_qual [[META5:![0-9]+]] !kernel_arg_type [[META6:![0-9]+]] !kernel_arg_base_type [[META6]] !kernel_arg_type_qual [[META7:![0-9]+]] { +// CHECK-SAME: ptr addrspace(1) noundef readonly align 8 captures(none) [[IN:%.*]], ptr addrspace(1) noundef writeonly align 8 captures(none) initializes((0, 8)) [[OUT:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] !kernel_arg_addr_space [[META8:![0-9]+]] !kernel_arg_access_qual [[META9:![0-9]+]] !kernel_arg_type [[META10:![0-9]+]] !kernel_arg_base_type [[META10]] !kernel_arg_type_qual [[META11:![0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[ARRAYIDX1_I:%.*]] = getelementptr inbounds nuw i8, ptr addrspace(1) [[IN]], i32 8 -// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr addrspace(1) [[ARRAYIDX1_I]], align 8, !tbaa [[LONG_TBAA8:![0-9]+]] -// CHECK-NEXT: store i64 [[TMP0]], ptr addrspace(1) [[OUT]], align 8, !tbaa [[LONG_TBAA8]] +// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr addrspace(1) [[ARRAYIDX1_I]], align 8, !tbaa [[LONG_TBAA12:![0-9]+]] +// CHECK-NEXT: store i64 [[TMP0]], ptr addrspace(1) [[OUT]], align 8, !tbaa [[LONG_TBAA12]] // CHECK-NEXT: ret void // __kernel void test(__global long *In, __global long *Out) { @@ -14,12 +14,12 @@ __kernel void test(__global long *In, __global long *Out) { *Out = m[1]; } //. -// CHECK: [[META4]] = !{i32 1, i32 1} -// CHECK: [[META5]] = !{!"none", !"none"} -// CHECK: [[META6]] = !{!"long*", !"long*"} -// CHECK: [[META7]] = !{!"", !""} -// CHECK: [[LONG_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0} -// CHECK: [[META9]] = !{!"long", [[META10:![0-9]+]], i64 0} -// CHECK: [[META10]] = !{!"omnipotent char", [[META11:![0-9]+]], i64 0} -// CHECK: [[META11]] = !{!"Simple C++ TBAA"} +// CHECK: [[META6:![0-9]+]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} +// CHECK: [[META7]] = !{!"Simple C++ TBAA"} +// CHECK: [[META8]] = !{i32 1, i32 1} +// CHECK: [[META9]] = !{!"none", !"none"} +// CHECK: [[META10]] = !{!"long*", !"long*"} +// CHECK: [[META11]] = !{!"", !""} +// CHECK: [[LONG_TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} +// CHECK: [[META13]] = !{!"long", [[META6]], i64 0} //. diff --git a/clang/test/DebugInfo/Generic/unsigned-promotion-debuginfo.c b/clang/test/DebugInfo/Generic/unsigned-promotion-debuginfo.c index 6ca17e1..d50e0be 100644 --- a/clang/test/DebugInfo/Generic/unsigned-promotion-debuginfo.c +++ b/clang/test/DebugInfo/Generic/unsigned-promotion-debuginfo.c @@ -12,34 +12,34 @@ unsigned short si, sj, sk; // CHECKS-LABEL: define dso_local void @testshortmul( -// CHECKS-SAME: ) local_unnamed_addr #[[ATTR0:[0-9]+]] !dbg [[DBG13:![0-9]+]] { +// CHECKS-SAME: ) local_unnamed_addr #[[ATTR0:[0-9]+]] !dbg [[DBG17:![0-9]+]] { // CHECKS-NEXT: [[ENTRY:.*:]] -// CHECKS-NEXT: [[TMP0:%.*]] = load i16, ptr @sj, align 2, !dbg [[DBG16:![0-9]+]], !tbaa [[SHORT_TBAA17:![0-9]+]] -// CHECKS-NEXT: [[CONV:%.*]] = zext i16 [[TMP0]] to i32, !dbg [[DBG16]] -// CHECKS-NEXT: [[TMP1:%.*]] = load i16, ptr @sk, align 2, !dbg [[DBG21:![0-9]+]], !tbaa [[SHORT_TBAA17]] -// CHECKS-NEXT: [[CONV1:%.*]] = zext i16 [[TMP1]] to i32, !dbg [[DBG21]] -// CHECKS-NEXT: [[TMP2:%.*]] = tail call { i32, i1 } @llvm.smul.with.overflow.i32(i32 [[CONV]], i32 [[CONV1]]), !dbg [[DBG22:![0-9]+]], !nosanitize [[META26:![0-9]+]] -// CHECKS-NEXT: [[TMP3:%.*]] = extractvalue { i32, i1 } [[TMP2]], 1, !dbg [[DBG22]], !nosanitize [[META26]] -// CHECKS-NEXT: br i1 [[TMP3]], label %[[HANDLER_MUL_OVERFLOW:.*]], label %[[CONT:.*]], !dbg [[DBG22]], !prof [[PROF27:![0-9]+]], !nosanitize [[META26]] +// CHECKS-NEXT: [[TMP0:%.*]] = load i16, ptr @sj, align 2, !dbg [[DBG20:![0-9]+]], !tbaa [[SHORT_TBAA21:![0-9]+]] +// CHECKS-NEXT: [[CONV:%.*]] = zext i16 [[TMP0]] to i32, !dbg [[DBG20]] +// CHECKS-NEXT: [[TMP1:%.*]] = load i16, ptr @sk, align 2, !dbg [[DBG23:![0-9]+]], !tbaa [[SHORT_TBAA21]] +// CHECKS-NEXT: [[CONV1:%.*]] = zext i16 [[TMP1]] to i32, !dbg [[DBG23]] +// CHECKS-NEXT: [[TMP2:%.*]] = tail call { i32, i1 } @llvm.smul.with.overflow.i32(i32 [[CONV]], i32 [[CONV1]]), !dbg [[DBG24:![0-9]+]], !nosanitize [[META28:![0-9]+]] +// CHECKS-NEXT: [[TMP3:%.*]] = extractvalue { i32, i1 } [[TMP2]], 1, !dbg [[DBG24]], !nosanitize [[META28]] +// CHECKS-NEXT: br i1 [[TMP3]], label %[[HANDLER_MUL_OVERFLOW:.*]], label %[[CONT:.*]], !dbg [[DBG24]], !prof [[PROF29:![0-9]+]], !nosanitize [[META28]] // CHECKS: [[HANDLER_MUL_OVERFLOW]]: -// CHECKS-NEXT: [[TMP4:%.*]] = zext i16 [[TMP0]] to i64, !dbg [[DBG22]] -// CHECKS-NEXT: [[TMP5:%.*]] = zext i16 [[TMP1]] to i64, !dbg [[DBG22]] -// CHECKS-NEXT: tail call void @__ubsan_handle_mul_overflow_abort(ptr nonnull @[[GLOB1:[0-9]+]], i64 [[TMP4]], i64 [[TMP5]]) #[[ATTR3:[0-9]+]], !dbg [[DBG22]], !nosanitize [[META26]] -// CHECKS-NEXT: unreachable, !dbg [[DBG22]], !nosanitize [[META26]] +// CHECKS-NEXT: [[TMP4:%.*]] = zext i16 [[TMP0]] to i64, !dbg [[DBG24]] +// CHECKS-NEXT: [[TMP5:%.*]] = zext i16 [[TMP1]] to i64, !dbg [[DBG24]] +// CHECKS-NEXT: tail call void @__ubsan_handle_mul_overflow_abort(ptr nonnull @[[GLOB1:[0-9]+]], i64 [[TMP4]], i64 [[TMP5]]) #[[ATTR3:[0-9]+]], !dbg [[DBG24]], !nosanitize [[META28]] +// CHECKS-NEXT: unreachable, !dbg [[DBG24]], !nosanitize [[META28]] // CHECKS: [[CONT]]: -// CHECKS-NEXT: [[TMP6:%.*]] = extractvalue { i32, i1 } [[TMP2]], 0, !dbg [[DBG22]], !nosanitize [[META26]] -// CHECKS-NEXT: [[CONV2:%.*]] = trunc i32 [[TMP6]] to i16, !dbg [[DBG16]] -// CHECKS-NEXT: store i16 [[CONV2]], ptr @si, align 2, !dbg [[DBG28:![0-9]+]], !tbaa [[SHORT_TBAA17]] -// CHECKS-NEXT: ret void, !dbg [[DBG29:![0-9]+]] +// CHECKS-NEXT: [[TMP6:%.*]] = extractvalue { i32, i1 } [[TMP2]], 0, !dbg [[DBG24]], !nosanitize [[META28]] +// CHECKS-NEXT: [[CONV2:%.*]] = trunc i32 [[TMP6]] to i16, !dbg [[DBG20]] +// CHECKS-NEXT: store i16 [[CONV2]], ptr @si, align 2, !dbg [[DBG30:![0-9]+]], !tbaa [[SHORT_TBAA21]] +// CHECKS-NEXT: ret void, !dbg [[DBG31:![0-9]+]] // // CHECKU-LABEL: define dso_local void @testshortmul( -// CHECKU-SAME: ) local_unnamed_addr #[[ATTR0:[0-9]+]] !dbg [[DBG13:![0-9]+]] { +// CHECKU-SAME: ) local_unnamed_addr #[[ATTR0:[0-9]+]] !dbg [[DBG17:![0-9]+]] { // CHECKU-NEXT: [[ENTRY:.*:]] -// CHECKU-NEXT: [[TMP0:%.*]] = load i16, ptr @sj, align 2, !dbg [[DBG16:![0-9]+]], !tbaa [[SHORT_TBAA17:![0-9]+]] -// CHECKU-NEXT: [[TMP1:%.*]] = load i16, ptr @sk, align 2, !dbg [[DBG21:![0-9]+]], !tbaa [[SHORT_TBAA17]] -// CHECKU-NEXT: [[MUL:%.*]] = mul i16 [[TMP1]], [[TMP0]], !dbg [[DBG22:![0-9]+]] -// CHECKU-NEXT: store i16 [[MUL]], ptr @si, align 2, !dbg [[DBG23:![0-9]+]], !tbaa [[SHORT_TBAA17]] -// CHECKU-NEXT: ret void, !dbg [[DBG24:![0-9]+]] +// CHECKU-NEXT: [[TMP0:%.*]] = load i16, ptr @sj, align 2, !dbg [[DBG20:![0-9]+]], !tbaa [[SHORT_TBAA21:![0-9]+]] +// CHECKU-NEXT: [[TMP1:%.*]] = load i16, ptr @sk, align 2, !dbg [[DBG23:![0-9]+]], !tbaa [[SHORT_TBAA21]] +// CHECKU-NEXT: [[MUL:%.*]] = mul i16 [[TMP1]], [[TMP0]], !dbg [[DBG24:![0-9]+]] +// CHECKU-NEXT: store i16 [[MUL]], ptr @si, align 2, !dbg [[DBG25:![0-9]+]], !tbaa [[SHORT_TBAA21]] +// CHECKU-NEXT: ret void, !dbg [[DBG26:![0-9]+]] // void testshortmul(void) { @@ -58,23 +58,23 @@ void testshortmul(void) { // CHECKS: [[META8]] = !DIBasicType(name: "unsigned short", size: 16, encoding: DW_ATE_unsigned) // CHECKS: [[META9]] = !DIGlobalVariableExpression(var: [[META10:![0-9]+]], expr: !DIExpression()) // CHECKS: [[META10]] = distinct !DIGlobalVariable(name: "sk", scope: [[META2]], file: [[META7]], line: 12, type: [[META8]], isLocal: false, isDefinition: true) -// CHECKS: [[DBG13]] = distinct !DISubprogram(name: "testshortmul", scope: [[META7]], file: [[META7]], line: 44, type: [[META14:![0-9]+]], scopeLine: 44, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: [[META2]]) -// CHECKS: [[META14]] = !DISubroutineType(types: [[META15:![0-9]+]]) -// CHECKS: [[META15]] = !{null} -// CHECKS: [[DBG16]] = !DILocation(line: 47, column: 8, scope: [[DBG13]]) -// CHECKS: [[SHORT_TBAA17]] = !{[[META18:![0-9]+]], [[META18]], i64 0} -// CHECKS: [[META18]] = !{!"short", [[META19:![0-9]+]], i64 0} -// CHECKS: [[META19]] = !{!"omnipotent char", [[META20:![0-9]+]], i64 0} -// CHECKS: [[META20]] = !{!"Simple C/C++ TBAA"} -// CHECKS: [[DBG21]] = !DILocation(line: 47, column: 13, scope: [[DBG13]]) -// CHECKS: [[DBG22]] = !DILocation(line: 0, scope: [[META23:![0-9]+]], inlinedAt: [[META25:![0-9]+]]) -// CHECKS: [[META23]] = distinct !DISubprogram(name: "__ubsan_check_mul_overflow", scope: [[META7]], file: [[META7]], type: [[META24:![0-9]+]], flags: DIFlagArtificial, spFlags: DISPFlagDefinition, unit: [[META2]]) -// CHECKS: [[META24]] = !DISubroutineType(types: null) -// CHECKS: [[META25]] = !DILocation(line: 47, column: 11, scope: [[DBG13]]) -// CHECKS: [[META26]] = !{} -// CHECKS: [[PROF27]] = !{!"branch_weights", i32 1, i32 1048575} -// CHECKS: [[DBG28]] = !DILocation(line: 47, column: 6, scope: [[DBG13]]) -// CHECKS: [[DBG29]] = !DILocation(line: 48, column: 1, scope: [[DBG13]]) +// CHECKS: [[META15:![0-9]+]] = !{!"omnipotent char", [[META16:![0-9]+]], i64 0} +// CHECKS: [[META16]] = !{!"Simple C/C++ TBAA"} +// CHECKS: [[DBG17]] = distinct !DISubprogram(name: "testshortmul", scope: [[META7]], file: [[META7]], line: 44, type: [[META18:![0-9]+]], scopeLine: 44, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: [[META2]]) +// CHECKS: [[META18]] = !DISubroutineType(types: [[META19:![0-9]+]]) +// CHECKS: [[META19]] = !{null} +// CHECKS: [[DBG20]] = !DILocation(line: 47, column: 8, scope: [[DBG17]]) +// CHECKS: [[SHORT_TBAA21]] = !{[[META22:![0-9]+]], [[META22]], i64 0} +// CHECKS: [[META22]] = !{!"short", [[META15]], i64 0} +// CHECKS: [[DBG23]] = !DILocation(line: 47, column: 13, scope: [[DBG17]]) +// CHECKS: [[DBG24]] = !DILocation(line: 0, scope: [[META25:![0-9]+]], inlinedAt: [[META27:![0-9]+]]) +// CHECKS: [[META25]] = distinct !DISubprogram(name: "__ubsan_check_mul_overflow", scope: [[META7]], file: [[META7]], type: [[META26:![0-9]+]], flags: DIFlagArtificial, spFlags: DISPFlagDefinition, unit: [[META2]]) +// CHECKS: [[META26]] = !DISubroutineType(types: null) +// CHECKS: [[META27]] = !DILocation(line: 47, column: 11, scope: [[DBG17]]) +// CHECKS: [[META28]] = !{} +// CHECKS: [[PROF29]] = !{!"branch_weights", i32 1, i32 1048575} +// CHECKS: [[DBG30]] = !DILocation(line: 47, column: 6, scope: [[DBG17]]) +// CHECKS: [[DBG31]] = !DILocation(line: 48, column: 1, scope: [[DBG17]]) //. // CHECKU: [[META0:![0-9]+]] = !DIGlobalVariableExpression(var: [[META1:![0-9]+]], expr: !DIExpression()) // CHECKU: [[META1]] = distinct !DIGlobalVariable(name: "sj", scope: [[META2:![0-9]+]], file: [[META7:![0-9]+]], line: 12, type: [[META8:![0-9]+]], isLocal: false, isDefinition: true) @@ -87,16 +87,16 @@ void testshortmul(void) { // CHECKU: [[META8]] = !DIBasicType(name: "unsigned short", size: 16, encoding: DW_ATE_unsigned) // CHECKU: [[META9]] = !DIGlobalVariableExpression(var: [[META10:![0-9]+]], expr: !DIExpression()) // CHECKU: [[META10]] = distinct !DIGlobalVariable(name: "sk", scope: [[META2]], file: [[META7]], line: 12, type: [[META8]], isLocal: false, isDefinition: true) -// CHECKU: [[DBG13]] = distinct !DISubprogram(name: "testshortmul", scope: [[META7]], file: [[META7]], line: 44, type: [[META14:![0-9]+]], scopeLine: 44, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: [[META2]]) -// CHECKU: [[META14]] = !DISubroutineType(types: [[META15:![0-9]+]]) -// CHECKU: [[META15]] = !{null} -// CHECKU: [[DBG16]] = !DILocation(line: 47, column: 8, scope: [[DBG13]]) -// CHECKU: [[SHORT_TBAA17]] = !{[[META18:![0-9]+]], [[META18]], i64 0} -// CHECKU: [[META18]] = !{!"short", [[META19:![0-9]+]], i64 0} -// CHECKU: [[META19]] = !{!"omnipotent char", [[META20:![0-9]+]], i64 0} -// CHECKU: [[META20]] = !{!"Simple C/C++ TBAA"} -// CHECKU: [[DBG21]] = !DILocation(line: 47, column: 13, scope: [[DBG13]]) -// CHECKU: [[DBG22]] = !DILocation(line: 47, column: 11, scope: [[DBG13]]) -// CHECKU: [[DBG23]] = !DILocation(line: 47, column: 6, scope: [[DBG13]]) -// CHECKU: [[DBG24]] = !DILocation(line: 48, column: 1, scope: [[DBG13]]) +// CHECKU: [[META15:![0-9]+]] = !{!"omnipotent char", [[META16:![0-9]+]], i64 0} +// CHECKU: [[META16]] = !{!"Simple C/C++ TBAA"} +// CHECKU: [[DBG17]] = distinct !DISubprogram(name: "testshortmul", scope: [[META7]], file: [[META7]], line: 44, type: [[META18:![0-9]+]], scopeLine: 44, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: [[META2]]) +// CHECKU: [[META18]] = !DISubroutineType(types: [[META19:![0-9]+]]) +// CHECKU: [[META19]] = !{null} +// CHECKU: [[DBG20]] = !DILocation(line: 47, column: 8, scope: [[DBG17]]) +// CHECKU: [[SHORT_TBAA21]] = !{[[META22:![0-9]+]], [[META22]], i64 0} +// CHECKU: [[META22]] = !{!"short", [[META15]], i64 0} +// CHECKU: [[DBG23]] = !DILocation(line: 47, column: 13, scope: [[DBG17]]) +// CHECKU: [[DBG24]] = !DILocation(line: 47, column: 11, scope: [[DBG17]]) +// CHECKU: [[DBG25]] = !DILocation(line: 47, column: 6, scope: [[DBG17]]) +// CHECKU: [[DBG26]] = !DILocation(line: 48, column: 1, scope: [[DBG17]]) //. diff --git a/clang/test/Driver/arm-abi.c b/clang/test/Driver/arm-abi.c index 139456c..b89b969 100644 --- a/clang/test/Driver/arm-abi.c +++ b/clang/test/Driver/arm-abi.c @@ -31,6 +31,8 @@ // FreeBSD / OpenBSD default to aapcs-linux // RUN: %clang -target arm--freebsd- %s -### -o %t.o 2>&1 \ // RUN: | FileCheck -check-prefix=CHECK-AAPCS-LINUX %s +// RUN: %clang -target arm--fuchsia- %s -### -o %t.o 2>&1 \ +// RUN: | FileCheck -check-prefix=CHECK-AAPCS-LINUX %s // RUN: %clang -target arm--openbsd- %s -### -o %t.o 2>&1 \ // RUN: | FileCheck -check-prefix=CHECK-AAPCS-LINUX %s // RUN: %clang -target arm--haiku- %s -### -o %t.o 2>&1 \ diff --git a/clang/test/Driver/fsanitize-alloc-token.c b/clang/test/Driver/fsanitize-alloc-token.c index 2964f60..6d8bda1 100644 --- a/clang/test/Driver/fsanitize-alloc-token.c +++ b/clang/test/Driver/fsanitize-alloc-token.c @@ -41,3 +41,14 @@ // CHECK-MAX: "-falloc-token-max=42" // RUN: not %clang --target=x86_64-linux-gnu -fsanitize=alloc-token -falloc-token-max=-1 %s 2>&1 | FileCheck -check-prefix=CHECK-INVALID-MAX %s // CHECK-INVALID-MAX: error: invalid value + +// RUN: %clang --target=x86_64-linux-gnu -Xclang -falloc-token-mode=increment %s -### 2>&1 | FileCheck -check-prefix=CHECK-MODE-INCREMENT %s +// CHECK-MODE-INCREMENT: "-falloc-token-mode=increment" +// RUN: %clang --target=x86_64-linux-gnu -Xclang -falloc-token-mode=random %s -### 2>&1 | FileCheck -check-prefix=CHECK-MODE-RANDOM %s +// CHECK-MODE-RANDOM: "-falloc-token-mode=random" +// RUN: %clang --target=x86_64-linux-gnu -Xclang -falloc-token-mode=typehash %s -### 2>&1 | FileCheck -check-prefix=CHECK-MODE-TYPEHASH %s +// CHECK-MODE-TYPEHASH: "-falloc-token-mode=typehash" +// RUN: %clang --target=x86_64-linux-gnu -Xclang -falloc-token-mode=typehashpointersplit %s -### 2>&1 | FileCheck -check-prefix=CHECK-MODE-TYPEHASHPTRSPLIT %s +// CHECK-MODE-TYPEHASHPTRSPLIT: "-falloc-token-mode=typehashpointersplit" +// RUN: not %clang --target=x86_64-linux-gnu -Xclang -falloc-token-mode=asdf %s 2>&1 | FileCheck -check-prefix=CHECK-INVALID-MODE %s +// CHECK-INVALID-MODE: error: invalid value 'asdf' diff --git a/clang/test/Driver/fuchsia.c b/clang/test/Driver/fuchsia.c index cf92f85..3fb2a94 100644 --- a/clang/test/Driver/fuchsia.c +++ b/clang/test/Driver/fuchsia.c @@ -2,6 +2,10 @@ // RUN: -resource-dir=%S/Inputs/resource_dir_with_per_target_subdir \ // RUN: --sysroot=%S/platform -fuse-ld=ld 2>&1 \ // RUN: | FileCheck -check-prefixes=CHECK,CHECK-X86_64 %s +// RUN: %clang -### %s --target=arm-unknown-fuchsia \ +// RUN: -resource-dir=%S/Inputs/resource_dir_with_per_target_subdir \ +// RUN: --sysroot=%S/platform -fuse-ld=ld 2>&1 \ +// RUN: | FileCheck -check-prefixes=CHECK,CHECK-ARMV8A %s // RUN: %clang -### %s --target=aarch64-unknown-fuchsia \ // RUN: -resource-dir=%S/Inputs/resource_dir_with_per_target_subdir \ // RUN: --sysroot=%S/platform -fuse-ld=ld 2>&1 \ @@ -14,6 +18,10 @@ // RUN: -resource-dir=%S/Inputs/resource_dir_with_per_target_subdir \ // RUN: --sysroot=%S/platform -fuse-ld=ld 2>&1 \ // RUN: | FileCheck -check-prefixes=CHECK,CHECK-X86_64 %s +// RUN: %clang -### %s --target=arm-fuchsia \ +// RUN: -resource-dir=%S/Inputs/resource_dir_with_per_target_subdir \ +// RUN: --sysroot=%S/platform -fuse-ld=ld 2>&1 \ +// RUN: | FileCheck -check-prefixes=CHECK,CHECK-ARMV8A %s // RUN: %clang -### %s --target=aarch64-fuchsia \ // RUN: -resource-dir=%S/Inputs/resource_dir_with_per_target_subdir \ // RUN: --sysroot=%S/platform -fuse-ld=ld 2>&1 \ @@ -24,6 +32,7 @@ // RUN: | FileCheck -check-prefixes=CHECK,CHECK-RISCV64 %s // CHECK: "-cc1" // CHECK-X86_64: "-triple" "x86_64-unknown-fuchsia" +// CHECK-ARMV8A: "-triple" "thumbv8a-unknown-fuchsia" // CHECK-AARCH64: "-triple" "aarch64-unknown-fuchsia" // CHECK-RISCV64: "-triple" "riscv64-unknown-fuchsia" // CHECK: "-funwind-tables=2" diff --git a/clang/test/Driver/hip-toolchain-no-rdc.hip b/clang/test/Driver/hip-toolchain-no-rdc.hip index 2f40fd4..840334e 100644 --- a/clang/test/Driver/hip-toolchain-no-rdc.hip +++ b/clang/test/Driver/hip-toolchain-no-rdc.hip @@ -214,3 +214,10 @@ // AMDGCNSPIRV: {{".*clang-offload-bundler.*"}} "-type=o" // AMDGCNSPIRV-SAME: "-targets={{.*}}hipv4-spirv64-amd-amdhsa--amdgcnspirv,hipv4-amdgcn-amd-amdhsa--gfx900" // AMDGCNSPIRV-SAME: "-input=[[AMDGCNSPV_CO]]" "-input=[[GFX900_CO]]" + +// Check verbose printing with the new driver. +// RUN: %clang -### --target=x86_64-linux-gnu -fno-gpu-rdc -nogpulib -nogpuinc \ +// RUN: --offload-new-driver --offload-arch=gfx908 -v %s 2>&1 | FileCheck %s --check-prefix=VERBOSE +// VERBOSE: clang-linker-wrapper +// VERBOSE-NOT: --device-compiler=amdgcn-amd-amdhsa=-v +// VERBOSE-NOT: --wrapper-verbose diff --git a/clang/test/Driver/print-supported-extensions-riscv.c b/clang/test/Driver/print-supported-extensions-riscv.c index 7972b9e..cb81273 100644 --- a/clang/test/Driver/print-supported-extensions-riscv.c +++ b/clang/test/Driver/print-supported-extensions-riscv.c @@ -190,6 +190,11 @@ // CHECK-NEXT: xsfmm64t 0.6 'XSfmm64t' (TE=64 configuration) // CHECK-NEXT: xsfmmbase 0.6 'XSfmmbase' (All non arithmetic instructions for all TEWs and sf.vtzero) // CHECK-NEXT: xsfvcp 1.0 'XSfvcp' (SiFive Custom Vector Coprocessor Interface Instructions) +// CHECK-NEXT: xsfvfbfexp16e 0.5 'XSfvfbfexp16e' (SiFive Vector Floating-Point Exponential Function Instruction, BFloat16) +// CHECK-NEXT: xsfvfexp16e 0.5 'XSfvfexp16e' (SiFive Vector Floating-Point Exponential Function Instruction, Half Precision) +// CHECK-NEXT: xsfvfexp32e 0.5 'XSfvfexp32e' (SiFive Vector Floating-Point Exponential Function Instruction, Single Precision) +// CHECK-NEXT: xsfvfexpa 0.2 'XSfvfexpa' (SiFive Vector Floating-Point Exponential Approximation Instruction) +// CHECK-NEXT: xsfvfexpa64e 0.2 'XSfvfexpa64e' (SiFive Vector Floating-Point Exponential Approximation Instruction with Double-Precision) // CHECK-NEXT: xsfvfnrclipxfqf 1.0 'XSfvfnrclipxfqf' (SiFive FP32-to-int8 Ranged Clip Instructions) // CHECK-NEXT: xsfvfwmaccqqq 1.0 'XSfvfwmaccqqq' (SiFive Matrix Multiply Accumulate Instruction (4-by-4)) // CHECK-NEXT: xsfvqmaccdod 1.0 'XSfvqmaccdod' (SiFive Int8 Matrix Multiplication Instructions (2-by-8 and 8-by-2)) diff --git a/clang/test/Frontend/arm-ignore-branch-protection-option.c b/clang/test/Frontend/arm-ignore-branch-protection-option.c index 99a2acc..45bdb37 100644 --- a/clang/test/Frontend/arm-ignore-branch-protection-option.c +++ b/clang/test/Frontend/arm-ignore-branch-protection-option.c @@ -15,4 +15,4 @@ __attribute__((target("arch=cortex-m0"))) void f() {} // CHECK-NOT: attributes { {{.*}} "branch-target-enforcement" /// Check that there are branch protection module attributes despite the warning. -// CHECK: !{i32 8, !"branch-target-enforcement", i32 1} +// CHECK: !{i32 8, !"branch-target-enforcement", i32 2} diff --git a/clang/test/Headers/__clang_hip_math.hip b/clang/test/Headers/__clang_hip_math.hip index b88aa3c..7e26916 100644 --- a/clang/test/Headers/__clang_hip_math.hip +++ b/clang/test/Headers/__clang_hip_math.hip @@ -50,7 +50,7 @@ typedef unsigned long long uint64_t; // CHECK-LABEL: define dso_local i64 @test___make_mantissa_base8( // CHECK-SAME: ptr noundef readonly captures(none) [[P:%.*]]) local_unnamed_addr #[[ATTR2:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*]]: -// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[P]], align 1, !tbaa [[CHAR_TBAA4:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[P]], align 1, !tbaa [[CHAR_TBAA8:![0-9]+]] // CHECK-NEXT: [[CMP_NOT_I1:%.*]] = icmp eq i8 [[TMP0]], 0 // CHECK-NEXT: br i1 [[CMP_NOT_I1]], label %[[_ZL21__MAKE_MANTISSA_BASE8PKC_EXIT:.*]], label %[[WHILE_BODY_I:.*]] // CHECK: [[WHILE_BODY_I]]: @@ -66,9 +66,9 @@ typedef unsigned long long uint64_t; // CHECK-NEXT: [[ADD_I:%.*]] = add i64 [[MUL_I]], -48 // CHECK-NEXT: [[SUB_I]] = add i64 [[ADD_I]], [[CONV5_I]] // CHECK-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I2]], i64 1 -// CHECK-NEXT: [[TMP3]] = load i8, ptr [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA4]] +// CHECK-NEXT: [[TMP3]] = load i8, ptr [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA8]] // CHECK-NEXT: [[CMP_NOT_I:%.*]] = icmp eq i8 [[TMP3]], 0 -// CHECK-NEXT: br i1 [[CMP_NOT_I]], label %[[_ZL21__MAKE_MANTISSA_BASE8PKC_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP7:![0-9]+]] +// CHECK-NEXT: br i1 [[CMP_NOT_I]], label %[[_ZL21__MAKE_MANTISSA_BASE8PKC_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP9:![0-9]+]] // CHECK: [[_ZL21__MAKE_MANTISSA_BASE8PKC_EXIT]]: // CHECK-NEXT: [[RETVAL_2_I:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ 0, %[[WHILE_BODY_I]] ], [ [[SUB_I]], %[[IF_THEN_I]] ] // CHECK-NEXT: ret i64 [[RETVAL_2_I]] @@ -80,7 +80,7 @@ typedef unsigned long long uint64_t; // AMDGCNSPIRV: [[WHILE_COND_I]]: // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I:%.*]] = phi ptr addrspace(4) [ [[P]], %[[ENTRY]] ], [ [[__TAGP_ADDR_1_I:%.*]], %[[WHILE_BODY_I:.*]] ] // AMDGCNSPIRV-NEXT: [[__R_0_I:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[__R_1_I:%.*]], %[[WHILE_BODY_I]] ] -// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I]], align 1, !tbaa [[CHAR_TBAA5:![0-9]+]] +// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I]], align 1, !tbaa [[CHAR_TBAA9:![0-9]+]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I:%.*]] = icmp eq i8 [[TMP0]], 0 // AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I]], label %[[_ZL21__MAKE_MANTISSA_BASE8PKC_EXIT:.*]], label %[[WHILE_BODY_I]] // AMDGCNSPIRV: [[WHILE_BODY_I]]: @@ -93,7 +93,7 @@ typedef unsigned long long uint64_t; // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I_IDX:%.*]] = zext i1 [[OR_COND_I]] to i64 // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I]], i64 [[__TAGP_ADDR_1_I_IDX]] // AMDGCNSPIRV-NEXT: [[__R_1_I]] = select i1 [[OR_COND_I]], i64 [[SUB_I]], i64 [[__R_0_I]] -// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I]], label %[[WHILE_COND_I]], label %[[_ZL21__MAKE_MANTISSA_BASE8PKC_EXIT]], !llvm.loop [[LOOP8:![0-9]+]] +// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I]], label %[[WHILE_COND_I]], label %[[_ZL21__MAKE_MANTISSA_BASE8PKC_EXIT]], !llvm.loop [[LOOP10:![0-9]+]] // AMDGCNSPIRV: [[_ZL21__MAKE_MANTISSA_BASE8PKC_EXIT]]: // AMDGCNSPIRV-NEXT: [[RETVAL_2_I:%.*]] = phi i64 [ 0, %[[WHILE_BODY_I]] ], [ [[__R_0_I]], %[[WHILE_COND_I]] ] // AMDGCNSPIRV-NEXT: ret i64 [[RETVAL_2_I]] @@ -105,7 +105,7 @@ extern "C" __device__ uint64_t test___make_mantissa_base8(const char *p) { // CHECK-LABEL: define dso_local i64 @test___make_mantissa_base10( // CHECK-SAME: ptr noundef readonly captures(none) [[P:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*]]: -// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[P]], align 1, !tbaa [[CHAR_TBAA4]] +// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[P]], align 1, !tbaa [[CHAR_TBAA8]] // CHECK-NEXT: [[CMP_NOT_I1:%.*]] = icmp eq i8 [[TMP0]], 0 // CHECK-NEXT: br i1 [[CMP_NOT_I1]], label %[[_ZL22__MAKE_MANTISSA_BASE10PKC_EXIT:.*]], label %[[WHILE_BODY_I:.*]] // CHECK: [[WHILE_BODY_I]]: @@ -121,9 +121,9 @@ extern "C" __device__ uint64_t test___make_mantissa_base8(const char *p) { // CHECK-NEXT: [[ADD_I:%.*]] = add i64 [[MUL_I]], -48 // CHECK-NEXT: [[SUB_I]] = add i64 [[ADD_I]], [[CONV5_I]] // CHECK-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I2]], i64 1 -// CHECK-NEXT: [[TMP3]] = load i8, ptr [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA4]] +// CHECK-NEXT: [[TMP3]] = load i8, ptr [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA8]] // CHECK-NEXT: [[CMP_NOT_I:%.*]] = icmp eq i8 [[TMP3]], 0 -// CHECK-NEXT: br i1 [[CMP_NOT_I]], label %[[_ZL22__MAKE_MANTISSA_BASE10PKC_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP10:![0-9]+]] +// CHECK-NEXT: br i1 [[CMP_NOT_I]], label %[[_ZL22__MAKE_MANTISSA_BASE10PKC_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP12:![0-9]+]] // CHECK: [[_ZL22__MAKE_MANTISSA_BASE10PKC_EXIT]]: // CHECK-NEXT: [[RETVAL_2_I:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ 0, %[[WHILE_BODY_I]] ], [ [[SUB_I]], %[[IF_THEN_I]] ] // CHECK-NEXT: ret i64 [[RETVAL_2_I]] @@ -135,7 +135,7 @@ extern "C" __device__ uint64_t test___make_mantissa_base8(const char *p) { // AMDGCNSPIRV: [[WHILE_COND_I]]: // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I:%.*]] = phi ptr addrspace(4) [ [[P]], %[[ENTRY]] ], [ [[__TAGP_ADDR_1_I:%.*]], %[[WHILE_BODY_I:.*]] ] // AMDGCNSPIRV-NEXT: [[__R_0_I:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[__R_1_I:%.*]], %[[WHILE_BODY_I]] ] -// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I:%.*]] = icmp eq i8 [[TMP0]], 0 // AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I]], label %[[_ZL22__MAKE_MANTISSA_BASE10PKC_EXIT:.*]], label %[[WHILE_BODY_I]] // AMDGCNSPIRV: [[WHILE_BODY_I]]: @@ -148,7 +148,7 @@ extern "C" __device__ uint64_t test___make_mantissa_base8(const char *p) { // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I_IDX:%.*]] = zext i1 [[OR_COND_I]] to i64 // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I]], i64 [[__TAGP_ADDR_1_I_IDX]] // AMDGCNSPIRV-NEXT: [[__R_1_I]] = select i1 [[OR_COND_I]], i64 [[SUB_I]], i64 [[__R_0_I]] -// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I]], label %[[WHILE_COND_I]], label %[[_ZL22__MAKE_MANTISSA_BASE10PKC_EXIT]], !llvm.loop [[LOOP11:![0-9]+]] +// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I]], label %[[WHILE_COND_I]], label %[[_ZL22__MAKE_MANTISSA_BASE10PKC_EXIT]], !llvm.loop [[LOOP13:![0-9]+]] // AMDGCNSPIRV: [[_ZL22__MAKE_MANTISSA_BASE10PKC_EXIT]]: // AMDGCNSPIRV-NEXT: [[RETVAL_2_I:%.*]] = phi i64 [ 0, %[[WHILE_BODY_I]] ], [ [[__R_0_I]], %[[WHILE_COND_I]] ] // AMDGCNSPIRV-NEXT: ret i64 [[RETVAL_2_I]] @@ -160,7 +160,7 @@ extern "C" __device__ uint64_t test___make_mantissa_base10(const char *p) { // CHECK-LABEL: define dso_local i64 @test___make_mantissa_base16( // CHECK-SAME: ptr noundef readonly captures(none) [[P:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*]]: -// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[P]], align 1, !tbaa [[CHAR_TBAA4]] +// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[P]], align 1, !tbaa [[CHAR_TBAA8]] // CHECK-NEXT: [[CMP_NOT_I1:%.*]] = icmp eq i8 [[TMP0]], 0 // CHECK-NEXT: br i1 [[CMP_NOT_I1]], label %[[_ZL22__MAKE_MANTISSA_BASE16PKC_EXIT:.*]], label %[[WHILE_BODY_I:.*]] // CHECK: [[WHILE_BODY_I]]: @@ -185,9 +185,9 @@ extern "C" __device__ uint64_t test___make_mantissa_base10(const char *p) { // CHECK-NEXT: [[ADD26_I:%.*]] = add i64 [[MUL24_I]], [[DOTSINK]] // CHECK-NEXT: [[ADD28_I]] = add i64 [[ADD26_I]], [[CONV25_I]] // CHECK-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I2]], i64 1 -// CHECK-NEXT: [[TMP5]] = load i8, ptr [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA4]] +// CHECK-NEXT: [[TMP5]] = load i8, ptr [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA8]] // CHECK-NEXT: [[CMP_NOT_I:%.*]] = icmp eq i8 [[TMP5]], 0 -// CHECK-NEXT: br i1 [[CMP_NOT_I]], label %[[_ZL22__MAKE_MANTISSA_BASE16PKC_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP11:![0-9]+]] +// CHECK-NEXT: br i1 [[CMP_NOT_I]], label %[[_ZL22__MAKE_MANTISSA_BASE16PKC_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP13:![0-9]+]] // CHECK: [[_ZL22__MAKE_MANTISSA_BASE16PKC_EXIT]]: // CHECK-NEXT: [[RETVAL_2_I:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ 0, %[[IF_ELSE17_I]] ], [ [[ADD28_I]], %[[IF_END31_I]] ] // CHECK-NEXT: ret i64 [[RETVAL_2_I]] @@ -195,7 +195,7 @@ extern "C" __device__ uint64_t test___make_mantissa_base10(const char *p) { // AMDGCNSPIRV-LABEL: define spir_func i64 @test___make_mantissa_base16( // AMDGCNSPIRV-SAME: ptr addrspace(4) noundef readonly captures(none) [[P:%.*]]) local_unnamed_addr addrspace(4) #[[ATTR2]] { // AMDGCNSPIRV-NEXT: [[ENTRY:.*]]: -// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i8, ptr addrspace(4) [[P]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i8, ptr addrspace(4) [[P]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I1:%.*]] = icmp eq i8 [[TMP0]], 0 // AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I1]], label %[[_ZL22__MAKE_MANTISSA_BASE16PKC_EXIT:.*]], label %[[WHILE_BODY_I:.*]] // AMDGCNSPIRV: [[WHILE_BODY_I]]: @@ -220,9 +220,9 @@ extern "C" __device__ uint64_t test___make_mantissa_base10(const char *p) { // AMDGCNSPIRV-NEXT: [[ADD26_I:%.*]] = add i64 [[MUL24_I]], [[DOTSINK]] // AMDGCNSPIRV-NEXT: [[ADD28_I]] = add i64 [[ADD26_I]], [[CONV25_I]] // AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I2]], i64 1 -// AMDGCNSPIRV-NEXT: [[TMP5]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP5]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I:%.*]] = icmp eq i8 [[TMP5]], 0 -// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I]], label %[[_ZL22__MAKE_MANTISSA_BASE16PKC_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP12:![0-9]+]] +// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I]], label %[[_ZL22__MAKE_MANTISSA_BASE16PKC_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP14:![0-9]+]] // AMDGCNSPIRV: [[_ZL22__MAKE_MANTISSA_BASE16PKC_EXIT]]: // AMDGCNSPIRV-NEXT: [[RETVAL_2_I:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ 0, %[[IF_ELSE17_I]] ], [ [[ADD28_I]], %[[IF_END31_I]] ] // AMDGCNSPIRV-NEXT: ret i64 [[RETVAL_2_I]] @@ -234,26 +234,26 @@ extern "C" __device__ uint64_t test___make_mantissa_base16(const char *p) { // CHECK-LABEL: define dso_local i64 @test___make_mantissa( // CHECK-SAME: ptr noundef readonly captures(none) [[P:%.*]]) local_unnamed_addr #[[ATTR2]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[P]], align 1, !tbaa [[CHAR_TBAA4]] +// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[P]], align 1, !tbaa [[CHAR_TBAA8]] // CHECK-NEXT: [[CMP_I:%.*]] = icmp eq i8 [[TMP0]], 48 // CHECK-NEXT: br i1 [[CMP_I]], label %[[IF_THEN_I:.*]], label %[[WHILE_COND_I14_I_PREHEADER:.*]] // CHECK: [[WHILE_COND_I14_I_PREHEADER]]: -// CHECK-NEXT: [[TMP1:%.*]] = load i8, ptr [[P]], align 1, !tbaa [[CHAR_TBAA4]] +// CHECK-NEXT: [[TMP1:%.*]] = load i8, ptr [[P]], align 1, !tbaa [[CHAR_TBAA8]] // CHECK-NEXT: [[CMP_NOT_I17_I5:%.*]] = icmp eq i8 [[TMP1]], 0 // CHECK-NEXT: br i1 [[CMP_NOT_I17_I5]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT:.*]], label %[[WHILE_BODY_I18_I:.*]] // CHECK: [[IF_THEN_I]]: // CHECK-NEXT: [[INCDEC_PTR_I:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 1 -// CHECK-NEXT: [[TMP2:%.*]] = load i8, ptr [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA4]] +// CHECK-NEXT: [[TMP2:%.*]] = load i8, ptr [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA8]] // CHECK-NEXT: switch i8 [[TMP2]], label %[[WHILE_COND_I_I_PREHEADER:.*]] [ // CHECK-NEXT: i8 120, label %[[IF_THEN5_I:.*]] // CHECK-NEXT: i8 88, label %[[IF_THEN5_I]] // CHECK-NEXT: ] // CHECK: [[WHILE_COND_I_I_PREHEADER]]: -// CHECK-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA4]] +// CHECK-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA8]] // CHECK-NEXT: [[CMP_NOT_I_I14:%.*]] = icmp eq i8 [[TMP3]], 0 // CHECK-NEXT: br i1 [[CMP_NOT_I_I14]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], label %[[WHILE_BODY_I_I:.*]] // CHECK: [[IF_THEN5_I]]: -// CHECK-NEXT: [[TMP4:%.*]] = load i8, ptr [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA4]] +// CHECK-NEXT: [[TMP4:%.*]] = load i8, ptr [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA8]] // CHECK-NEXT: [[CMP_NOT_I30_I9:%.*]] = icmp eq i8 [[TMP4]], 0 // CHECK-NEXT: br i1 [[CMP_NOT_I30_I9]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], label %[[WHILE_BODY_I31_I:.*]] // CHECK: [[WHILE_BODY_I31_I]]: @@ -278,9 +278,9 @@ extern "C" __device__ uint64_t test___make_mantissa_base16(const char *p) { // CHECK-NEXT: [[ADD26_I_I:%.*]] = add i64 [[MUL24_I_I]], [[DOTSINK]] // CHECK-NEXT: [[ADD28_I_I]] = add i64 [[ADD26_I_I]], [[CONV25_I_I]] // CHECK-NEXT: [[INCDEC_PTR_I34_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I28_I10]], i64 1 -// CHECK-NEXT: [[TMP9]] = load i8, ptr [[INCDEC_PTR_I34_I]], align 1, !tbaa [[CHAR_TBAA4]] +// CHECK-NEXT: [[TMP9]] = load i8, ptr [[INCDEC_PTR_I34_I]], align 1, !tbaa [[CHAR_TBAA8]] // CHECK-NEXT: [[CMP_NOT_I30_I:%.*]] = icmp eq i8 [[TMP9]], 0 -// CHECK-NEXT: br i1 [[CMP_NOT_I30_I]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], label %[[WHILE_BODY_I31_I]], !llvm.loop [[LOOP11]] +// CHECK-NEXT: br i1 [[CMP_NOT_I30_I]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], label %[[WHILE_BODY_I31_I]], !llvm.loop [[LOOP13]] // CHECK: [[WHILE_BODY_I_I]]: // CHECK-NEXT: [[TMP10:%.*]] = phi i8 [ [[TMP12:%.*]], %[[IF_THEN_I_I:.*]] ], [ [[TMP3]], %[[WHILE_COND_I_I_PREHEADER]] ] // CHECK-NEXT: [[__R_0_I_I16:%.*]] = phi i64 [ [[SUB_I_I:%.*]], %[[IF_THEN_I_I]] ], [ 0, %[[WHILE_COND_I_I_PREHEADER]] ] @@ -294,9 +294,9 @@ extern "C" __device__ uint64_t test___make_mantissa_base16(const char *p) { // CHECK-NEXT: [[ADD_I_I:%.*]] = add i64 [[MUL_I_I]], -48 // CHECK-NEXT: [[SUB_I_I]] = add i64 [[ADD_I_I]], [[CONV5_I_I]] // CHECK-NEXT: [[INCDEC_PTR_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I_I15]], i64 1 -// CHECK-NEXT: [[TMP12]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// CHECK-NEXT: [[TMP12]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // CHECK-NEXT: [[CMP_NOT_I_I:%.*]] = icmp eq i8 [[TMP12]], 0 -// CHECK-NEXT: br i1 [[CMP_NOT_I_I]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], label %[[WHILE_BODY_I_I]], !llvm.loop [[LOOP7]] +// CHECK-NEXT: br i1 [[CMP_NOT_I_I]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], label %[[WHILE_BODY_I_I]], !llvm.loop [[LOOP9]] // CHECK: [[WHILE_BODY_I18_I]]: // CHECK-NEXT: [[TMP13:%.*]] = phi i8 [ [[TMP15:%.*]], %[[IF_THEN_I21_I:.*]] ], [ [[TMP1]], %[[WHILE_COND_I14_I_PREHEADER]] ] // CHECK-NEXT: [[__R_0_I16_I7:%.*]] = phi i64 [ [[SUB_I25_I:%.*]], %[[IF_THEN_I21_I]] ], [ 0, %[[WHILE_COND_I14_I_PREHEADER]] ] @@ -310,9 +310,9 @@ extern "C" __device__ uint64_t test___make_mantissa_base16(const char *p) { // CHECK-NEXT: [[ADD_I24_I:%.*]] = add i64 [[MUL_I22_I]], -48 // CHECK-NEXT: [[SUB_I25_I]] = add i64 [[ADD_I24_I]], [[CONV5_I23_I]] // CHECK-NEXT: [[INCDEC_PTR_I26_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I15_I6]], i64 1 -// CHECK-NEXT: [[TMP15]] = load i8, ptr [[INCDEC_PTR_I26_I]], align 1, !tbaa [[CHAR_TBAA4]] +// CHECK-NEXT: [[TMP15]] = load i8, ptr [[INCDEC_PTR_I26_I]], align 1, !tbaa [[CHAR_TBAA8]] // CHECK-NEXT: [[CMP_NOT_I17_I:%.*]] = icmp eq i8 [[TMP15]], 0 -// CHECK-NEXT: br i1 [[CMP_NOT_I17_I]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], label %[[WHILE_BODY_I18_I]], !llvm.loop [[LOOP10]] +// CHECK-NEXT: br i1 [[CMP_NOT_I17_I]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], label %[[WHILE_BODY_I18_I]], !llvm.loop [[LOOP12]] // CHECK: [[_ZL15__MAKE_MANTISSAPKC_EXIT]]: // CHECK-NEXT: [[RETVAL_0_I:%.*]] = phi i64 [ 0, %[[WHILE_COND_I_I_PREHEADER]] ], [ 0, %[[IF_THEN5_I]] ], [ 0, %[[WHILE_COND_I14_I_PREHEADER]] ], [ [[SUB_I_I]], %[[IF_THEN_I_I]] ], [ 0, %[[WHILE_BODY_I_I]] ], [ [[ADD28_I_I]], %[[IF_END31_I_I]] ], [ 0, %[[IF_ELSE17_I_I]] ], [ [[SUB_I25_I]], %[[IF_THEN_I21_I]] ], [ 0, %[[WHILE_BODY_I18_I]] ] // CHECK-NEXT: ret i64 [[RETVAL_0_I]] @@ -320,18 +320,18 @@ extern "C" __device__ uint64_t test___make_mantissa_base16(const char *p) { // AMDGCNSPIRV-LABEL: define spir_func i64 @test___make_mantissa( // AMDGCNSPIRV-SAME: ptr addrspace(4) noundef readonly captures(none) [[P:%.*]]) local_unnamed_addr addrspace(4) #[[ATTR2]] { // AMDGCNSPIRV-NEXT: [[ENTRY:.*]]: -// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i8, ptr addrspace(4) [[P]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i8, ptr addrspace(4) [[P]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_I:%.*]] = icmp eq i8 [[TMP0]], 48 // AMDGCNSPIRV-NEXT: br i1 [[CMP_I]], label %[[IF_THEN_I:.*]], label %[[WHILE_COND_I14_I:.*]] // AMDGCNSPIRV: [[IF_THEN_I]]: // AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I:%.*]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[P]], i64 1 -// AMDGCNSPIRV-NEXT: [[TMP1:%.*]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP1:%.*]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: switch i8 [[TMP1]], label %[[WHILE_COND_I_I:.*]] [ // AMDGCNSPIRV-NEXT: i8 120, label %[[IF_THEN5_I:.*]] // AMDGCNSPIRV-NEXT: i8 88, label %[[IF_THEN5_I]] // AMDGCNSPIRV-NEXT: ] // AMDGCNSPIRV: [[IF_THEN5_I]]: -// AMDGCNSPIRV-NEXT: [[TMP2:%.*]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP2:%.*]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I31_I5:%.*]] = icmp eq i8 [[TMP2]], 0 // AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I5]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT:.*]], label %[[WHILE_BODY_I32_I:.*]] // AMDGCNSPIRV: [[WHILE_BODY_I32_I]]: @@ -356,13 +356,13 @@ extern "C" __device__ uint64_t test___make_mantissa_base16(const char *p) { // AMDGCNSPIRV-NEXT: [[ADD26_I_I:%.*]] = add i64 [[MUL24_I_I]], [[DOTSINK]] // AMDGCNSPIRV-NEXT: [[ADD28_I_I]] = add i64 [[ADD26_I_I]], [[CONV25_I_I]] // AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I36_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I29_I6]], i64 1 -// AMDGCNSPIRV-NEXT: [[TMP7]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I36_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP7]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I36_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I31_I:%.*]] = icmp eq i8 [[TMP7]], 0 -// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], label %[[WHILE_BODY_I32_I]], !llvm.loop [[LOOP12]] +// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], label %[[WHILE_BODY_I32_I]], !llvm.loop [[LOOP14]] // AMDGCNSPIRV: [[WHILE_COND_I_I]]: // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I_I:%.*]], %[[WHILE_BODY_I_I:.*]] ], [ [[INCDEC_PTR_I]], %[[IF_THEN_I]] ] // AMDGCNSPIRV-NEXT: [[__R_0_I_I:%.*]] = phi i64 [ [[__R_1_I_I:%.*]], %[[WHILE_BODY_I_I]] ], [ 0, %[[IF_THEN_I]] ] -// AMDGCNSPIRV-NEXT: [[TMP8:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP8:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I_I:%.*]] = icmp eq i8 [[TMP8]], 0 // AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I_I]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], label %[[WHILE_BODY_I_I]] // AMDGCNSPIRV: [[WHILE_BODY_I_I]]: @@ -375,11 +375,11 @@ extern "C" __device__ uint64_t test___make_mantissa_base16(const char *p) { // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I_I_IDX:%.*]] = zext i1 [[OR_COND_I_I]] to i64 // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I]], i64 [[__TAGP_ADDR_1_I_I_IDX]] // AMDGCNSPIRV-NEXT: [[__R_1_I_I]] = select i1 [[OR_COND_I_I]], i64 [[SUB_I_I]], i64 [[__R_0_I_I]] -// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I_I]], label %[[WHILE_COND_I_I]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], !llvm.loop [[LOOP8]] +// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I_I]], label %[[WHILE_COND_I_I]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], !llvm.loop [[LOOP10]] // AMDGCNSPIRV: [[WHILE_COND_I14_I]]: // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I15_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I25_I:%.*]], %[[WHILE_BODY_I18_I:.*]] ], [ [[P]], %[[ENTRY]] ] // AMDGCNSPIRV-NEXT: [[__R_0_I16_I:%.*]] = phi i64 [ [[__R_1_I26_I:%.*]], %[[WHILE_BODY_I18_I]] ], [ 0, %[[ENTRY]] ] -// AMDGCNSPIRV-NEXT: [[TMP10:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP10:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I17_I:%.*]] = icmp eq i8 [[TMP10]], 0 // AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I17_I]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], label %[[WHILE_BODY_I18_I]] // AMDGCNSPIRV: [[WHILE_BODY_I18_I]]: @@ -392,7 +392,7 @@ extern "C" __device__ uint64_t test___make_mantissa_base16(const char *p) { // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I25_I_IDX:%.*]] = zext i1 [[OR_COND_I19_I]] to i64 // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I25_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I]], i64 [[__TAGP_ADDR_1_I25_I_IDX]] // AMDGCNSPIRV-NEXT: [[__R_1_I26_I]] = select i1 [[OR_COND_I19_I]], i64 [[SUB_I23_I]], i64 [[__R_0_I16_I]] -// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I19_I]], label %[[WHILE_COND_I14_I]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], !llvm.loop [[LOOP11]] +// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I19_I]], label %[[WHILE_COND_I14_I]], label %[[_ZL15__MAKE_MANTISSAPKC_EXIT]], !llvm.loop [[LOOP13]] // AMDGCNSPIRV: [[_ZL15__MAKE_MANTISSAPKC_EXIT]]: // AMDGCNSPIRV-NEXT: [[RETVAL_0_I:%.*]] = phi i64 [ 0, %[[IF_THEN5_I]] ], [ 0, %[[WHILE_BODY_I_I]] ], [ [[__R_0_I_I]], %[[WHILE_COND_I_I]] ], [ [[ADD28_I_I]], %[[IF_END31_I_I]] ], [ 0, %[[IF_ELSE17_I_I]] ], [ 0, %[[WHILE_BODY_I18_I]] ], [ [[__R_0_I16_I]], %[[WHILE_COND_I14_I]] ] // AMDGCNSPIRV-NEXT: ret i64 [[RETVAL_0_I]] @@ -452,25 +452,25 @@ extern "C" __device__ long long test_llabs(long x) { // DEFAULT-LABEL: define dso_local noundef float @test_acosf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4:[0-9]+]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acos_f32(float noundef [[X]]) #[[ATTR14:[0-9]+]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acos_f32(float noundef [[X]]) #[[ATTR12:[0-9]+]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_acosf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR4:[0-9]+]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_acos_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14:[0-9]+]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_acos_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR12:[0-9]+]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_acosf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4:[0-9]+]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acos_f32(float noundef [[X]]) #[[ATTR14:[0-9]+]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acos_f32(float noundef [[X]]) #[[ATTR12:[0-9]+]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_acosf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4:[0-9]+]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acos_f32(float noundef [[X]]) #[[ATTR14:[0-9]+]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acos_f32(float noundef [[X]]) #[[ATTR12:[0-9]+]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_acosf( @@ -486,25 +486,25 @@ extern "C" __device__ float test_acosf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_acos( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acos_f64(double noundef [[X]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acos_f64(double noundef [[X]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_acos( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_acos_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_acos_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_acos( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acos_f64(double noundef [[X]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acos_f64(double noundef [[X]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_acos( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acos_f64(double noundef [[X]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acos_f64(double noundef [[X]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_acos( @@ -520,25 +520,25 @@ extern "C" __device__ double test_acos(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_acoshf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5:[0-9]+]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acosh_f32(float noundef [[X]]) #[[ATTR15:[0-9]+]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acosh_f32(float noundef [[X]]) #[[ATTR13:[0-9]+]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_acoshf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5:[0-9]+]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_acosh_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR15:[0-9]+]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_acosh_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR13:[0-9]+]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_acoshf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5:[0-9]+]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acosh_f32(float noundef [[X]]) #[[ATTR15:[0-9]+]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acosh_f32(float noundef [[X]]) #[[ATTR13:[0-9]+]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_acoshf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5:[0-9]+]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acosh_f32(float noundef [[X]]) #[[ATTR15:[0-9]+]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acosh_f32(float noundef [[X]]) #[[ATTR13:[0-9]+]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_acoshf( @@ -554,25 +554,25 @@ extern "C" __device__ float test_acoshf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_acosh( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acosh_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acosh_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_acosh( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_acosh_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_acosh_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_acosh( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acosh_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acosh_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_acosh( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acosh_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acosh_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_acosh( @@ -588,25 +588,25 @@ extern "C" __device__ double test_acosh(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_asinf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asin_f32(float noundef [[X]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asin_f32(float noundef [[X]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_asinf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_asin_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_asin_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_asinf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asin_f32(float noundef [[X]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asin_f32(float noundef [[X]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_asinf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asin_f32(float noundef [[X]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asin_f32(float noundef [[X]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_asinf( @@ -622,25 +622,25 @@ extern "C" __device__ float test_asinf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_asin( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asin_f64(double noundef [[X]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asin_f64(double noundef [[X]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_asin( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_asin_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_asin_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_asin( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asin_f64(double noundef [[X]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asin_f64(double noundef [[X]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_asin( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asin_f64(double noundef [[X]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asin_f64(double noundef [[X]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_asin( @@ -657,25 +657,25 @@ extern "C" __device__ double test_asin(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_asinhf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asinh_f32(float noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asinh_f32(float noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_asinhf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_asinh_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_asinh_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_asinhf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asinh_f32(float noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asinh_f32(float noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_asinhf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asinh_f32(float noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asinh_f32(float noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_asinhf( @@ -691,25 +691,25 @@ extern "C" __device__ float test_asinhf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_asinh( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asinh_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asinh_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_asinh( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_asinh_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_asinh_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_asinh( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asinh_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asinh_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_asinh( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asinh_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asinh_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_asinh( @@ -725,25 +725,25 @@ extern "C" __device__ double test_asinh(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_atan2f( // DEFAULT-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan2_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan2_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_atan2f( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_atan2_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_atan2_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_atan2f( // APPROX-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan2_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan2_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_atan2f( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan2_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan2_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_atan2f( @@ -759,25 +759,25 @@ extern "C" __device__ float test_atan2f(float x, float y) { // DEFAULT-LABEL: define dso_local noundef double @test_atan2( // DEFAULT-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan2_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan2_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_atan2( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_atan2_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_atan2_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_atan2( // APPROX-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan2_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan2_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_atan2( // NCRDIV-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan2_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan2_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_atan2( @@ -793,25 +793,25 @@ extern "C" __device__ double test_atan2(double x, double y) { // DEFAULT-LABEL: define dso_local noundef float @test_atanf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan_f32(float noundef [[X]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan_f32(float noundef [[X]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_atanf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_atan_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_atan_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_atanf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan_f32(float noundef [[X]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan_f32(float noundef [[X]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_atanf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan_f32(float noundef [[X]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan_f32(float noundef [[X]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_atanf( @@ -827,25 +827,25 @@ extern "C" __device__ float test_atanf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_atan( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan_f64(double noundef [[X]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan_f64(double noundef [[X]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_atan( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_atan_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_atan_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_atan( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan_f64(double noundef [[X]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan_f64(double noundef [[X]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_atan( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan_f64(double noundef [[X]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan_f64(double noundef [[X]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_atan( @@ -861,25 +861,25 @@ extern "C" __device__ double test_atan(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_atanhf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atanh_f32(float noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atanh_f32(float noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_atanhf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_atanh_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_atanh_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_atanhf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atanh_f32(float noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atanh_f32(float noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_atanhf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atanh_f32(float noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atanh_f32(float noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_atanhf( @@ -895,25 +895,25 @@ extern "C" __device__ float test_atanhf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_atanh( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atanh_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atanh_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_atanh( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_atanh_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_atanh_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_atanh( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atanh_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atanh_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_atanh( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atanh_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atanh_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_atanh( @@ -929,25 +929,25 @@ extern "C" __device__ double test_atanh(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_cbrtf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cbrt_f32(float noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cbrt_f32(float noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_cbrtf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cbrt_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cbrt_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_cbrtf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cbrt_f32(float noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cbrt_f32(float noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_cbrtf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cbrt_f32(float noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cbrt_f32(float noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_cbrtf( @@ -963,25 +963,25 @@ extern "C" __device__ float test_cbrtf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_cbrt( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cbrt_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cbrt_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_cbrt( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cbrt_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cbrt_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_cbrt( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cbrt_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cbrt_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_cbrt( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cbrt_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cbrt_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_cbrt( @@ -1133,25 +1133,25 @@ extern "C" __device__ double test_copysign(double x, double y) { // DEFAULT-LABEL: define dso_local noundef float @test_cosf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6:[0-9]+]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cos_f32(float noundef [[X]]) #[[ATTR16:[0-9]+]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cos_f32(float noundef [[X]]) #[[ATTR14:[0-9]+]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_cosf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6:[0-9]+]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cos_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16:[0-9]+]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cos_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14:[0-9]+]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_cosf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6:[0-9]+]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I1:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16:[0-9]+]] +// APPROX-NEXT: [[CALL_I1:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14:[0-9]+]] // APPROX-NEXT: ret float [[CALL_I1]] // // NCRDIV-LABEL: define dso_local noundef float @test_cosf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6:[0-9]+]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cos_f32(float noundef [[X]]) #[[ATTR16:[0-9]+]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cos_f32(float noundef [[X]]) #[[ATTR14:[0-9]+]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_cosf( @@ -1167,25 +1167,25 @@ extern "C" __device__ float test_cosf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_cos( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cos_f64(double noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cos_f64(double noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_cos( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cos_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cos_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_cos( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cos_f64(double noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cos_f64(double noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_cos( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cos_f64(double noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cos_f64(double noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_cos( @@ -1201,25 +1201,25 @@ extern "C" __device__ double test_cos(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_coshf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cosh_f32(float noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cosh_f32(float noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_coshf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cosh_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cosh_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_coshf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cosh_f32(float noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cosh_f32(float noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_coshf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cosh_f32(float noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cosh_f32(float noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_coshf( @@ -1235,25 +1235,25 @@ extern "C" __device__ float test_coshf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_cosh( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cosh_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cosh_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_cosh( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cosh_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cosh_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_cosh( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cosh_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cosh_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_cosh( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cosh_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cosh_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_cosh( @@ -1269,25 +1269,25 @@ extern "C" __device__ double test_cosh(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_cospif( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cospi_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cospi_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_cospif( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cospi_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cospi_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_cospif( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cospi_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cospi_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_cospif( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cospi_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cospi_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_cospif( @@ -1303,25 +1303,25 @@ extern "C" __device__ float test_cospif(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_cospi( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cospi_f64(double noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cospi_f64(double noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_cospi( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cospi_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cospi_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_cospi( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cospi_f64(double noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cospi_f64(double noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_cospi( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cospi_f64(double noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cospi_f64(double noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_cospi( @@ -1337,25 +1337,25 @@ extern "C" __device__ double test_cospi(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_cyl_bessel_i0f( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i0_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i0_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_cyl_bessel_i0f( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_i0_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_i0_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_cyl_bessel_i0f( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i0_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i0_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_cyl_bessel_i0f( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i0_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i0_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_cyl_bessel_i0f( @@ -1371,25 +1371,25 @@ extern "C" __device__ float test_cyl_bessel_i0f(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_cyl_bessel_i0( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i0_f64(double noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i0_f64(double noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_cyl_bessel_i0( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_i0_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_i0_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_cyl_bessel_i0( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i0_f64(double noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i0_f64(double noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_cyl_bessel_i0( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i0_f64(double noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i0_f64(double noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_cyl_bessel_i0( @@ -1405,25 +1405,25 @@ extern "C" __device__ double test_cyl_bessel_i0(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_cyl_bessel_i1f( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i1_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i1_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_cyl_bessel_i1f( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_i1_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_i1_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_cyl_bessel_i1f( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i1_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i1_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_cyl_bessel_i1f( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i1_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i1_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_cyl_bessel_i1f( @@ -1439,25 +1439,25 @@ extern "C" __device__ float test_cyl_bessel_i1f(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_cyl_bessel_i1( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i1_f64(double noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i1_f64(double noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_cyl_bessel_i1( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_i1_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_i1_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_cyl_bessel_i1( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i1_f64(double noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i1_f64(double noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_cyl_bessel_i1( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i1_f64(double noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i1_f64(double noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_cyl_bessel_i1( @@ -1473,25 +1473,25 @@ extern "C" __device__ double test_cyl_bessel_i1(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_erfcf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfc_f32(float noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfc_f32(float noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_erfcf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_erfc_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_erfc_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_erfcf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfc_f32(float noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfc_f32(float noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_erfcf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfc_f32(float noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfc_f32(float noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_erfcf( @@ -1507,25 +1507,25 @@ extern "C" __device__ float test_erfcf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_erfc( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfc_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfc_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_erfc( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_erfc_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_erfc_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_erfc( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfc_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfc_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_erfc( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfc_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfc_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_erfc( @@ -1541,25 +1541,25 @@ extern "C" __device__ double test_erfc(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_erfinvf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfinv_f32(float noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfinv_f32(float noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_erfinvf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_erfinv_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_erfinv_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_erfinvf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfinv_f32(float noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfinv_f32(float noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_erfinvf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfinv_f32(float noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfinv_f32(float noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_erfinvf( @@ -1575,25 +1575,25 @@ extern "C" __device__ float test_erfinvf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_erfinv( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfinv_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfinv_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_erfinv( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_erfinv_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_erfinv_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_erfinv( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfinv_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfinv_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_erfinv( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfinv_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfinv_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_erfinv( @@ -1643,25 +1643,25 @@ extern "C" __device__ float test_exp10f(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_exp10( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp10_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp10_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_exp10( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_exp10_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_exp10_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_exp10( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp10_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp10_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_exp10( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp10_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp10_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_exp10( @@ -1711,25 +1711,25 @@ extern "C" __device__ float test_exp2f(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_exp2( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp2_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp2_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_exp2( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_exp2_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_exp2_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_exp2( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp2_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp2_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_exp2( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp2_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp2_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_exp2( @@ -1779,25 +1779,25 @@ extern "C" __device__ float test_expf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_exp( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_exp( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_exp_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_exp_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_exp( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_exp( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_exp( @@ -1813,25 +1813,25 @@ extern "C" __device__ double test_exp(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_expm1f( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_expm1_f32(float noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_expm1_f32(float noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_expm1f( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_expm1_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_expm1_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_expm1f( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_expm1_f32(float noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_expm1_f32(float noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_expm1f( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_expm1_f32(float noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_expm1_f32(float noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_expm1f( @@ -1847,25 +1847,25 @@ extern "C" __device__ float test_expm1f(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_expm1( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_expm1_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_expm1_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_expm1( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_expm1_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_expm1_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_expm1( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_expm1_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_expm1_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_expm1( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_expm1_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_expm1_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_expm1( @@ -1949,25 +1949,25 @@ extern "C" __device__ double test_fabs(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_fdimf( // DEFAULT-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fdim_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fdim_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_fdimf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_fdim_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_fdim_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_fdimf( // APPROX-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fdim_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fdim_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_fdimf( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fdim_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fdim_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_fdimf( @@ -1983,25 +1983,25 @@ extern "C" __device__ float test_fdimf(float x, float y) { // DEFAULT-LABEL: define dso_local noundef double @test_fdim( // DEFAULT-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fdim_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fdim_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_fdim( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_fdim_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_fdim_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_fdim( // APPROX-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fdim_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fdim_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_fdim( // NCRDIV-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fdim_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fdim_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_fdim( @@ -2035,7 +2035,7 @@ extern "C" __device__ double test_fdim(double x, double y) { // NCRDIV-LABEL: define dso_local noundef float @test_fdividef( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR3]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[DIV_I:%.*]] = fdiv contract float [[X]], [[Y]], !fpmath [[META12:![0-9]+]] +// NCRDIV-NEXT: [[DIV_I:%.*]] = fdiv contract float [[X]], [[Y]], !fpmath [[META14:![0-9]+]] // NCRDIV-NEXT: ret float [[DIV_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_fdividef( @@ -2357,25 +2357,25 @@ extern "C" __device__ double test_fmin(double x, double y) { // DEFAULT-LABEL: define dso_local noundef float @test_fmodf( // DEFAULT-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fmod_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fmod_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_fmodf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_fmod_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_fmod_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_fmodf( // APPROX-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fmod_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fmod_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_fmodf( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fmod_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fmod_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_fmodf( @@ -2391,25 +2391,25 @@ extern "C" __device__ float test_fmodf(float x, float y) { // DEFAULT-LABEL: define dso_local noundef double @test_fmod( // DEFAULT-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fmod_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fmod_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_fmod( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_fmod_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_fmod_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_fmod( // APPROX-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fmod_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fmod_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_fmod( // NCRDIV-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fmod_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fmod_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_fmod( @@ -2427,7 +2427,7 @@ extern "C" __device__ double test_fmod(double x, double y) { // DEFAULT-NEXT: [[ENTRY:.*:]] // DEFAULT-NEXT: [[TMP0:%.*]] = tail call { float, i32 } @llvm.frexp.f32.i32(float [[X]]) // DEFAULT-NEXT: [[TMP1:%.*]] = extractvalue { float, i32 } [[TMP0]], 1 -// DEFAULT-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA12:![0-9]+]] +// DEFAULT-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA4:![0-9]+]] // DEFAULT-NEXT: [[TMP2:%.*]] = extractvalue { float, i32 } [[TMP0]], 0 // DEFAULT-NEXT: ret float [[TMP2]] // @@ -2436,7 +2436,7 @@ extern "C" __device__ double test_fmod(double x, double y) { // FINITEONLY-NEXT: [[ENTRY:.*:]] // FINITEONLY-NEXT: [[TMP0:%.*]] = tail call { float, i32 } @llvm.frexp.f32.i32(float nofpclass(nan inf) [[X]]) // FINITEONLY-NEXT: [[TMP1:%.*]] = extractvalue { float, i32 } [[TMP0]], 1 -// FINITEONLY-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA12:![0-9]+]] +// FINITEONLY-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA4:![0-9]+]] // FINITEONLY-NEXT: [[TMP2:%.*]] = extractvalue { float, i32 } [[TMP0]], 0 // FINITEONLY-NEXT: ret float [[TMP2]] // @@ -2445,7 +2445,7 @@ extern "C" __device__ double test_fmod(double x, double y) { // APPROX-NEXT: [[ENTRY:.*:]] // APPROX-NEXT: [[TMP0:%.*]] = tail call { float, i32 } @llvm.frexp.f32.i32(float [[X]]) // APPROX-NEXT: [[TMP1:%.*]] = extractvalue { float, i32 } [[TMP0]], 1 -// APPROX-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA12:![0-9]+]] +// APPROX-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA4:![0-9]+]] // APPROX-NEXT: [[TMP2:%.*]] = extractvalue { float, i32 } [[TMP0]], 0 // APPROX-NEXT: ret float [[TMP2]] // @@ -2454,7 +2454,7 @@ extern "C" __device__ double test_fmod(double x, double y) { // NCRDIV-NEXT: [[ENTRY:.*:]] // NCRDIV-NEXT: [[TMP0:%.*]] = tail call { float, i32 } @llvm.frexp.f32.i32(float [[X]]) // NCRDIV-NEXT: [[TMP1:%.*]] = extractvalue { float, i32 } [[TMP0]], 1 -// NCRDIV-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA13:![0-9]+]] +// NCRDIV-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA4:![0-9]+]] // NCRDIV-NEXT: [[TMP2:%.*]] = extractvalue { float, i32 } [[TMP0]], 0 // NCRDIV-NEXT: ret float [[TMP2]] // @@ -2463,7 +2463,7 @@ extern "C" __device__ double test_fmod(double x, double y) { // AMDGCNSPIRV-NEXT: [[ENTRY:.*:]] // AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = tail call addrspace(4) { float, i32 } @llvm.frexp.f32.i32(float [[X]]) // AMDGCNSPIRV-NEXT: [[TMP1:%.*]] = extractvalue { float, i32 } [[TMP0]], 1 -// AMDGCNSPIRV-NEXT: store i32 [[TMP1]], ptr addrspace(4) [[Y]], align 4, !tbaa [[INT_TBAA13:![0-9]+]] +// AMDGCNSPIRV-NEXT: store i32 [[TMP1]], ptr addrspace(4) [[Y]], align 4, !tbaa [[INT_TBAA5:![0-9]+]] // AMDGCNSPIRV-NEXT: [[TMP2:%.*]] = extractvalue { float, i32 } [[TMP0]], 0 // AMDGCNSPIRV-NEXT: ret float [[TMP2]] // @@ -2476,7 +2476,7 @@ extern "C" __device__ float test_frexpf(float x, int* y) { // DEFAULT-NEXT: [[ENTRY:.*:]] // DEFAULT-NEXT: [[TMP0:%.*]] = tail call { double, i32 } @llvm.frexp.f64.i32(double [[X]]) // DEFAULT-NEXT: [[TMP1:%.*]] = extractvalue { double, i32 } [[TMP0]], 1 -// DEFAULT-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA12]] +// DEFAULT-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA4]] // DEFAULT-NEXT: [[TMP2:%.*]] = extractvalue { double, i32 } [[TMP0]], 0 // DEFAULT-NEXT: ret double [[TMP2]] // @@ -2485,7 +2485,7 @@ extern "C" __device__ float test_frexpf(float x, int* y) { // FINITEONLY-NEXT: [[ENTRY:.*:]] // FINITEONLY-NEXT: [[TMP0:%.*]] = tail call { double, i32 } @llvm.frexp.f64.i32(double nofpclass(nan inf) [[X]]) // FINITEONLY-NEXT: [[TMP1:%.*]] = extractvalue { double, i32 } [[TMP0]], 1 -// FINITEONLY-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA12]] +// FINITEONLY-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA4]] // FINITEONLY-NEXT: [[TMP2:%.*]] = extractvalue { double, i32 } [[TMP0]], 0 // FINITEONLY-NEXT: ret double [[TMP2]] // @@ -2494,7 +2494,7 @@ extern "C" __device__ float test_frexpf(float x, int* y) { // APPROX-NEXT: [[ENTRY:.*:]] // APPROX-NEXT: [[TMP0:%.*]] = tail call { double, i32 } @llvm.frexp.f64.i32(double [[X]]) // APPROX-NEXT: [[TMP1:%.*]] = extractvalue { double, i32 } [[TMP0]], 1 -// APPROX-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA12]] +// APPROX-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA4]] // APPROX-NEXT: [[TMP2:%.*]] = extractvalue { double, i32 } [[TMP0]], 0 // APPROX-NEXT: ret double [[TMP2]] // @@ -2503,7 +2503,7 @@ extern "C" __device__ float test_frexpf(float x, int* y) { // NCRDIV-NEXT: [[ENTRY:.*:]] // NCRDIV-NEXT: [[TMP0:%.*]] = tail call { double, i32 } @llvm.frexp.f64.i32(double [[X]]) // NCRDIV-NEXT: [[TMP1:%.*]] = extractvalue { double, i32 } [[TMP0]], 1 -// NCRDIV-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA13]] +// NCRDIV-NEXT: store i32 [[TMP1]], ptr [[Y]], align 4, !tbaa [[INT_TBAA4]] // NCRDIV-NEXT: [[TMP2:%.*]] = extractvalue { double, i32 } [[TMP0]], 0 // NCRDIV-NEXT: ret double [[TMP2]] // @@ -2512,7 +2512,7 @@ extern "C" __device__ float test_frexpf(float x, int* y) { // AMDGCNSPIRV-NEXT: [[ENTRY:.*:]] // AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = tail call addrspace(4) { double, i32 } @llvm.frexp.f64.i32(double [[X]]) // AMDGCNSPIRV-NEXT: [[TMP1:%.*]] = extractvalue { double, i32 } [[TMP0]], 1 -// AMDGCNSPIRV-NEXT: store i32 [[TMP1]], ptr addrspace(4) [[Y]], align 4, !tbaa [[INT_TBAA13]] +// AMDGCNSPIRV-NEXT: store i32 [[TMP1]], ptr addrspace(4) [[Y]], align 4, !tbaa [[INT_TBAA5]] // AMDGCNSPIRV-NEXT: [[TMP2:%.*]] = extractvalue { double, i32 } [[TMP0]], 0 // AMDGCNSPIRV-NEXT: ret double [[TMP2]] // @@ -2523,25 +2523,25 @@ extern "C" __device__ double test_frexp(double x, int* y) { // DEFAULT-LABEL: define dso_local noundef float @test_hypotf( // DEFAULT-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_hypot_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_hypot_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_hypotf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_hypot_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_hypot_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_hypotf( // APPROX-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_hypot_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_hypot_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_hypotf( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_hypot_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_hypot_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_hypotf( @@ -2557,25 +2557,25 @@ extern "C" __device__ float test_hypotf(float x, float y) { // DEFAULT-LABEL: define dso_local noundef double @test_hypot( // DEFAULT-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_hypot_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_hypot_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_hypot( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_hypot_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_hypot_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_hypot( // APPROX-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_hypot_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_hypot_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_hypot( // NCRDIV-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_hypot_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_hypot_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_hypot( @@ -2591,25 +2591,25 @@ extern "C" __device__ double test_hypot(double x, double y) { // DEFAULT-LABEL: define dso_local noundef i32 @test_ilogbf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef [[X]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef [[X]]) #[[ATTR12]] // DEFAULT-NEXT: ret i32 [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef i32 @test_ilogbf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR12]] // FINITEONLY-NEXT: ret i32 [[CALL_I]] // // APPROX-LABEL: define dso_local noundef i32 @test_ilogbf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef [[X]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef [[X]]) #[[ATTR12]] // APPROX-NEXT: ret i32 [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef i32 @test_ilogbf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef [[X]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef [[X]]) #[[ATTR12]] // NCRDIV-NEXT: ret i32 [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef i32 @test_ilogbf( @@ -2625,25 +2625,25 @@ extern "C" __device__ int test_ilogbf(float x) { // DEFAULT-LABEL: define dso_local noundef i32 @test_ilogb( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef [[X]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef [[X]]) #[[ATTR12]] // DEFAULT-NEXT: ret i32 [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef i32 @test_ilogb( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR12]] // FINITEONLY-NEXT: ret i32 [[CALL_I]] // // APPROX-LABEL: define dso_local noundef i32 @test_ilogb( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef [[X]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef [[X]]) #[[ATTR12]] // APPROX-NEXT: ret i32 [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef i32 @test_ilogb( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef [[X]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef [[X]]) #[[ATTR12]] // NCRDIV-NEXT: ret i32 [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef i32 @test_ilogb( @@ -2897,25 +2897,25 @@ extern "C" __device__ BOOL_TYPE test___isnan(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_j0f( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_j0f( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j0_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j0_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_j0f( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_j0f( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_j0f( @@ -2931,25 +2931,25 @@ extern "C" __device__ float test_j0f(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_j0( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_j0( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j0_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j0_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_j0( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_j0( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_j0( @@ -2965,25 +2965,25 @@ extern "C" __device__ double test_j0(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_j1f( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_j1f( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j1_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j1_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_j1f( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_j1f( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_j1f( @@ -2999,25 +2999,25 @@ extern "C" __device__ float test_j1f(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_j1( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_j1( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j1_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j1_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_j1( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_j1( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_j1( @@ -3038,14 +3038,14 @@ extern "C" __device__ double test_j1(double x) { // DEFAULT-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // DEFAULT-NEXT: ] // DEFAULT: [[IF_THEN_I]]: -// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR14]] // DEFAULT-NEXT: br label %[[_ZL3JNFIF_EXIT:.*]] // DEFAULT: [[IF_THEN2_I]]: -// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR14]] // DEFAULT-NEXT: br label %[[_ZL3JNFIF_EXIT]] // DEFAULT: [[IF_END4_I]]: -// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR16]] -// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR14]] // DEFAULT-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // DEFAULT-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL3JNFIF_EXIT]] // DEFAULT: [[FOR_BODY_I]]: @@ -3072,14 +3072,14 @@ extern "C" __device__ double test_j1(double x) { // FINITEONLY-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // FINITEONLY-NEXT: ] // FINITEONLY: [[IF_THEN_I]]: -// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j0_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j0_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] // FINITEONLY-NEXT: br label %[[_ZL3JNFIF_EXIT:.*]] // FINITEONLY: [[IF_THEN2_I]]: -// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] // FINITEONLY-NEXT: br label %[[_ZL3JNFIF_EXIT]] // FINITEONLY: [[IF_END4_I]]: -// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j0_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] -// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j0_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] // FINITEONLY-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // FINITEONLY-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL3JNFIF_EXIT]] // FINITEONLY: [[FOR_BODY_I]]: @@ -3106,14 +3106,14 @@ extern "C" __device__ double test_j1(double x) { // APPROX-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // APPROX-NEXT: ] // APPROX: [[IF_THEN_I]]: -// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR14]] // APPROX-NEXT: br label %[[_ZL3JNFIF_EXIT:.*]] // APPROX: [[IF_THEN2_I]]: -// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR14]] // APPROX-NEXT: br label %[[_ZL3JNFIF_EXIT]] // APPROX: [[IF_END4_I]]: -// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR16]] -// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR14]] // APPROX-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // APPROX-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL3JNFIF_EXIT]] // APPROX: [[FOR_BODY_I]]: @@ -3140,14 +3140,14 @@ extern "C" __device__ double test_j1(double x) { // NCRDIV-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // NCRDIV-NEXT: ] // NCRDIV: [[IF_THEN_I]]: -// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR14]] // NCRDIV-NEXT: br label %[[_ZL3JNFIF_EXIT:.*]] // NCRDIV: [[IF_THEN2_I]]: -// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR14]] // NCRDIV-NEXT: br label %[[_ZL3JNFIF_EXIT]] // NCRDIV: [[IF_END4_I]]: -// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR16]] -// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR14]] // NCRDIV-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // NCRDIV-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL3JNFIF_EXIT]] // NCRDIV: [[FOR_BODY_I]]: @@ -3156,7 +3156,7 @@ extern "C" __device__ double test_j1(double x) { // NCRDIV-NEXT: [[__X0_0_I2:%.*]] = phi float [ [[__X1_0_I3]], %[[FOR_BODY_I]] ], [ [[CALL_I_I]], %[[IF_END4_I]] ] // NCRDIV-NEXT: [[MUL_I:%.*]] = shl nuw nsw i32 [[__I_0_I4]], 1 // NCRDIV-NEXT: [[CONV_I:%.*]] = uitofp nneg i32 [[MUL_I]] to float -// NCRDIV-NEXT: [[DIV_I:%.*]] = fdiv contract float [[CONV_I]], [[Y]], !fpmath [[META12]] +// NCRDIV-NEXT: [[DIV_I:%.*]] = fdiv contract float [[CONV_I]], [[Y]], !fpmath [[META14]] // NCRDIV-NEXT: [[MUL8_I:%.*]] = fmul contract float [[__X1_0_I3]], [[DIV_I]] // NCRDIV-NEXT: [[SUB_I]] = fsub contract float [[MUL8_I]], [[__X0_0_I2]] // NCRDIV-NEXT: [[INC_I]] = add nuw nsw i32 [[__I_0_I4]], 1 @@ -3212,14 +3212,14 @@ extern "C" __device__ float test_jnf(int x, float y) { // DEFAULT-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // DEFAULT-NEXT: ] // DEFAULT: [[IF_THEN_I]]: -// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR14]] // DEFAULT-NEXT: br label %[[_ZL2JNID_EXIT:.*]] // DEFAULT: [[IF_THEN2_I]]: -// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR14]] // DEFAULT-NEXT: br label %[[_ZL2JNID_EXIT]] // DEFAULT: [[IF_END4_I]]: -// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR16]] -// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR14]] // DEFAULT-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // DEFAULT-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL2JNID_EXIT]] // DEFAULT: [[FOR_BODY_I]]: @@ -3246,14 +3246,14 @@ extern "C" __device__ float test_jnf(int x, float y) { // FINITEONLY-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // FINITEONLY-NEXT: ] // FINITEONLY: [[IF_THEN_I]]: -// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j0_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j0_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] // FINITEONLY-NEXT: br label %[[_ZL2JNID_EXIT:.*]] // FINITEONLY: [[IF_THEN2_I]]: -// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] // FINITEONLY-NEXT: br label %[[_ZL2JNID_EXIT]] // FINITEONLY: [[IF_END4_I]]: -// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j0_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] -// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j0_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] // FINITEONLY-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // FINITEONLY-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL2JNID_EXIT]] // FINITEONLY: [[FOR_BODY_I]]: @@ -3280,14 +3280,14 @@ extern "C" __device__ float test_jnf(int x, float y) { // APPROX-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // APPROX-NEXT: ] // APPROX: [[IF_THEN_I]]: -// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR14]] // APPROX-NEXT: br label %[[_ZL2JNID_EXIT:.*]] // APPROX: [[IF_THEN2_I]]: -// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR14]] // APPROX-NEXT: br label %[[_ZL2JNID_EXIT]] // APPROX: [[IF_END4_I]]: -// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR16]] -// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR14]] // APPROX-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // APPROX-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL2JNID_EXIT]] // APPROX: [[FOR_BODY_I]]: @@ -3314,14 +3314,14 @@ extern "C" __device__ float test_jnf(int x, float y) { // NCRDIV-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // NCRDIV-NEXT: ] // NCRDIV: [[IF_THEN_I]]: -// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR14]] // NCRDIV-NEXT: br label %[[_ZL2JNID_EXIT:.*]] // NCRDIV: [[IF_THEN2_I]]: -// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR14]] // NCRDIV-NEXT: br label %[[_ZL2JNID_EXIT]] // NCRDIV: [[IF_END4_I]]: -// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR16]] -// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR14]] // NCRDIV-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // NCRDIV-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL2JNID_EXIT]] // NCRDIV: [[FOR_BODY_I]]: @@ -3449,25 +3449,25 @@ extern "C" __device__ double test_ldexp(double x, int y) { // DEFAULT-LABEL: define dso_local noundef float @test_lgammaf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_lgamma_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_lgamma_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_lgammaf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_lgamma_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_lgamma_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_lgammaf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_lgamma_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_lgamma_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_lgammaf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_lgamma_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_lgamma_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_lgammaf( @@ -3483,25 +3483,25 @@ extern "C" __device__ float test_lgammaf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_lgamma( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_lgamma_f64(double noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_lgamma_f64(double noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_lgamma( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_lgamma_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_lgamma_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_lgamma( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_lgamma_f64(double noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_lgamma_f64(double noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_lgamma( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_lgamma_f64(double noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_lgamma_f64(double noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_lgamma( @@ -3707,25 +3707,25 @@ extern "C" __device__ float test_log10f(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_log10( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log10_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log10_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_log10( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_log10_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_log10_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_log10( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log10_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log10_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_log10( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log10_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log10_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_log10( @@ -3741,25 +3741,25 @@ extern "C" __device__ double test_log10(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_log1pf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_log1p_f32(float noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_log1p_f32(float noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_log1pf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_log1p_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_log1p_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_log1pf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_log1p_f32(float noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_log1p_f32(float noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_log1pf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_log1p_f32(float noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_log1p_f32(float noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_log1pf( @@ -3775,25 +3775,25 @@ extern "C" __device__ float test_log1pf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_log1p( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log1p_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log1p_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_log1p( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_log1p_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_log1p_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_log1p( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log1p_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log1p_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_log1p( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log1p_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log1p_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_log1p( @@ -3843,25 +3843,25 @@ extern "C" __device__ float test_log2f(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_log2( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log2_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log2_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_log2( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_log2_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_log2_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_log2( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log2_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log2_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_log2( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log2_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log2_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_log2( @@ -3877,25 +3877,25 @@ extern "C" __device__ double test_log2(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_logbf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_logb_f32(float noundef [[X]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_logb_f32(float noundef [[X]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_logbf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_logb_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_logb_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_logbf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_logb_f32(float noundef [[X]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_logb_f32(float noundef [[X]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_logbf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_logb_f32(float noundef [[X]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_logb_f32(float noundef [[X]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_logbf( @@ -3911,25 +3911,25 @@ extern "C" __device__ float test_logbf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_logb( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_logb_f64(double noundef [[X]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_logb_f64(double noundef [[X]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_logb( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_logb_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_logb_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_logb( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_logb_f64(double noundef [[X]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_logb_f64(double noundef [[X]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_logb( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_logb_f64(double noundef [[X]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_logb_f64(double noundef [[X]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_logb( @@ -4136,44 +4136,44 @@ extern "C" __device__ long int test_lround(double x) { // DEFAULT-SAME: float noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] // DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5) -// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17:[0-9]+]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_modf_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15:[0-9]+]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_modf_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // DEFAULT-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[FLOAT_TBAA16:![0-9]+]] // DEFAULT-NEXT: store float [[TMP0]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA16]] -// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_modff( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] // FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5) -// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17:[0-9]+]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_modf_f32(float noundef nofpclass(nan inf) [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15:[0-9]+]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_modf_f32(float noundef nofpclass(nan inf) [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // FINITEONLY-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[FLOAT_TBAA16:![0-9]+]] // FINITEONLY-NEXT: store float [[TMP0]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA16]] -// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_modff( // APPROX-SAME: float noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] // APPROX-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5) -// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17:[0-9]+]] -// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_modf_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15:[0-9]+]] +// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_modf_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // APPROX-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[FLOAT_TBAA16:![0-9]+]] // APPROX-NEXT: store float [[TMP0]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA16]] -// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_modff( // NCRDIV-SAME: float noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] // NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5) -// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17:[0-9]+]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_modf_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15:[0-9]+]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_modf_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // NCRDIV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[FLOAT_TBAA17:![0-9]+]] // NCRDIV-NEXT: store float [[TMP0]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA17]] -// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_modff( @@ -4196,44 +4196,44 @@ extern "C" __device__ float test_modff(float x, float* y) { // DEFAULT-SAME: double noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Y:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] // DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5) -// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_modf_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_modf_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // DEFAULT-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[DOUBLE_TBAA18:![0-9]+]] // DEFAULT-NEXT: store double [[TMP0]], ptr [[Y]], align 8, !tbaa [[DOUBLE_TBAA18]] -// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_modf( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Y:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] // FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5) -// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_modf_f64(double noundef nofpclass(nan inf) [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_modf_f64(double noundef nofpclass(nan inf) [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // FINITEONLY-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[DOUBLE_TBAA18:![0-9]+]] // FINITEONLY-NEXT: store double [[TMP0]], ptr [[Y]], align 8, !tbaa [[DOUBLE_TBAA18]] -// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_modf( // APPROX-SAME: double noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Y:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] // APPROX-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5) -// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_modf_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_modf_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // APPROX-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[DOUBLE_TBAA18:![0-9]+]] // APPROX-NEXT: store double [[TMP0]], ptr [[Y]], align 8, !tbaa [[DOUBLE_TBAA18]] -// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_modf( // NCRDIV-SAME: double noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Y:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] // NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5) -// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_modf_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_modf_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // NCRDIV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[DOUBLE_TBAA19:![0-9]+]] // NCRDIV-NEXT: store double [[TMP0]], ptr [[Y]], align 8, !tbaa [[DOUBLE_TBAA19]] -// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_modf( @@ -4255,26 +4255,26 @@ extern "C" __device__ double test_modf(double x, double* y) { // DEFAULT-LABEL: define dso_local float @test_nanf( // DEFAULT-SAME: ptr noundef readonly captures(none) [[TAG:%.*]]) local_unnamed_addr #[[ATTR2]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[TMP0:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP0:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: [[CMP_I_I:%.*]] = icmp eq i8 [[TMP0]], 48 // DEFAULT-NEXT: br i1 [[CMP_I_I]], label %[[IF_THEN_I_I:.*]], label %[[WHILE_COND_I14_I_I_PREHEADER:.*]] // DEFAULT: [[WHILE_COND_I14_I_I_PREHEADER]]: -// DEFAULT-NEXT: [[TMP1:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP1:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: [[CMP_NOT_I17_I_I5:%.*]] = icmp eq i8 [[TMP1]], 0 // DEFAULT-NEXT: br i1 [[CMP_NOT_I17_I_I5]], label %[[_ZL4NANFPKC_EXIT:.*]], label %[[WHILE_BODY_I18_I_I:.*]] // DEFAULT: [[IF_THEN_I_I]]: // DEFAULT-NEXT: [[INCDEC_PTR_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[TAG]], i64 1 -// DEFAULT-NEXT: [[TMP2:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP2:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: switch i8 [[TMP2]], label %[[WHILE_COND_I_I_I_PREHEADER:.*]] [ // DEFAULT-NEXT: i8 120, label %[[IF_THEN5_I_I:.*]] // DEFAULT-NEXT: i8 88, label %[[IF_THEN5_I_I]] // DEFAULT-NEXT: ] // DEFAULT: [[WHILE_COND_I_I_I_PREHEADER]]: -// DEFAULT-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: [[CMP_NOT_I_I_I14:%.*]] = icmp eq i8 [[TMP3]], 0 // DEFAULT-NEXT: br i1 [[CMP_NOT_I_I_I14]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I_I_I:.*]] // DEFAULT: [[IF_THEN5_I_I]]: -// DEFAULT-NEXT: [[TMP4:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP4:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: [[CMP_NOT_I30_I_I9:%.*]] = icmp eq i8 [[TMP4]], 0 // DEFAULT-NEXT: br i1 [[CMP_NOT_I30_I_I9]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I31_I_I:.*]] // DEFAULT: [[WHILE_BODY_I31_I_I]]: @@ -4299,9 +4299,9 @@ extern "C" __device__ double test_modf(double x, double* y) { // DEFAULT-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]] // DEFAULT-NEXT: [[ADD28_I_I_I]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]] // DEFAULT-NEXT: [[INCDEC_PTR_I34_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I28_I_I10]], i64 1 -// DEFAULT-NEXT: [[TMP9]] = load i8, ptr [[INCDEC_PTR_I34_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP9]] = load i8, ptr [[INCDEC_PTR_I34_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: [[CMP_NOT_I30_I_I:%.*]] = icmp eq i8 [[TMP9]], 0 -// DEFAULT-NEXT: br i1 [[CMP_NOT_I30_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I31_I_I]], !llvm.loop [[LOOP11]] +// DEFAULT-NEXT: br i1 [[CMP_NOT_I30_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I31_I_I]], !llvm.loop [[LOOP13]] // DEFAULT: [[WHILE_BODY_I_I_I]]: // DEFAULT-NEXT: [[TMP10:%.*]] = phi i8 [ [[TMP12:%.*]], %[[IF_THEN_I_I_I:.*]] ], [ [[TMP3]], %[[WHILE_COND_I_I_I_PREHEADER]] ] // DEFAULT-NEXT: [[__R_0_I_I_I16:%.*]] = phi i64 [ [[SUB_I_I_I:%.*]], %[[IF_THEN_I_I_I]] ], [ 0, %[[WHILE_COND_I_I_I_PREHEADER]] ] @@ -4315,9 +4315,9 @@ extern "C" __device__ double test_modf(double x, double* y) { // DEFAULT-NEXT: [[ADD_I_I_I:%.*]] = add i64 [[MUL_I_I_I]], -48 // DEFAULT-NEXT: [[SUB_I_I_I]] = add i64 [[ADD_I_I_I]], [[CONV5_I_I_I]] // DEFAULT-NEXT: [[INCDEC_PTR_I_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I_I_I15]], i64 1 -// DEFAULT-NEXT: [[TMP12]] = load i8, ptr [[INCDEC_PTR_I_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP12]] = load i8, ptr [[INCDEC_PTR_I_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP12]], 0 -// DEFAULT-NEXT: br i1 [[CMP_NOT_I_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I_I_I]], !llvm.loop [[LOOP7]] +// DEFAULT-NEXT: br i1 [[CMP_NOT_I_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I_I_I]], !llvm.loop [[LOOP9]] // DEFAULT: [[WHILE_BODY_I18_I_I]]: // DEFAULT-NEXT: [[TMP13:%.*]] = phi i8 [ [[TMP15:%.*]], %[[IF_THEN_I21_I_I:.*]] ], [ [[TMP1]], %[[WHILE_COND_I14_I_I_PREHEADER]] ] // DEFAULT-NEXT: [[__R_0_I16_I_I7:%.*]] = phi i64 [ [[SUB_I25_I_I:%.*]], %[[IF_THEN_I21_I_I]] ], [ 0, %[[WHILE_COND_I14_I_I_PREHEADER]] ] @@ -4331,9 +4331,9 @@ extern "C" __device__ double test_modf(double x, double* y) { // DEFAULT-NEXT: [[ADD_I24_I_I:%.*]] = add i64 [[MUL_I22_I_I]], -48 // DEFAULT-NEXT: [[SUB_I25_I_I]] = add i64 [[ADD_I24_I_I]], [[CONV5_I23_I_I]] // DEFAULT-NEXT: [[INCDEC_PTR_I26_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I15_I_I6]], i64 1 -// DEFAULT-NEXT: [[TMP15]] = load i8, ptr [[INCDEC_PTR_I26_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP15]] = load i8, ptr [[INCDEC_PTR_I26_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP15]], 0 -// DEFAULT-NEXT: br i1 [[CMP_NOT_I17_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I18_I_I]], !llvm.loop [[LOOP10]] +// DEFAULT-NEXT: br i1 [[CMP_NOT_I17_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I18_I_I]], !llvm.loop [[LOOP12]] // DEFAULT: [[_ZL4NANFPKC_EXIT]]: // DEFAULT-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, %[[WHILE_COND_I_I_I_PREHEADER]] ], [ 0, %[[IF_THEN5_I_I]] ], [ 0, %[[WHILE_COND_I14_I_I_PREHEADER]] ], [ [[SUB_I_I_I]], %[[IF_THEN_I_I_I]] ], [ 0, %[[WHILE_BODY_I_I_I]] ], [ [[ADD28_I_I_I]], %[[IF_END31_I_I_I]] ], [ 0, %[[IF_ELSE17_I_I_I]] ], [ [[SUB_I25_I_I]], %[[IF_THEN_I21_I_I]] ], [ 0, %[[WHILE_BODY_I18_I_I]] ] // DEFAULT-NEXT: [[CONV_I:%.*]] = trunc i64 [[RETVAL_0_I_I]] to i32 @@ -4350,26 +4350,26 @@ extern "C" __device__ double test_modf(double x, double* y) { // APPROX-LABEL: define dso_local float @test_nanf( // APPROX-SAME: ptr noundef readonly captures(none) [[TAG:%.*]]) local_unnamed_addr #[[ATTR2]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[TMP0:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP0:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: [[CMP_I_I:%.*]] = icmp eq i8 [[TMP0]], 48 // APPROX-NEXT: br i1 [[CMP_I_I]], label %[[IF_THEN_I_I:.*]], label %[[WHILE_COND_I14_I_I_PREHEADER:.*]] // APPROX: [[WHILE_COND_I14_I_I_PREHEADER]]: -// APPROX-NEXT: [[TMP1:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP1:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: [[CMP_NOT_I17_I_I5:%.*]] = icmp eq i8 [[TMP1]], 0 // APPROX-NEXT: br i1 [[CMP_NOT_I17_I_I5]], label %[[_ZL4NANFPKC_EXIT:.*]], label %[[WHILE_BODY_I18_I_I:.*]] // APPROX: [[IF_THEN_I_I]]: // APPROX-NEXT: [[INCDEC_PTR_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[TAG]], i64 1 -// APPROX-NEXT: [[TMP2:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP2:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: switch i8 [[TMP2]], label %[[WHILE_COND_I_I_I_PREHEADER:.*]] [ // APPROX-NEXT: i8 120, label %[[IF_THEN5_I_I:.*]] // APPROX-NEXT: i8 88, label %[[IF_THEN5_I_I]] // APPROX-NEXT: ] // APPROX: [[WHILE_COND_I_I_I_PREHEADER]]: -// APPROX-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: [[CMP_NOT_I_I_I14:%.*]] = icmp eq i8 [[TMP3]], 0 // APPROX-NEXT: br i1 [[CMP_NOT_I_I_I14]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I_I_I:.*]] // APPROX: [[IF_THEN5_I_I]]: -// APPROX-NEXT: [[TMP4:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP4:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: [[CMP_NOT_I30_I_I9:%.*]] = icmp eq i8 [[TMP4]], 0 // APPROX-NEXT: br i1 [[CMP_NOT_I30_I_I9]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I31_I_I:.*]] // APPROX: [[WHILE_BODY_I31_I_I]]: @@ -4394,9 +4394,9 @@ extern "C" __device__ double test_modf(double x, double* y) { // APPROX-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]] // APPROX-NEXT: [[ADD28_I_I_I]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]] // APPROX-NEXT: [[INCDEC_PTR_I34_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I28_I_I10]], i64 1 -// APPROX-NEXT: [[TMP9]] = load i8, ptr [[INCDEC_PTR_I34_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP9]] = load i8, ptr [[INCDEC_PTR_I34_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: [[CMP_NOT_I30_I_I:%.*]] = icmp eq i8 [[TMP9]], 0 -// APPROX-NEXT: br i1 [[CMP_NOT_I30_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I31_I_I]], !llvm.loop [[LOOP11]] +// APPROX-NEXT: br i1 [[CMP_NOT_I30_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I31_I_I]], !llvm.loop [[LOOP13]] // APPROX: [[WHILE_BODY_I_I_I]]: // APPROX-NEXT: [[TMP10:%.*]] = phi i8 [ [[TMP12:%.*]], %[[IF_THEN_I_I_I:.*]] ], [ [[TMP3]], %[[WHILE_COND_I_I_I_PREHEADER]] ] // APPROX-NEXT: [[__R_0_I_I_I16:%.*]] = phi i64 [ [[SUB_I_I_I:%.*]], %[[IF_THEN_I_I_I]] ], [ 0, %[[WHILE_COND_I_I_I_PREHEADER]] ] @@ -4410,9 +4410,9 @@ extern "C" __device__ double test_modf(double x, double* y) { // APPROX-NEXT: [[ADD_I_I_I:%.*]] = add i64 [[MUL_I_I_I]], -48 // APPROX-NEXT: [[SUB_I_I_I]] = add i64 [[ADD_I_I_I]], [[CONV5_I_I_I]] // APPROX-NEXT: [[INCDEC_PTR_I_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I_I_I15]], i64 1 -// APPROX-NEXT: [[TMP12]] = load i8, ptr [[INCDEC_PTR_I_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP12]] = load i8, ptr [[INCDEC_PTR_I_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP12]], 0 -// APPROX-NEXT: br i1 [[CMP_NOT_I_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I_I_I]], !llvm.loop [[LOOP7]] +// APPROX-NEXT: br i1 [[CMP_NOT_I_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I_I_I]], !llvm.loop [[LOOP9]] // APPROX: [[WHILE_BODY_I18_I_I]]: // APPROX-NEXT: [[TMP13:%.*]] = phi i8 [ [[TMP15:%.*]], %[[IF_THEN_I21_I_I:.*]] ], [ [[TMP1]], %[[WHILE_COND_I14_I_I_PREHEADER]] ] // APPROX-NEXT: [[__R_0_I16_I_I7:%.*]] = phi i64 [ [[SUB_I25_I_I:%.*]], %[[IF_THEN_I21_I_I]] ], [ 0, %[[WHILE_COND_I14_I_I_PREHEADER]] ] @@ -4426,9 +4426,9 @@ extern "C" __device__ double test_modf(double x, double* y) { // APPROX-NEXT: [[ADD_I24_I_I:%.*]] = add i64 [[MUL_I22_I_I]], -48 // APPROX-NEXT: [[SUB_I25_I_I]] = add i64 [[ADD_I24_I_I]], [[CONV5_I23_I_I]] // APPROX-NEXT: [[INCDEC_PTR_I26_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I15_I_I6]], i64 1 -// APPROX-NEXT: [[TMP15]] = load i8, ptr [[INCDEC_PTR_I26_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP15]] = load i8, ptr [[INCDEC_PTR_I26_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP15]], 0 -// APPROX-NEXT: br i1 [[CMP_NOT_I17_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I18_I_I]], !llvm.loop [[LOOP10]] +// APPROX-NEXT: br i1 [[CMP_NOT_I17_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I18_I_I]], !llvm.loop [[LOOP12]] // APPROX: [[_ZL4NANFPKC_EXIT]]: // APPROX-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, %[[WHILE_COND_I_I_I_PREHEADER]] ], [ 0, %[[IF_THEN5_I_I]] ], [ 0, %[[WHILE_COND_I14_I_I_PREHEADER]] ], [ [[SUB_I_I_I]], %[[IF_THEN_I_I_I]] ], [ 0, %[[WHILE_BODY_I_I_I]] ], [ [[ADD28_I_I_I]], %[[IF_END31_I_I_I]] ], [ 0, %[[IF_ELSE17_I_I_I]] ], [ [[SUB_I25_I_I]], %[[IF_THEN_I21_I_I]] ], [ 0, %[[WHILE_BODY_I18_I_I]] ] // APPROX-NEXT: [[CONV_I:%.*]] = trunc i64 [[RETVAL_0_I_I]] to i32 @@ -4440,26 +4440,26 @@ extern "C" __device__ double test_modf(double x, double* y) { // NCRDIV-LABEL: define dso_local float @test_nanf( // NCRDIV-SAME: ptr noundef readonly captures(none) [[TAG:%.*]]) local_unnamed_addr #[[ATTR2]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[TMP0:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP0:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: [[CMP_I_I:%.*]] = icmp eq i8 [[TMP0]], 48 // NCRDIV-NEXT: br i1 [[CMP_I_I]], label %[[IF_THEN_I_I:.*]], label %[[WHILE_COND_I14_I_I_PREHEADER:.*]] // NCRDIV: [[WHILE_COND_I14_I_I_PREHEADER]]: -// NCRDIV-NEXT: [[TMP1:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP1:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: [[CMP_NOT_I17_I_I5:%.*]] = icmp eq i8 [[TMP1]], 0 // NCRDIV-NEXT: br i1 [[CMP_NOT_I17_I_I5]], label %[[_ZL4NANFPKC_EXIT:.*]], label %[[WHILE_BODY_I18_I_I:.*]] // NCRDIV: [[IF_THEN_I_I]]: // NCRDIV-NEXT: [[INCDEC_PTR_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[TAG]], i64 1 -// NCRDIV-NEXT: [[TMP2:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP2:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: switch i8 [[TMP2]], label %[[WHILE_COND_I_I_I_PREHEADER:.*]] [ // NCRDIV-NEXT: i8 120, label %[[IF_THEN5_I_I:.*]] // NCRDIV-NEXT: i8 88, label %[[IF_THEN5_I_I]] // NCRDIV-NEXT: ] // NCRDIV: [[WHILE_COND_I_I_I_PREHEADER]]: -// NCRDIV-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: [[CMP_NOT_I_I_I14:%.*]] = icmp eq i8 [[TMP3]], 0 // NCRDIV-NEXT: br i1 [[CMP_NOT_I_I_I14]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I_I_I:.*]] // NCRDIV: [[IF_THEN5_I_I]]: -// NCRDIV-NEXT: [[TMP4:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP4:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: [[CMP_NOT_I30_I_I9:%.*]] = icmp eq i8 [[TMP4]], 0 // NCRDIV-NEXT: br i1 [[CMP_NOT_I30_I_I9]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I31_I_I:.*]] // NCRDIV: [[WHILE_BODY_I31_I_I]]: @@ -4484,9 +4484,9 @@ extern "C" __device__ double test_modf(double x, double* y) { // NCRDIV-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]] // NCRDIV-NEXT: [[ADD28_I_I_I]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]] // NCRDIV-NEXT: [[INCDEC_PTR_I34_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I28_I_I10]], i64 1 -// NCRDIV-NEXT: [[TMP9]] = load i8, ptr [[INCDEC_PTR_I34_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP9]] = load i8, ptr [[INCDEC_PTR_I34_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: [[CMP_NOT_I30_I_I:%.*]] = icmp eq i8 [[TMP9]], 0 -// NCRDIV-NEXT: br i1 [[CMP_NOT_I30_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I31_I_I]], !llvm.loop [[LOOP11]] +// NCRDIV-NEXT: br i1 [[CMP_NOT_I30_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I31_I_I]], !llvm.loop [[LOOP13]] // NCRDIV: [[WHILE_BODY_I_I_I]]: // NCRDIV-NEXT: [[TMP10:%.*]] = phi i8 [ [[TMP12:%.*]], %[[IF_THEN_I_I_I:.*]] ], [ [[TMP3]], %[[WHILE_COND_I_I_I_PREHEADER]] ] // NCRDIV-NEXT: [[__R_0_I_I_I16:%.*]] = phi i64 [ [[SUB_I_I_I:%.*]], %[[IF_THEN_I_I_I]] ], [ 0, %[[WHILE_COND_I_I_I_PREHEADER]] ] @@ -4500,9 +4500,9 @@ extern "C" __device__ double test_modf(double x, double* y) { // NCRDIV-NEXT: [[ADD_I_I_I:%.*]] = add i64 [[MUL_I_I_I]], -48 // NCRDIV-NEXT: [[SUB_I_I_I]] = add i64 [[ADD_I_I_I]], [[CONV5_I_I_I]] // NCRDIV-NEXT: [[INCDEC_PTR_I_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I_I_I15]], i64 1 -// NCRDIV-NEXT: [[TMP12]] = load i8, ptr [[INCDEC_PTR_I_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP12]] = load i8, ptr [[INCDEC_PTR_I_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP12]], 0 -// NCRDIV-NEXT: br i1 [[CMP_NOT_I_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I_I_I]], !llvm.loop [[LOOP7]] +// NCRDIV-NEXT: br i1 [[CMP_NOT_I_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I_I_I]], !llvm.loop [[LOOP9]] // NCRDIV: [[WHILE_BODY_I18_I_I]]: // NCRDIV-NEXT: [[TMP13:%.*]] = phi i8 [ [[TMP15:%.*]], %[[IF_THEN_I21_I_I:.*]] ], [ [[TMP1]], %[[WHILE_COND_I14_I_I_PREHEADER]] ] // NCRDIV-NEXT: [[__R_0_I16_I_I7:%.*]] = phi i64 [ [[SUB_I25_I_I:%.*]], %[[IF_THEN_I21_I_I]] ], [ 0, %[[WHILE_COND_I14_I_I_PREHEADER]] ] @@ -4516,9 +4516,9 @@ extern "C" __device__ double test_modf(double x, double* y) { // NCRDIV-NEXT: [[ADD_I24_I_I:%.*]] = add i64 [[MUL_I22_I_I]], -48 // NCRDIV-NEXT: [[SUB_I25_I_I]] = add i64 [[ADD_I24_I_I]], [[CONV5_I23_I_I]] // NCRDIV-NEXT: [[INCDEC_PTR_I26_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I15_I_I6]], i64 1 -// NCRDIV-NEXT: [[TMP15]] = load i8, ptr [[INCDEC_PTR_I26_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP15]] = load i8, ptr [[INCDEC_PTR_I26_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP15]], 0 -// NCRDIV-NEXT: br i1 [[CMP_NOT_I17_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I18_I_I]], !llvm.loop [[LOOP10]] +// NCRDIV-NEXT: br i1 [[CMP_NOT_I17_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I18_I_I]], !llvm.loop [[LOOP12]] // NCRDIV: [[_ZL4NANFPKC_EXIT]]: // NCRDIV-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, %[[WHILE_COND_I_I_I_PREHEADER]] ], [ 0, %[[IF_THEN5_I_I]] ], [ 0, %[[WHILE_COND_I14_I_I_PREHEADER]] ], [ [[SUB_I_I_I]], %[[IF_THEN_I_I_I]] ], [ 0, %[[WHILE_BODY_I_I_I]] ], [ [[ADD28_I_I_I]], %[[IF_END31_I_I_I]] ], [ 0, %[[IF_ELSE17_I_I_I]] ], [ [[SUB_I25_I_I]], %[[IF_THEN_I21_I_I]] ], [ 0, %[[WHILE_BODY_I18_I_I]] ] // NCRDIV-NEXT: [[CONV_I:%.*]] = trunc i64 [[RETVAL_0_I_I]] to i32 @@ -4530,18 +4530,18 @@ extern "C" __device__ double test_modf(double x, double* y) { // AMDGCNSPIRV-LABEL: define spir_func float @test_nanf( // AMDGCNSPIRV-SAME: ptr addrspace(4) noundef readonly captures(none) [[TAG:%.*]]) local_unnamed_addr addrspace(4) #[[ATTR2]] { // AMDGCNSPIRV-NEXT: [[ENTRY:.*]]: -// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i8, ptr addrspace(4) [[TAG]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i8, ptr addrspace(4) [[TAG]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_I_I:%.*]] = icmp eq i8 [[TMP0]], 48 // AMDGCNSPIRV-NEXT: br i1 [[CMP_I_I]], label %[[IF_THEN_I_I:.*]], label %[[WHILE_COND_I14_I_I:.*]] // AMDGCNSPIRV: [[IF_THEN_I_I]]: // AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I_I:%.*]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[TAG]], i64 1 -// AMDGCNSPIRV-NEXT: [[TMP1:%.*]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP1:%.*]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: switch i8 [[TMP1]], label %[[WHILE_COND_I_I_I:.*]] [ // AMDGCNSPIRV-NEXT: i8 120, label %[[IF_THEN5_I_I:.*]] // AMDGCNSPIRV-NEXT: i8 88, label %[[IF_THEN5_I_I]] // AMDGCNSPIRV-NEXT: ] // AMDGCNSPIRV: [[IF_THEN5_I_I]]: -// AMDGCNSPIRV-NEXT: [[TMP2:%.*]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP2:%.*]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I31_I_I5:%.*]] = icmp eq i8 [[TMP2]], 0 // AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I_I5]], label %[[_ZL4NANFPKC_EXIT:.*]], label %[[WHILE_BODY_I32_I_I:.*]] // AMDGCNSPIRV: [[WHILE_BODY_I32_I_I]]: @@ -4566,13 +4566,13 @@ extern "C" __device__ double test_modf(double x, double* y) { // AMDGCNSPIRV-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]] // AMDGCNSPIRV-NEXT: [[ADD28_I_I_I]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]] // AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I36_I_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I29_I_I6]], i64 1 -// AMDGCNSPIRV-NEXT: [[TMP7]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I36_I_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP7]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I36_I_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I31_I_I:%.*]] = icmp eq i8 [[TMP7]], 0 -// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I32_I_I]], !llvm.loop [[LOOP12]] +// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I32_I_I]], !llvm.loop [[LOOP14]] // AMDGCNSPIRV: [[WHILE_COND_I_I_I]]: // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I_I_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I_I_I:%.*]], %[[WHILE_BODY_I_I_I:.*]] ], [ [[INCDEC_PTR_I_I]], %[[IF_THEN_I_I]] ] // AMDGCNSPIRV-NEXT: [[__R_0_I_I_I:%.*]] = phi i64 [ [[__R_1_I_I_I:%.*]], %[[WHILE_BODY_I_I_I]] ], [ 0, %[[IF_THEN_I_I]] ] -// AMDGCNSPIRV-NEXT: [[TMP8:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP8:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP8]], 0 // AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I_I_I]] // AMDGCNSPIRV: [[WHILE_BODY_I_I_I]]: @@ -4585,11 +4585,11 @@ extern "C" __device__ double test_modf(double x, double* y) { // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I_I_I_IDX:%.*]] = zext i1 [[OR_COND_I_I_I]] to i64 // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I_I_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I_I]], i64 [[__TAGP_ADDR_1_I_I_I_IDX]] // AMDGCNSPIRV-NEXT: [[__R_1_I_I_I]] = select i1 [[OR_COND_I_I_I]], i64 [[SUB_I_I_I]], i64 [[__R_0_I_I_I]] -// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I_I_I]], label %[[WHILE_COND_I_I_I]], label %[[_ZL4NANFPKC_EXIT]], !llvm.loop [[LOOP8]] +// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I_I_I]], label %[[WHILE_COND_I_I_I]], label %[[_ZL4NANFPKC_EXIT]], !llvm.loop [[LOOP10]] // AMDGCNSPIRV: [[WHILE_COND_I14_I_I]]: // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I15_I_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I25_I_I:%.*]], %[[WHILE_BODY_I18_I_I:.*]] ], [ [[TAG]], %[[ENTRY]] ] // AMDGCNSPIRV-NEXT: [[__R_0_I16_I_I:%.*]] = phi i64 [ [[__R_1_I26_I_I:%.*]], %[[WHILE_BODY_I18_I_I]] ], [ 0, %[[ENTRY]] ] -// AMDGCNSPIRV-NEXT: [[TMP10:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP10:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP10]], 0 // AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I17_I_I]], label %[[_ZL4NANFPKC_EXIT]], label %[[WHILE_BODY_I18_I_I]] // AMDGCNSPIRV: [[WHILE_BODY_I18_I_I]]: @@ -4602,7 +4602,7 @@ extern "C" __device__ double test_modf(double x, double* y) { // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I25_I_I_IDX:%.*]] = zext i1 [[OR_COND_I19_I_I]] to i64 // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I25_I_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I_I]], i64 [[__TAGP_ADDR_1_I25_I_I_IDX]] // AMDGCNSPIRV-NEXT: [[__R_1_I26_I_I]] = select i1 [[OR_COND_I19_I_I]], i64 [[SUB_I23_I_I]], i64 [[__R_0_I16_I_I]] -// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I19_I_I]], label %[[WHILE_COND_I14_I_I]], label %[[_ZL4NANFPKC_EXIT]], !llvm.loop [[LOOP11]] +// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I19_I_I]], label %[[WHILE_COND_I14_I_I]], label %[[_ZL4NANFPKC_EXIT]], !llvm.loop [[LOOP13]] // AMDGCNSPIRV: [[_ZL4NANFPKC_EXIT]]: // AMDGCNSPIRV-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, %[[IF_THEN5_I_I]] ], [ 0, %[[WHILE_BODY_I_I_I]] ], [ [[__R_0_I_I_I]], %[[WHILE_COND_I_I_I]] ], [ [[ADD28_I_I_I]], %[[IF_END31_I_I_I]] ], [ 0, %[[IF_ELSE17_I_I_I]] ], [ 0, %[[WHILE_BODY_I18_I_I]] ], [ [[__R_0_I16_I_I]], %[[WHILE_COND_I14_I_I]] ] // AMDGCNSPIRV-NEXT: [[CONV_I:%.*]] = trunc i64 [[RETVAL_0_I_I]] to i32 @@ -4618,26 +4618,26 @@ extern "C" __device__ float test_nanf(const char *tag) { // DEFAULT-LABEL: define dso_local double @test_nan( // DEFAULT-SAME: ptr noundef readonly captures(none) [[TAG:%.*]]) local_unnamed_addr #[[ATTR2]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[TMP0:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP0:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: [[CMP_I_I:%.*]] = icmp eq i8 [[TMP0]], 48 // DEFAULT-NEXT: br i1 [[CMP_I_I]], label %[[IF_THEN_I_I:.*]], label %[[WHILE_COND_I14_I_I_PREHEADER:.*]] // DEFAULT: [[WHILE_COND_I14_I_I_PREHEADER]]: -// DEFAULT-NEXT: [[TMP1:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP1:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: [[CMP_NOT_I17_I_I5:%.*]] = icmp eq i8 [[TMP1]], 0 // DEFAULT-NEXT: br i1 [[CMP_NOT_I17_I_I5]], label %[[_ZL3NANPKC_EXIT:.*]], label %[[WHILE_BODY_I18_I_I:.*]] // DEFAULT: [[IF_THEN_I_I]]: // DEFAULT-NEXT: [[INCDEC_PTR_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[TAG]], i64 1 -// DEFAULT-NEXT: [[TMP2:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP2:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: switch i8 [[TMP2]], label %[[WHILE_COND_I_I_I_PREHEADER:.*]] [ // DEFAULT-NEXT: i8 120, label %[[IF_THEN5_I_I:.*]] // DEFAULT-NEXT: i8 88, label %[[IF_THEN5_I_I]] // DEFAULT-NEXT: ] // DEFAULT: [[WHILE_COND_I_I_I_PREHEADER]]: -// DEFAULT-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: [[CMP_NOT_I_I_I14:%.*]] = icmp eq i8 [[TMP3]], 0 // DEFAULT-NEXT: br i1 [[CMP_NOT_I_I_I14]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I_I_I:.*]] // DEFAULT: [[IF_THEN5_I_I]]: -// DEFAULT-NEXT: [[TMP4:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP4:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: [[CMP_NOT_I30_I_I9:%.*]] = icmp eq i8 [[TMP4]], 0 // DEFAULT-NEXT: br i1 [[CMP_NOT_I30_I_I9]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I31_I_I:.*]] // DEFAULT: [[WHILE_BODY_I31_I_I]]: @@ -4662,9 +4662,9 @@ extern "C" __device__ float test_nanf(const char *tag) { // DEFAULT-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]] // DEFAULT-NEXT: [[ADD28_I_I_I]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]] // DEFAULT-NEXT: [[INCDEC_PTR_I34_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I28_I_I10]], i64 1 -// DEFAULT-NEXT: [[TMP9]] = load i8, ptr [[INCDEC_PTR_I34_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP9]] = load i8, ptr [[INCDEC_PTR_I34_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: [[CMP_NOT_I30_I_I:%.*]] = icmp eq i8 [[TMP9]], 0 -// DEFAULT-NEXT: br i1 [[CMP_NOT_I30_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I31_I_I]], !llvm.loop [[LOOP11]] +// DEFAULT-NEXT: br i1 [[CMP_NOT_I30_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I31_I_I]], !llvm.loop [[LOOP13]] // DEFAULT: [[WHILE_BODY_I_I_I]]: // DEFAULT-NEXT: [[TMP10:%.*]] = phi i8 [ [[TMP12:%.*]], %[[IF_THEN_I_I_I:.*]] ], [ [[TMP3]], %[[WHILE_COND_I_I_I_PREHEADER]] ] // DEFAULT-NEXT: [[__R_0_I_I_I16:%.*]] = phi i64 [ [[SUB_I_I_I:%.*]], %[[IF_THEN_I_I_I]] ], [ 0, %[[WHILE_COND_I_I_I_PREHEADER]] ] @@ -4678,9 +4678,9 @@ extern "C" __device__ float test_nanf(const char *tag) { // DEFAULT-NEXT: [[ADD_I_I_I:%.*]] = add i64 [[MUL_I_I_I]], -48 // DEFAULT-NEXT: [[SUB_I_I_I]] = add i64 [[ADD_I_I_I]], [[CONV5_I_I_I]] // DEFAULT-NEXT: [[INCDEC_PTR_I_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I_I_I15]], i64 1 -// DEFAULT-NEXT: [[TMP12]] = load i8, ptr [[INCDEC_PTR_I_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP12]] = load i8, ptr [[INCDEC_PTR_I_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP12]], 0 -// DEFAULT-NEXT: br i1 [[CMP_NOT_I_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I_I_I]], !llvm.loop [[LOOP7]] +// DEFAULT-NEXT: br i1 [[CMP_NOT_I_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I_I_I]], !llvm.loop [[LOOP9]] // DEFAULT: [[WHILE_BODY_I18_I_I]]: // DEFAULT-NEXT: [[TMP13:%.*]] = phi i8 [ [[TMP15:%.*]], %[[IF_THEN_I21_I_I:.*]] ], [ [[TMP1]], %[[WHILE_COND_I14_I_I_PREHEADER]] ] // DEFAULT-NEXT: [[__R_0_I16_I_I7:%.*]] = phi i64 [ [[SUB_I25_I_I:%.*]], %[[IF_THEN_I21_I_I]] ], [ 0, %[[WHILE_COND_I14_I_I_PREHEADER]] ] @@ -4694,9 +4694,9 @@ extern "C" __device__ float test_nanf(const char *tag) { // DEFAULT-NEXT: [[ADD_I24_I_I:%.*]] = add i64 [[MUL_I22_I_I]], -48 // DEFAULT-NEXT: [[SUB_I25_I_I]] = add i64 [[ADD_I24_I_I]], [[CONV5_I23_I_I]] // DEFAULT-NEXT: [[INCDEC_PTR_I26_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I15_I_I6]], i64 1 -// DEFAULT-NEXT: [[TMP15]] = load i8, ptr [[INCDEC_PTR_I26_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// DEFAULT-NEXT: [[TMP15]] = load i8, ptr [[INCDEC_PTR_I26_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // DEFAULT-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP15]], 0 -// DEFAULT-NEXT: br i1 [[CMP_NOT_I17_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I18_I_I]], !llvm.loop [[LOOP10]] +// DEFAULT-NEXT: br i1 [[CMP_NOT_I17_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I18_I_I]], !llvm.loop [[LOOP12]] // DEFAULT: [[_ZL3NANPKC_EXIT]]: // DEFAULT-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, %[[WHILE_COND_I_I_I_PREHEADER]] ], [ 0, %[[IF_THEN5_I_I]] ], [ 0, %[[WHILE_COND_I14_I_I_PREHEADER]] ], [ [[SUB_I_I_I]], %[[IF_THEN_I_I_I]] ], [ 0, %[[WHILE_BODY_I_I_I]] ], [ [[ADD28_I_I_I]], %[[IF_END31_I_I_I]] ], [ 0, %[[IF_ELSE17_I_I_I]] ], [ [[SUB_I25_I_I]], %[[IF_THEN_I21_I_I]] ], [ 0, %[[WHILE_BODY_I18_I_I]] ] // DEFAULT-NEXT: [[BF_VALUE_I:%.*]] = and i64 [[RETVAL_0_I_I]], 2251799813685247 @@ -4712,26 +4712,26 @@ extern "C" __device__ float test_nanf(const char *tag) { // APPROX-LABEL: define dso_local double @test_nan( // APPROX-SAME: ptr noundef readonly captures(none) [[TAG:%.*]]) local_unnamed_addr #[[ATTR2]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[TMP0:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP0:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: [[CMP_I_I:%.*]] = icmp eq i8 [[TMP0]], 48 // APPROX-NEXT: br i1 [[CMP_I_I]], label %[[IF_THEN_I_I:.*]], label %[[WHILE_COND_I14_I_I_PREHEADER:.*]] // APPROX: [[WHILE_COND_I14_I_I_PREHEADER]]: -// APPROX-NEXT: [[TMP1:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP1:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: [[CMP_NOT_I17_I_I5:%.*]] = icmp eq i8 [[TMP1]], 0 // APPROX-NEXT: br i1 [[CMP_NOT_I17_I_I5]], label %[[_ZL3NANPKC_EXIT:.*]], label %[[WHILE_BODY_I18_I_I:.*]] // APPROX: [[IF_THEN_I_I]]: // APPROX-NEXT: [[INCDEC_PTR_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[TAG]], i64 1 -// APPROX-NEXT: [[TMP2:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP2:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: switch i8 [[TMP2]], label %[[WHILE_COND_I_I_I_PREHEADER:.*]] [ // APPROX-NEXT: i8 120, label %[[IF_THEN5_I_I:.*]] // APPROX-NEXT: i8 88, label %[[IF_THEN5_I_I]] // APPROX-NEXT: ] // APPROX: [[WHILE_COND_I_I_I_PREHEADER]]: -// APPROX-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: [[CMP_NOT_I_I_I14:%.*]] = icmp eq i8 [[TMP3]], 0 // APPROX-NEXT: br i1 [[CMP_NOT_I_I_I14]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I_I_I:.*]] // APPROX: [[IF_THEN5_I_I]]: -// APPROX-NEXT: [[TMP4:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP4:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: [[CMP_NOT_I30_I_I9:%.*]] = icmp eq i8 [[TMP4]], 0 // APPROX-NEXT: br i1 [[CMP_NOT_I30_I_I9]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I31_I_I:.*]] // APPROX: [[WHILE_BODY_I31_I_I]]: @@ -4756,9 +4756,9 @@ extern "C" __device__ float test_nanf(const char *tag) { // APPROX-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]] // APPROX-NEXT: [[ADD28_I_I_I]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]] // APPROX-NEXT: [[INCDEC_PTR_I34_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I28_I_I10]], i64 1 -// APPROX-NEXT: [[TMP9]] = load i8, ptr [[INCDEC_PTR_I34_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP9]] = load i8, ptr [[INCDEC_PTR_I34_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: [[CMP_NOT_I30_I_I:%.*]] = icmp eq i8 [[TMP9]], 0 -// APPROX-NEXT: br i1 [[CMP_NOT_I30_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I31_I_I]], !llvm.loop [[LOOP11]] +// APPROX-NEXT: br i1 [[CMP_NOT_I30_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I31_I_I]], !llvm.loop [[LOOP13]] // APPROX: [[WHILE_BODY_I_I_I]]: // APPROX-NEXT: [[TMP10:%.*]] = phi i8 [ [[TMP12:%.*]], %[[IF_THEN_I_I_I:.*]] ], [ [[TMP3]], %[[WHILE_COND_I_I_I_PREHEADER]] ] // APPROX-NEXT: [[__R_0_I_I_I16:%.*]] = phi i64 [ [[SUB_I_I_I:%.*]], %[[IF_THEN_I_I_I]] ], [ 0, %[[WHILE_COND_I_I_I_PREHEADER]] ] @@ -4772,9 +4772,9 @@ extern "C" __device__ float test_nanf(const char *tag) { // APPROX-NEXT: [[ADD_I_I_I:%.*]] = add i64 [[MUL_I_I_I]], -48 // APPROX-NEXT: [[SUB_I_I_I]] = add i64 [[ADD_I_I_I]], [[CONV5_I_I_I]] // APPROX-NEXT: [[INCDEC_PTR_I_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I_I_I15]], i64 1 -// APPROX-NEXT: [[TMP12]] = load i8, ptr [[INCDEC_PTR_I_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP12]] = load i8, ptr [[INCDEC_PTR_I_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP12]], 0 -// APPROX-NEXT: br i1 [[CMP_NOT_I_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I_I_I]], !llvm.loop [[LOOP7]] +// APPROX-NEXT: br i1 [[CMP_NOT_I_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I_I_I]], !llvm.loop [[LOOP9]] // APPROX: [[WHILE_BODY_I18_I_I]]: // APPROX-NEXT: [[TMP13:%.*]] = phi i8 [ [[TMP15:%.*]], %[[IF_THEN_I21_I_I:.*]] ], [ [[TMP1]], %[[WHILE_COND_I14_I_I_PREHEADER]] ] // APPROX-NEXT: [[__R_0_I16_I_I7:%.*]] = phi i64 [ [[SUB_I25_I_I:%.*]], %[[IF_THEN_I21_I_I]] ], [ 0, %[[WHILE_COND_I14_I_I_PREHEADER]] ] @@ -4788,9 +4788,9 @@ extern "C" __device__ float test_nanf(const char *tag) { // APPROX-NEXT: [[ADD_I24_I_I:%.*]] = add i64 [[MUL_I22_I_I]], -48 // APPROX-NEXT: [[SUB_I25_I_I]] = add i64 [[ADD_I24_I_I]], [[CONV5_I23_I_I]] // APPROX-NEXT: [[INCDEC_PTR_I26_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I15_I_I6]], i64 1 -// APPROX-NEXT: [[TMP15]] = load i8, ptr [[INCDEC_PTR_I26_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// APPROX-NEXT: [[TMP15]] = load i8, ptr [[INCDEC_PTR_I26_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // APPROX-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP15]], 0 -// APPROX-NEXT: br i1 [[CMP_NOT_I17_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I18_I_I]], !llvm.loop [[LOOP10]] +// APPROX-NEXT: br i1 [[CMP_NOT_I17_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I18_I_I]], !llvm.loop [[LOOP12]] // APPROX: [[_ZL3NANPKC_EXIT]]: // APPROX-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, %[[WHILE_COND_I_I_I_PREHEADER]] ], [ 0, %[[IF_THEN5_I_I]] ], [ 0, %[[WHILE_COND_I14_I_I_PREHEADER]] ], [ [[SUB_I_I_I]], %[[IF_THEN_I_I_I]] ], [ 0, %[[WHILE_BODY_I_I_I]] ], [ [[ADD28_I_I_I]], %[[IF_END31_I_I_I]] ], [ 0, %[[IF_ELSE17_I_I_I]] ], [ [[SUB_I25_I_I]], %[[IF_THEN_I21_I_I]] ], [ 0, %[[WHILE_BODY_I18_I_I]] ] // APPROX-NEXT: [[BF_VALUE_I:%.*]] = and i64 [[RETVAL_0_I_I]], 2251799813685247 @@ -4801,26 +4801,26 @@ extern "C" __device__ float test_nanf(const char *tag) { // NCRDIV-LABEL: define dso_local double @test_nan( // NCRDIV-SAME: ptr noundef readonly captures(none) [[TAG:%.*]]) local_unnamed_addr #[[ATTR2]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[TMP0:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP0:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: [[CMP_I_I:%.*]] = icmp eq i8 [[TMP0]], 48 // NCRDIV-NEXT: br i1 [[CMP_I_I]], label %[[IF_THEN_I_I:.*]], label %[[WHILE_COND_I14_I_I_PREHEADER:.*]] // NCRDIV: [[WHILE_COND_I14_I_I_PREHEADER]]: -// NCRDIV-NEXT: [[TMP1:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP1:%.*]] = load i8, ptr [[TAG]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: [[CMP_NOT_I17_I_I5:%.*]] = icmp eq i8 [[TMP1]], 0 // NCRDIV-NEXT: br i1 [[CMP_NOT_I17_I_I5]], label %[[_ZL3NANPKC_EXIT:.*]], label %[[WHILE_BODY_I18_I_I:.*]] // NCRDIV: [[IF_THEN_I_I]]: // NCRDIV-NEXT: [[INCDEC_PTR_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[TAG]], i64 1 -// NCRDIV-NEXT: [[TMP2:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP2:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: switch i8 [[TMP2]], label %[[WHILE_COND_I_I_I_PREHEADER:.*]] [ // NCRDIV-NEXT: i8 120, label %[[IF_THEN5_I_I:.*]] // NCRDIV-NEXT: i8 88, label %[[IF_THEN5_I_I]] // NCRDIV-NEXT: ] // NCRDIV: [[WHILE_COND_I_I_I_PREHEADER]]: -// NCRDIV-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: [[CMP_NOT_I_I_I14:%.*]] = icmp eq i8 [[TMP3]], 0 // NCRDIV-NEXT: br i1 [[CMP_NOT_I_I_I14]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I_I_I:.*]] // NCRDIV: [[IF_THEN5_I_I]]: -// NCRDIV-NEXT: [[TMP4:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP4:%.*]] = load i8, ptr [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: [[CMP_NOT_I30_I_I9:%.*]] = icmp eq i8 [[TMP4]], 0 // NCRDIV-NEXT: br i1 [[CMP_NOT_I30_I_I9]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I31_I_I:.*]] // NCRDIV: [[WHILE_BODY_I31_I_I]]: @@ -4845,9 +4845,9 @@ extern "C" __device__ float test_nanf(const char *tag) { // NCRDIV-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]] // NCRDIV-NEXT: [[ADD28_I_I_I]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]] // NCRDIV-NEXT: [[INCDEC_PTR_I34_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I28_I_I10]], i64 1 -// NCRDIV-NEXT: [[TMP9]] = load i8, ptr [[INCDEC_PTR_I34_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP9]] = load i8, ptr [[INCDEC_PTR_I34_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: [[CMP_NOT_I30_I_I:%.*]] = icmp eq i8 [[TMP9]], 0 -// NCRDIV-NEXT: br i1 [[CMP_NOT_I30_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I31_I_I]], !llvm.loop [[LOOP11]] +// NCRDIV-NEXT: br i1 [[CMP_NOT_I30_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I31_I_I]], !llvm.loop [[LOOP13]] // NCRDIV: [[WHILE_BODY_I_I_I]]: // NCRDIV-NEXT: [[TMP10:%.*]] = phi i8 [ [[TMP12:%.*]], %[[IF_THEN_I_I_I:.*]] ], [ [[TMP3]], %[[WHILE_COND_I_I_I_PREHEADER]] ] // NCRDIV-NEXT: [[__R_0_I_I_I16:%.*]] = phi i64 [ [[SUB_I_I_I:%.*]], %[[IF_THEN_I_I_I]] ], [ 0, %[[WHILE_COND_I_I_I_PREHEADER]] ] @@ -4861,9 +4861,9 @@ extern "C" __device__ float test_nanf(const char *tag) { // NCRDIV-NEXT: [[ADD_I_I_I:%.*]] = add i64 [[MUL_I_I_I]], -48 // NCRDIV-NEXT: [[SUB_I_I_I]] = add i64 [[ADD_I_I_I]], [[CONV5_I_I_I]] // NCRDIV-NEXT: [[INCDEC_PTR_I_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I_I_I15]], i64 1 -// NCRDIV-NEXT: [[TMP12]] = load i8, ptr [[INCDEC_PTR_I_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP12]] = load i8, ptr [[INCDEC_PTR_I_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP12]], 0 -// NCRDIV-NEXT: br i1 [[CMP_NOT_I_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I_I_I]], !llvm.loop [[LOOP7]] +// NCRDIV-NEXT: br i1 [[CMP_NOT_I_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I_I_I]], !llvm.loop [[LOOP9]] // NCRDIV: [[WHILE_BODY_I18_I_I]]: // NCRDIV-NEXT: [[TMP13:%.*]] = phi i8 [ [[TMP15:%.*]], %[[IF_THEN_I21_I_I:.*]] ], [ [[TMP1]], %[[WHILE_COND_I14_I_I_PREHEADER]] ] // NCRDIV-NEXT: [[__R_0_I16_I_I7:%.*]] = phi i64 [ [[SUB_I25_I_I:%.*]], %[[IF_THEN_I21_I_I]] ], [ 0, %[[WHILE_COND_I14_I_I_PREHEADER]] ] @@ -4877,9 +4877,9 @@ extern "C" __device__ float test_nanf(const char *tag) { // NCRDIV-NEXT: [[ADD_I24_I_I:%.*]] = add i64 [[MUL_I22_I_I]], -48 // NCRDIV-NEXT: [[SUB_I25_I_I]] = add i64 [[ADD_I24_I_I]], [[CONV5_I23_I_I]] // NCRDIV-NEXT: [[INCDEC_PTR_I26_I_I]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I15_I_I6]], i64 1 -// NCRDIV-NEXT: [[TMP15]] = load i8, ptr [[INCDEC_PTR_I26_I_I]], align 1, !tbaa [[CHAR_TBAA4]] +// NCRDIV-NEXT: [[TMP15]] = load i8, ptr [[INCDEC_PTR_I26_I_I]], align 1, !tbaa [[CHAR_TBAA8]] // NCRDIV-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP15]], 0 -// NCRDIV-NEXT: br i1 [[CMP_NOT_I17_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I18_I_I]], !llvm.loop [[LOOP10]] +// NCRDIV-NEXT: br i1 [[CMP_NOT_I17_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I18_I_I]], !llvm.loop [[LOOP12]] // NCRDIV: [[_ZL3NANPKC_EXIT]]: // NCRDIV-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, %[[WHILE_COND_I_I_I_PREHEADER]] ], [ 0, %[[IF_THEN5_I_I]] ], [ 0, %[[WHILE_COND_I14_I_I_PREHEADER]] ], [ [[SUB_I_I_I]], %[[IF_THEN_I_I_I]] ], [ 0, %[[WHILE_BODY_I_I_I]] ], [ [[ADD28_I_I_I]], %[[IF_END31_I_I_I]] ], [ 0, %[[IF_ELSE17_I_I_I]] ], [ [[SUB_I25_I_I]], %[[IF_THEN_I21_I_I]] ], [ 0, %[[WHILE_BODY_I18_I_I]] ] // NCRDIV-NEXT: [[BF_VALUE_I:%.*]] = and i64 [[RETVAL_0_I_I]], 2251799813685247 @@ -4890,18 +4890,18 @@ extern "C" __device__ float test_nanf(const char *tag) { // AMDGCNSPIRV-LABEL: define spir_func double @test_nan( // AMDGCNSPIRV-SAME: ptr addrspace(4) noundef readonly captures(none) [[TAG:%.*]]) local_unnamed_addr addrspace(4) #[[ATTR2]] { // AMDGCNSPIRV-NEXT: [[ENTRY:.*]]: -// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i8, ptr addrspace(4) [[TAG]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i8, ptr addrspace(4) [[TAG]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_I_I:%.*]] = icmp eq i8 [[TMP0]], 48 // AMDGCNSPIRV-NEXT: br i1 [[CMP_I_I]], label %[[IF_THEN_I_I:.*]], label %[[WHILE_COND_I14_I_I:.*]] // AMDGCNSPIRV: [[IF_THEN_I_I]]: // AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I_I:%.*]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[TAG]], i64 1 -// AMDGCNSPIRV-NEXT: [[TMP1:%.*]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP1:%.*]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: switch i8 [[TMP1]], label %[[WHILE_COND_I_I_I:.*]] [ // AMDGCNSPIRV-NEXT: i8 120, label %[[IF_THEN5_I_I:.*]] // AMDGCNSPIRV-NEXT: i8 88, label %[[IF_THEN5_I_I]] // AMDGCNSPIRV-NEXT: ] // AMDGCNSPIRV: [[IF_THEN5_I_I]]: -// AMDGCNSPIRV-NEXT: [[TMP2:%.*]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP2:%.*]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I31_I_I5:%.*]] = icmp eq i8 [[TMP2]], 0 // AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I_I5]], label %[[_ZL3NANPKC_EXIT:.*]], label %[[WHILE_BODY_I32_I_I:.*]] // AMDGCNSPIRV: [[WHILE_BODY_I32_I_I]]: @@ -4926,13 +4926,13 @@ extern "C" __device__ float test_nanf(const char *tag) { // AMDGCNSPIRV-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]] // AMDGCNSPIRV-NEXT: [[ADD28_I_I_I]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]] // AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I36_I_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I29_I_I6]], i64 1 -// AMDGCNSPIRV-NEXT: [[TMP7]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I36_I_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP7]] = load i8, ptr addrspace(4) [[INCDEC_PTR_I36_I_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I31_I_I:%.*]] = icmp eq i8 [[TMP7]], 0 -// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I32_I_I]], !llvm.loop [[LOOP12]] +// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I32_I_I]], !llvm.loop [[LOOP14]] // AMDGCNSPIRV: [[WHILE_COND_I_I_I]]: // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I_I_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I_I_I:%.*]], %[[WHILE_BODY_I_I_I:.*]] ], [ [[INCDEC_PTR_I_I]], %[[IF_THEN_I_I]] ] // AMDGCNSPIRV-NEXT: [[__R_0_I_I_I:%.*]] = phi i64 [ [[__R_1_I_I_I:%.*]], %[[WHILE_BODY_I_I_I]] ], [ 0, %[[IF_THEN_I_I]] ] -// AMDGCNSPIRV-NEXT: [[TMP8:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP8:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP8]], 0 // AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I_I_I]] // AMDGCNSPIRV: [[WHILE_BODY_I_I_I]]: @@ -4945,11 +4945,11 @@ extern "C" __device__ float test_nanf(const char *tag) { // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I_I_I_IDX:%.*]] = zext i1 [[OR_COND_I_I_I]] to i64 // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I_I_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I_I]], i64 [[__TAGP_ADDR_1_I_I_I_IDX]] // AMDGCNSPIRV-NEXT: [[__R_1_I_I_I]] = select i1 [[OR_COND_I_I_I]], i64 [[SUB_I_I_I]], i64 [[__R_0_I_I_I]] -// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I_I_I]], label %[[WHILE_COND_I_I_I]], label %[[_ZL3NANPKC_EXIT]], !llvm.loop [[LOOP8]] +// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I_I_I]], label %[[WHILE_COND_I_I_I]], label %[[_ZL3NANPKC_EXIT]], !llvm.loop [[LOOP10]] // AMDGCNSPIRV: [[WHILE_COND_I14_I_I]]: // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I15_I_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I25_I_I:%.*]], %[[WHILE_BODY_I18_I_I:.*]] ], [ [[TAG]], %[[ENTRY]] ] // AMDGCNSPIRV-NEXT: [[__R_0_I16_I_I:%.*]] = phi i64 [ [[__R_1_I26_I_I:%.*]], %[[WHILE_BODY_I18_I_I]] ], [ 0, %[[ENTRY]] ] -// AMDGCNSPIRV-NEXT: [[TMP10:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[CHAR_TBAA5]] +// AMDGCNSPIRV-NEXT: [[TMP10:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[CHAR_TBAA9]] // AMDGCNSPIRV-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP10]], 0 // AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I17_I_I]], label %[[_ZL3NANPKC_EXIT]], label %[[WHILE_BODY_I18_I_I]] // AMDGCNSPIRV: [[WHILE_BODY_I18_I_I]]: @@ -4962,7 +4962,7 @@ extern "C" __device__ float test_nanf(const char *tag) { // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I25_I_I_IDX:%.*]] = zext i1 [[OR_COND_I19_I_I]] to i64 // AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I25_I_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I_I]], i64 [[__TAGP_ADDR_1_I25_I_I_IDX]] // AMDGCNSPIRV-NEXT: [[__R_1_I26_I_I]] = select i1 [[OR_COND_I19_I_I]], i64 [[SUB_I23_I_I]], i64 [[__R_0_I16_I_I]] -// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I19_I_I]], label %[[WHILE_COND_I14_I_I]], label %[[_ZL3NANPKC_EXIT]], !llvm.loop [[LOOP11]] +// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I19_I_I]], label %[[WHILE_COND_I14_I_I]], label %[[_ZL3NANPKC_EXIT]], !llvm.loop [[LOOP13]] // AMDGCNSPIRV: [[_ZL3NANPKC_EXIT]]: // AMDGCNSPIRV-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, %[[IF_THEN5_I_I]] ], [ 0, %[[WHILE_BODY_I_I_I]] ], [ [[__R_0_I_I_I]], %[[WHILE_COND_I_I_I]] ], [ [[ADD28_I_I_I]], %[[IF_END31_I_I_I]] ], [ 0, %[[IF_ELSE17_I_I_I]] ], [ 0, %[[WHILE_BODY_I18_I_I]] ], [ [[__R_0_I16_I_I]], %[[WHILE_COND_I14_I_I]] ] // AMDGCNSPIRV-NEXT: [[BF_VALUE_I:%.*]] = and i64 [[RETVAL_0_I_I]], 2251799813685247 @@ -5161,25 +5161,25 @@ extern "C" __device__ double test_nearbyint(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_nextafterf( // DEFAULT-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_nextafter_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_nextafter_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_nextafterf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_nextafter_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_nextafter_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_nextafterf( // APPROX-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_nextafter_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_nextafter_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_nextafterf( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_nextafter_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_nextafter_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_nextafterf( @@ -5195,25 +5195,25 @@ extern "C" __device__ float test_nextafterf(float x, float y) { // DEFAULT-LABEL: define dso_local noundef double @test_nextafter( // DEFAULT-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_nextafter_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_nextafter_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_nextafter( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_nextafter_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_nextafter_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_nextafter( // APPROX-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_nextafter_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_nextafter_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_nextafter( // NCRDIV-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_nextafter_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_nextafter_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_nextafter( @@ -5229,25 +5229,25 @@ extern "C" __device__ double test_nextafter(double x, double y) { // DEFAULT-LABEL: define dso_local noundef float @test_norm3df( // DEFAULT-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len3_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len3_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_norm3df( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], float noundef nofpclass(nan inf) [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_len3_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]], float noundef nofpclass(nan inf) [[Z]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_len3_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]], float noundef nofpclass(nan inf) [[Z]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_norm3df( // APPROX-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len3_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len3_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_norm3df( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len3_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len3_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_norm3df( @@ -5263,25 +5263,25 @@ extern "C" __device__ float test_norm3df(float x, float y, float z) { // DEFAULT-LABEL: define dso_local noundef double @test_norm3d( // DEFAULT-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len3_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len3_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_norm3d( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], double noundef nofpclass(nan inf) [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_len3_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]], double noundef nofpclass(nan inf) [[Z]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_len3_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]], double noundef nofpclass(nan inf) [[Z]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_norm3d( // APPROX-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len3_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len3_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_norm3d( // NCRDIV-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len3_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len3_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_norm3d( @@ -5297,25 +5297,25 @@ extern "C" __device__ double test_norm3d(double x, double y, double z) { // DEFAULT-LABEL: define dso_local noundef float @test_norm4df( // DEFAULT-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len4_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]], float noundef [[W]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len4_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]], float noundef [[W]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_norm4df( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], float noundef nofpclass(nan inf) [[Z:%.*]], float noundef nofpclass(nan inf) [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_len4_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]], float noundef nofpclass(nan inf) [[Z]], float noundef nofpclass(nan inf) [[W]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_len4_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]], float noundef nofpclass(nan inf) [[Z]], float noundef nofpclass(nan inf) [[W]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_norm4df( // APPROX-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len4_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]], float noundef [[W]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len4_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]], float noundef [[W]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_norm4df( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len4_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]], float noundef [[W]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len4_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]], float noundef [[W]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_norm4df( @@ -5331,25 +5331,25 @@ extern "C" __device__ float test_norm4df(float x, float y, float z, float w) { // DEFAULT-LABEL: define dso_local noundef double @test_norm4d( // DEFAULT-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len4_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]], double noundef [[W]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len4_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]], double noundef [[W]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_norm4d( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], double noundef nofpclass(nan inf) [[Z:%.*]], double noundef nofpclass(nan inf) [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_len4_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]], double noundef nofpclass(nan inf) [[Z]], double noundef nofpclass(nan inf) [[W]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_len4_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]], double noundef nofpclass(nan inf) [[Z]], double noundef nofpclass(nan inf) [[W]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_norm4d( // APPROX-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len4_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]], double noundef [[W]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len4_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]], double noundef [[W]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_norm4d( // NCRDIV-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len4_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]], double noundef [[W]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len4_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]], double noundef [[W]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_norm4d( @@ -5365,25 +5365,25 @@ extern "C" __device__ double test_norm4d(double x, double y, double z, double w) // DEFAULT-LABEL: define dso_local noundef float @test_normcdff( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdf_f32(float noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdf_f32(float noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_normcdff( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_ncdf_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_ncdf_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_normcdff( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdf_f32(float noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdf_f32(float noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_normcdff( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdf_f32(float noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdf_f32(float noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_normcdff( @@ -5399,25 +5399,25 @@ extern "C" __device__ float test_normcdff(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_normcdf( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdf_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdf_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_normcdf( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_ncdf_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_ncdf_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_normcdf( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdf_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdf_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_normcdf( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdf_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdf_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_normcdf( @@ -5433,25 +5433,25 @@ extern "C" __device__ double test_normcdf(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_normcdfinvf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdfinv_f32(float noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdfinv_f32(float noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_normcdfinvf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_ncdfinv_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_ncdfinv_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_normcdfinvf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdfinv_f32(float noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdfinv_f32(float noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_normcdfinvf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdfinv_f32(float noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdfinv_f32(float noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_normcdfinvf( @@ -5467,25 +5467,25 @@ extern "C" __device__ float test_normcdfinvf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_normcdfinv( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdfinv_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdfinv_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_normcdfinv( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_ncdfinv_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_ncdfinv_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_normcdfinv( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdfinv_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdfinv_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_normcdfinv( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdfinv_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdfinv_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_normcdfinv( @@ -5739,25 +5739,25 @@ extern "C" __device__ double test_norm(int x, const double *y) { // DEFAULT-LABEL: define dso_local noundef float @test_powf( // DEFAULT-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_powf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_pow_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_pow_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_powf( // APPROX-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_powf( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_powf( @@ -5773,25 +5773,25 @@ extern "C" __device__ float test_powf(float x, float y) { // DEFAULT-LABEL: define dso_local noundef double @test_pow( // DEFAULT-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pow_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pow_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_pow( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_pow_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_pow_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_pow( // APPROX-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pow_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pow_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_pow( // NCRDIV-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pow_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pow_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_pow( @@ -5807,25 +5807,25 @@ extern "C" __device__ double test_pow(double x, double y) { // DEFAULT-LABEL: define dso_local noundef float @test_powif( // DEFAULT-SAME: float noundef [[X:%.*]], i32 noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pown_f32(float noundef [[X]], i32 noundef [[Y]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pown_f32(float noundef [[X]], i32 noundef [[Y]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_powif( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], i32 noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_pown_f32(float noundef nofpclass(nan inf) [[X]], i32 noundef [[Y]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_pown_f32(float noundef nofpclass(nan inf) [[X]], i32 noundef [[Y]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_powif( // APPROX-SAME: float noundef [[X:%.*]], i32 noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pown_f32(float noundef [[X]], i32 noundef [[Y]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pown_f32(float noundef [[X]], i32 noundef [[Y]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_powif( // NCRDIV-SAME: float noundef [[X:%.*]], i32 noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pown_f32(float noundef [[X]], i32 noundef [[Y]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pown_f32(float noundef [[X]], i32 noundef [[Y]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_powif( @@ -5841,25 +5841,25 @@ extern "C" __device__ float test_powif(float x, int y) { // DEFAULT-LABEL: define dso_local noundef double @test_powi( // DEFAULT-SAME: double noundef [[X:%.*]], i32 noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pown_f64(double noundef [[X]], i32 noundef [[Y]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pown_f64(double noundef [[X]], i32 noundef [[Y]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_powi( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], i32 noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_pown_f64(double noundef nofpclass(nan inf) [[X]], i32 noundef [[Y]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_pown_f64(double noundef nofpclass(nan inf) [[X]], i32 noundef [[Y]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_powi( // APPROX-SAME: double noundef [[X:%.*]], i32 noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pown_f64(double noundef [[X]], i32 noundef [[Y]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pown_f64(double noundef [[X]], i32 noundef [[Y]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_powi( // NCRDIV-SAME: double noundef [[X:%.*]], i32 noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pown_f64(double noundef [[X]], i32 noundef [[Y]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pown_f64(double noundef [[X]], i32 noundef [[Y]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_powi( @@ -5875,25 +5875,25 @@ extern "C" __device__ double test_powi(double x, int y) { // DEFAULT-LABEL: define dso_local noundef float @test_rcbrtf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rcbrt_f32(float noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rcbrt_f32(float noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_rcbrtf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rcbrt_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rcbrt_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_rcbrtf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rcbrt_f32(float noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rcbrt_f32(float noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_rcbrtf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rcbrt_f32(float noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rcbrt_f32(float noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_rcbrtf( @@ -5909,25 +5909,25 @@ extern "C" __device__ float test_rcbrtf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_rcbrt( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rcbrt_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rcbrt_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_rcbrt( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rcbrt_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rcbrt_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_rcbrt( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rcbrt_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rcbrt_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_rcbrt( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rcbrt_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rcbrt_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_rcbrt( @@ -5943,25 +5943,25 @@ extern "C" __device__ double test_rcbrt(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_remainderf( // DEFAULT-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_remainder_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_remainder_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_remainderf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_remainder_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_remainder_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_remainderf( // APPROX-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_remainder_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_remainder_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_remainderf( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_remainder_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_remainder_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_remainderf( @@ -5977,25 +5977,25 @@ extern "C" __device__ float test_remainderf(float x, float y) { // DEFAULT-LABEL: define dso_local noundef double @test_remainder( // DEFAULT-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_remainder_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_remainder_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_remainder( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_remainder_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_remainder_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_remainder( // APPROX-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_remainder_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_remainder_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_remainder( // NCRDIV-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_remainder_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_remainder_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_remainder( @@ -6012,44 +6012,44 @@ extern "C" __device__ double test_remainder(double x, double y) { // DEFAULT-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] // DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5) -// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_remquo_f32(float noundef [[X]], float noundef [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] -// DEFAULT-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA12]] -// DEFAULT-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA12]] -// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_remquo_f32(float noundef [[X]], float noundef [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] +// DEFAULT-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA4]] +// DEFAULT-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA4]] +// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_remquof( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] // FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5) -// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_remquo_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] -// FINITEONLY-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA12]] -// FINITEONLY-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA12]] -// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_remquo_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA4]] +// FINITEONLY-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA4]] +// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_remquof( // APPROX-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] // APPROX-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5) -// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_remquo_f32(float noundef [[X]], float noundef [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] -// APPROX-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA12]] -// APPROX-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA12]] -// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_remquo_f32(float noundef [[X]], float noundef [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] +// APPROX-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA4]] +// APPROX-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA4]] +// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_remquof( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] // NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5) -// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_remquo_f32(float noundef [[X]], float noundef [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] -// NCRDIV-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA13]] -// NCRDIV-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA13]] -// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_remquo_f32(float noundef [[X]], float noundef [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] +// NCRDIV-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA4]] +// NCRDIV-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA4]] +// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_remquof( @@ -6059,8 +6059,8 @@ extern "C" __device__ double test_remainder(double x, double y) { // AMDGCNSPIRV-NEXT: [[__TMP_ASCAST_I:%.*]] = addrspacecast ptr [[__TMP_I]] to ptr addrspace(4) // AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.start.p0(ptr nonnull [[__TMP_I]]) #[[ATTR15]] // AMDGCNSPIRV-NEXT: [[CALL_I:%.*]] = call contract spir_func noundef addrspace(4) float @__ocml_remquo_f32(float noundef [[X]], float noundef [[Y]], ptr noundef nonnull [[__TMP_I]]) #[[ATTR14]] -// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(4) [[__TMP_ASCAST_I]], align 4, !tbaa [[INT_TBAA13]] -// AMDGCNSPIRV-NEXT: store i32 [[TMP0]], ptr addrspace(4) [[Z]], align 4, !tbaa [[INT_TBAA13]] +// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(4) [[__TMP_ASCAST_I]], align 4, !tbaa [[INT_TBAA5]] +// AMDGCNSPIRV-NEXT: store i32 [[TMP0]], ptr addrspace(4) [[Z]], align 4, !tbaa [[INT_TBAA5]] // AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.end.p0(ptr nonnull [[__TMP_I]]) #[[ATTR15]] // AMDGCNSPIRV-NEXT: ret float [[CALL_I]] // @@ -6072,44 +6072,44 @@ extern "C" __device__ float test_remquof(float x, float y, int* z) { // DEFAULT-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] // DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5) -// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_remquo_f64(double noundef [[X]], double noundef [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] -// DEFAULT-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA12]] -// DEFAULT-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA12]] -// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_remquo_f64(double noundef [[X]], double noundef [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] +// DEFAULT-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA4]] +// DEFAULT-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA4]] +// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_remquo( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] // FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5) -// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_remquo_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] -// FINITEONLY-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA12]] -// FINITEONLY-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA12]] -// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_remquo_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA4]] +// FINITEONLY-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA4]] +// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_remquo( // APPROX-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] // APPROX-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5) -// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_remquo_f64(double noundef [[X]], double noundef [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] -// APPROX-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA12]] -// APPROX-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA12]] -// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_remquo_f64(double noundef [[X]], double noundef [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] +// APPROX-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA4]] +// APPROX-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA4]] +// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_remquo( // NCRDIV-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] // NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5) -// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_remquo_f64(double noundef [[X]], double noundef [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] -// NCRDIV-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA13]] -// NCRDIV-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA13]] -// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_remquo_f64(double noundef [[X]], double noundef [[Y]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] +// NCRDIV-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[INT_TBAA4]] +// NCRDIV-NEXT: store i32 [[TMP0]], ptr [[Z]], align 4, !tbaa [[INT_TBAA4]] +// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_remquo( @@ -6119,8 +6119,8 @@ extern "C" __device__ float test_remquof(float x, float y, int* z) { // AMDGCNSPIRV-NEXT: [[__TMP_ASCAST_I:%.*]] = addrspacecast ptr [[__TMP_I]] to ptr addrspace(4) // AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.start.p0(ptr nonnull [[__TMP_I]]) #[[ATTR15]] // AMDGCNSPIRV-NEXT: [[CALL_I:%.*]] = call contract spir_func noundef addrspace(4) double @__ocml_remquo_f64(double noundef [[X]], double noundef [[Y]], ptr noundef nonnull [[__TMP_I]]) #[[ATTR14]] -// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(4) [[__TMP_ASCAST_I]], align 4, !tbaa [[INT_TBAA13]] -// AMDGCNSPIRV-NEXT: store i32 [[TMP0]], ptr addrspace(4) [[Z]], align 4, !tbaa [[INT_TBAA13]] +// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(4) [[__TMP_ASCAST_I]], align 4, !tbaa [[INT_TBAA5]] +// AMDGCNSPIRV-NEXT: store i32 [[TMP0]], ptr addrspace(4) [[Z]], align 4, !tbaa [[INT_TBAA5]] // AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.end.p0(ptr nonnull [[__TMP_I]]) #[[ATTR15]] // AMDGCNSPIRV-NEXT: ret double [[CALL_I]] // @@ -6131,25 +6131,25 @@ extern "C" __device__ double test_remquo(double x, double y, int* z) { // DEFAULT-LABEL: define dso_local noundef float @test_rhypotf( // DEFAULT-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rhypot_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rhypot_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_rhypotf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rhypot_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rhypot_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_rhypotf( // APPROX-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rhypot_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rhypot_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_rhypotf( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rhypot_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rhypot_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_rhypotf( @@ -6165,25 +6165,25 @@ extern "C" __device__ float test_rhypotf(float x, float y) { // DEFAULT-LABEL: define dso_local noundef double @test_rhypot( // DEFAULT-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rhypot_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rhypot_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_rhypot( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rhypot_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rhypot_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_rhypot( // APPROX-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rhypot_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rhypot_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_rhypot( // NCRDIV-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rhypot_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rhypot_f64(double noundef [[X]], double noundef [[Y]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_rhypot( @@ -6282,7 +6282,7 @@ extern "C" __device__ double test_rint(double x) { // DEFAULT-NEXT: br i1 [[TOBOOL_NOT_I]], label %[[_ZL6RNORMFIPKF_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP22:![0-9]+]] // DEFAULT: [[_ZL6RNORMFIPKF_EXIT]]: // DEFAULT-NEXT: [[__R_0_I_LCSSA:%.*]] = phi float [ 0.000000e+00, %[[ENTRY]] ], [ [[ADD_I]], %[[WHILE_BODY_I]] ] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[__R_0_I_LCSSA]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[__R_0_I_LCSSA]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_rnormf( @@ -6303,7 +6303,7 @@ extern "C" __device__ double test_rint(double x) { // FINITEONLY-NEXT: br i1 [[TOBOOL_NOT_I]], label %[[_ZL6RNORMFIPKF_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP22:![0-9]+]] // FINITEONLY: [[_ZL6RNORMFIPKF_EXIT]]: // FINITEONLY-NEXT: [[__R_0_I_LCSSA:%.*]] = phi float [ 0.000000e+00, %[[ENTRY]] ], [ [[ADD_I]], %[[WHILE_BODY_I]] ] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rsqrt_f32(float noundef nofpclass(nan inf) [[__R_0_I_LCSSA]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rsqrt_f32(float noundef nofpclass(nan inf) [[__R_0_I_LCSSA]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_rnormf( @@ -6324,7 +6324,7 @@ extern "C" __device__ double test_rint(double x) { // APPROX-NEXT: br i1 [[TOBOOL_NOT_I]], label %[[_ZL6RNORMFIPKF_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP22:![0-9]+]] // APPROX: [[_ZL6RNORMFIPKF_EXIT]]: // APPROX-NEXT: [[__R_0_I_LCSSA:%.*]] = phi float [ 0.000000e+00, %[[ENTRY]] ], [ [[ADD_I]], %[[WHILE_BODY_I]] ] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[__R_0_I_LCSSA]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[__R_0_I_LCSSA]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_rnormf( @@ -6345,7 +6345,7 @@ extern "C" __device__ double test_rint(double x) { // NCRDIV-NEXT: br i1 [[TOBOOL_NOT_I]], label %[[_ZL6RNORMFIPKF_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP23:![0-9]+]] // NCRDIV: [[_ZL6RNORMFIPKF_EXIT]]: // NCRDIV-NEXT: [[__R_0_I_LCSSA:%.*]] = phi float [ 0.000000e+00, %[[ENTRY]] ], [ [[ADD_I]], %[[WHILE_BODY_I]] ] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[__R_0_I_LCSSA]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[__R_0_I_LCSSA]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_rnormf( @@ -6391,7 +6391,7 @@ extern "C" __device__ float test_rnormf(int x, const float* y) { // DEFAULT-NEXT: br i1 [[TOBOOL_NOT_I]], label %[[_ZL5RNORMIPKD_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP23:![0-9]+]] // DEFAULT: [[_ZL5RNORMIPKD_EXIT]]: // DEFAULT-NEXT: [[__R_0_I_LCSSA:%.*]] = phi double [ 0.000000e+00, %[[ENTRY]] ], [ [[ADD_I]], %[[WHILE_BODY_I]] ] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[__R_0_I_LCSSA]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[__R_0_I_LCSSA]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_rnorm( @@ -6412,7 +6412,7 @@ extern "C" __device__ float test_rnormf(int x, const float* y) { // FINITEONLY-NEXT: br i1 [[TOBOOL_NOT_I]], label %[[_ZL5RNORMIPKD_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP23:![0-9]+]] // FINITEONLY: [[_ZL5RNORMIPKD_EXIT]]: // FINITEONLY-NEXT: [[__R_0_I_LCSSA:%.*]] = phi double [ 0.000000e+00, %[[ENTRY]] ], [ [[ADD_I]], %[[WHILE_BODY_I]] ] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rsqrt_f64(double noundef nofpclass(nan inf) [[__R_0_I_LCSSA]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rsqrt_f64(double noundef nofpclass(nan inf) [[__R_0_I_LCSSA]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_rnorm( @@ -6433,7 +6433,7 @@ extern "C" __device__ float test_rnormf(int x, const float* y) { // APPROX-NEXT: br i1 [[TOBOOL_NOT_I]], label %[[_ZL5RNORMIPKD_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP23:![0-9]+]] // APPROX: [[_ZL5RNORMIPKD_EXIT]]: // APPROX-NEXT: [[__R_0_I_LCSSA:%.*]] = phi double [ 0.000000e+00, %[[ENTRY]] ], [ [[ADD_I]], %[[WHILE_BODY_I]] ] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[__R_0_I_LCSSA]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[__R_0_I_LCSSA]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_rnorm( @@ -6454,7 +6454,7 @@ extern "C" __device__ float test_rnormf(int x, const float* y) { // NCRDIV-NEXT: br i1 [[TOBOOL_NOT_I]], label %[[_ZL5RNORMIPKD_EXIT]], label %[[WHILE_BODY_I]], !llvm.loop [[LOOP24:![0-9]+]] // NCRDIV: [[_ZL5RNORMIPKD_EXIT]]: // NCRDIV-NEXT: [[__R_0_I_LCSSA:%.*]] = phi double [ 0.000000e+00, %[[ENTRY]] ], [ [[ADD_I]], %[[WHILE_BODY_I]] ] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[__R_0_I_LCSSA]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[__R_0_I_LCSSA]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_rnorm( @@ -6485,25 +6485,25 @@ extern "C" __device__ double test_rnorm(int x, const double* y) { // DEFAULT-LABEL: define dso_local noundef float @test_rnorm3df( // DEFAULT-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen3_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen3_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_rnorm3df( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], float noundef nofpclass(nan inf) [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rlen3_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]], float noundef nofpclass(nan inf) [[Z]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rlen3_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]], float noundef nofpclass(nan inf) [[Z]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_rnorm3df( // APPROX-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen3_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen3_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_rnorm3df( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen3_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen3_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_rnorm3df( @@ -6519,25 +6519,25 @@ extern "C" __device__ float test_rnorm3df(float x, float y, float z) { // DEFAULT-LABEL: define dso_local noundef double @test_rnorm3d( // DEFAULT-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen3_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen3_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_rnorm3d( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], double noundef nofpclass(nan inf) [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rlen3_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]], double noundef nofpclass(nan inf) [[Z]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rlen3_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]], double noundef nofpclass(nan inf) [[Z]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_rnorm3d( // APPROX-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen3_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen3_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_rnorm3d( // NCRDIV-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen3_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen3_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_rnorm3d( @@ -6553,25 +6553,25 @@ extern "C" __device__ double test_rnorm3d(double x, double y, double z) { // DEFAULT-LABEL: define dso_local noundef float @test_rnorm4df( // DEFAULT-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen4_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]], float noundef [[W]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen4_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]], float noundef [[W]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_rnorm4df( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], float noundef nofpclass(nan inf) [[Z:%.*]], float noundef nofpclass(nan inf) [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rlen4_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]], float noundef nofpclass(nan inf) [[Z]], float noundef nofpclass(nan inf) [[W]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rlen4_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]], float noundef nofpclass(nan inf) [[Z]], float noundef nofpclass(nan inf) [[W]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_rnorm4df( // APPROX-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen4_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]], float noundef [[W]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen4_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]], float noundef [[W]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_rnorm4df( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen4_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]], float noundef [[W]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen4_f32(float noundef [[X]], float noundef [[Y]], float noundef [[Z]], float noundef [[W]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_rnorm4df( @@ -6587,25 +6587,25 @@ extern "C" __device__ float test_rnorm4df(float x, float y, float z, float w) { // DEFAULT-LABEL: define dso_local noundef double @test_rnorm4d( // DEFAULT-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen4_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]], double noundef [[W]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen4_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]], double noundef [[W]]) #[[ATTR12]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_rnorm4d( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], double noundef nofpclass(nan inf) [[Z:%.*]], double noundef nofpclass(nan inf) [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rlen4_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]], double noundef nofpclass(nan inf) [[Z]], double noundef nofpclass(nan inf) [[W]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rlen4_f64(double noundef nofpclass(nan inf) [[X]], double noundef nofpclass(nan inf) [[Y]], double noundef nofpclass(nan inf) [[Z]], double noundef nofpclass(nan inf) [[W]]) #[[ATTR12]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_rnorm4d( // APPROX-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen4_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]], double noundef [[W]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen4_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]], double noundef [[W]]) #[[ATTR12]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_rnorm4d( // NCRDIV-SAME: double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen4_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]], double noundef [[W]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen4_f64(double noundef [[X]], double noundef [[Y]], double noundef [[Z]], double noundef [[W]]) #[[ATTR12]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_rnorm4d( @@ -6689,25 +6689,25 @@ extern "C" __device__ double test_round(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_rsqrtf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_rsqrtf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rsqrt_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rsqrt_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_rsqrtf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_rsqrtf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_rsqrtf( @@ -6723,25 +6723,25 @@ extern "C" __device__ float test_rsqrtf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_rsqrt( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_rsqrt( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rsqrt_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rsqrt_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_rsqrt( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_rsqrt( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_rsqrt( @@ -6997,48 +6997,48 @@ extern "C" __device__ BOOL_TYPE test___signbit(double x) { // DEFAULT-SAME: float noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] // DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5) -// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincos_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincos_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // DEFAULT-NEXT: store float [[CALL_I]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA16]] // DEFAULT-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[FLOAT_TBAA16]] // DEFAULT-NEXT: store float [[TMP0]], ptr [[Z]], align 4, !tbaa [[FLOAT_TBAA16]] -// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // DEFAULT-NEXT: ret void // // FINITEONLY-LABEL: define dso_local void @test_sincosf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] // FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5) -// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) float @__ocml_sincos_f32(float noundef nofpclass(nan inf) [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) float @__ocml_sincos_f32(float noundef nofpclass(nan inf) [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // FINITEONLY-NEXT: store float [[CALL_I]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA16]] // FINITEONLY-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[FLOAT_TBAA16]] // FINITEONLY-NEXT: store float [[TMP0]], ptr [[Z]], align 4, !tbaa [[FLOAT_TBAA16]] -// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // FINITEONLY-NEXT: ret void // // APPROX-LABEL: define dso_local void @test_sincosf( // APPROX-SAME: float noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] // APPROX-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5) -// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// APPROX-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincos_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincos_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // APPROX-NEXT: store float [[CALL_I]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA16]] // APPROX-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[FLOAT_TBAA16]] // APPROX-NEXT: store float [[TMP0]], ptr [[Z]], align 4, !tbaa [[FLOAT_TBAA16]] -// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // APPROX-NEXT: ret void // // NCRDIV-LABEL: define dso_local void @test_sincosf( // NCRDIV-SAME: float noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] // NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5) -// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincos_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincos_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // NCRDIV-NEXT: store float [[CALL_I]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA17]] // NCRDIV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[FLOAT_TBAA17]] // NCRDIV-NEXT: store float [[TMP0]], ptr [[Z]], align 4, !tbaa [[FLOAT_TBAA17]] -// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // NCRDIV-NEXT: ret void // // AMDGCNSPIRV-LABEL: define spir_func void @test_sincosf( @@ -7062,48 +7062,48 @@ extern "C" __device__ void test_sincosf(float x, float *y, float *z) { // DEFAULT-SAME: double noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] // DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5) -// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincos_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincos_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // DEFAULT-NEXT: store double [[CALL_I]], ptr [[Y]], align 8, !tbaa [[DOUBLE_TBAA18]] // DEFAULT-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[DOUBLE_TBAA18]] // DEFAULT-NEXT: store double [[TMP0]], ptr [[Z]], align 8, !tbaa [[DOUBLE_TBAA18]] -// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // DEFAULT-NEXT: ret void // // FINITEONLY-LABEL: define dso_local void @test_sincos( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] // FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5) -// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) double @__ocml_sincos_f64(double noundef nofpclass(nan inf) [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) double @__ocml_sincos_f64(double noundef nofpclass(nan inf) [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // FINITEONLY-NEXT: store double [[CALL_I]], ptr [[Y]], align 8, !tbaa [[DOUBLE_TBAA18]] // FINITEONLY-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[DOUBLE_TBAA18]] // FINITEONLY-NEXT: store double [[TMP0]], ptr [[Z]], align 8, !tbaa [[DOUBLE_TBAA18]] -// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // FINITEONLY-NEXT: ret void // // APPROX-LABEL: define dso_local void @test_sincos( // APPROX-SAME: double noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] // APPROX-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5) -// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// APPROX-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincos_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincos_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // APPROX-NEXT: store double [[CALL_I]], ptr [[Y]], align 8, !tbaa [[DOUBLE_TBAA18]] // APPROX-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[DOUBLE_TBAA18]] // APPROX-NEXT: store double [[TMP0]], ptr [[Z]], align 8, !tbaa [[DOUBLE_TBAA18]] -// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // APPROX-NEXT: ret void // // NCRDIV-LABEL: define dso_local void @test_sincos( // NCRDIV-SAME: double noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] // NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5) -// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincos_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincos_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // NCRDIV-NEXT: store double [[CALL_I]], ptr [[Y]], align 8, !tbaa [[DOUBLE_TBAA19]] // NCRDIV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[DOUBLE_TBAA19]] // NCRDIV-NEXT: store double [[TMP0]], ptr [[Z]], align 8, !tbaa [[DOUBLE_TBAA19]] -// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // NCRDIV-NEXT: ret void // // AMDGCNSPIRV-LABEL: define spir_func void @test_sincos( @@ -7127,48 +7127,48 @@ extern "C" __device__ void test_sincos(double x, double *y, double *z) { // DEFAULT-SAME: float noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] // DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5) -// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincospi_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincospi_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // DEFAULT-NEXT: store float [[CALL_I]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA16]] // DEFAULT-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[FLOAT_TBAA16]] // DEFAULT-NEXT: store float [[TMP0]], ptr [[Z]], align 4, !tbaa [[FLOAT_TBAA16]] -// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // DEFAULT-NEXT: ret void // // FINITEONLY-LABEL: define dso_local void @test_sincospif( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] // FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5) -// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) float @__ocml_sincospi_f32(float noundef nofpclass(nan inf) [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) float @__ocml_sincospi_f32(float noundef nofpclass(nan inf) [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // FINITEONLY-NEXT: store float [[CALL_I]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA16]] // FINITEONLY-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[FLOAT_TBAA16]] // FINITEONLY-NEXT: store float [[TMP0]], ptr [[Z]], align 4, !tbaa [[FLOAT_TBAA16]] -// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // FINITEONLY-NEXT: ret void // // APPROX-LABEL: define dso_local void @test_sincospif( // APPROX-SAME: float noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] // APPROX-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5) -// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// APPROX-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincospi_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincospi_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // APPROX-NEXT: store float [[CALL_I]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA16]] // APPROX-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[FLOAT_TBAA16]] // APPROX-NEXT: store float [[TMP0]], ptr [[Z]], align 4, !tbaa [[FLOAT_TBAA16]] -// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // APPROX-NEXT: ret void // // NCRDIV-LABEL: define dso_local void @test_sincospif( // NCRDIV-SAME: float noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] // NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5) -// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincospi_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincospi_f32(float noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // NCRDIV-NEXT: store float [[CALL_I]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA17]] // NCRDIV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[FLOAT_TBAA17]] // NCRDIV-NEXT: store float [[TMP0]], ptr [[Z]], align 4, !tbaa [[FLOAT_TBAA17]] -// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // NCRDIV-NEXT: ret void // // AMDGCNSPIRV-LABEL: define spir_func void @test_sincospif( @@ -7192,48 +7192,48 @@ extern "C" __device__ void test_sincospif(float x, float *y, float *z) { // DEFAULT-SAME: double noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] // DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5) -// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincospi_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincospi_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // DEFAULT-NEXT: store double [[CALL_I]], ptr [[Y]], align 8, !tbaa [[DOUBLE_TBAA18]] // DEFAULT-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[DOUBLE_TBAA18]] // DEFAULT-NEXT: store double [[TMP0]], ptr [[Z]], align 8, !tbaa [[DOUBLE_TBAA18]] -// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // DEFAULT-NEXT: ret void // // FINITEONLY-LABEL: define dso_local void @test_sincospi( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] // FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5) -// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) double @__ocml_sincospi_f64(double noundef nofpclass(nan inf) [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) double @__ocml_sincospi_f64(double noundef nofpclass(nan inf) [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // FINITEONLY-NEXT: store double [[CALL_I]], ptr [[Y]], align 8, !tbaa [[DOUBLE_TBAA18]] // FINITEONLY-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[DOUBLE_TBAA18]] // FINITEONLY-NEXT: store double [[TMP0]], ptr [[Z]], align 8, !tbaa [[DOUBLE_TBAA18]] -// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // FINITEONLY-NEXT: ret void // // APPROX-LABEL: define dso_local void @test_sincospi( // APPROX-SAME: double noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] // APPROX-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5) -// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// APPROX-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincospi_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// APPROX-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincospi_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // APPROX-NEXT: store double [[CALL_I]], ptr [[Y]], align 8, !tbaa [[DOUBLE_TBAA18]] // APPROX-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[DOUBLE_TBAA18]] // APPROX-NEXT: store double [[TMP0]], ptr [[Z]], align 8, !tbaa [[DOUBLE_TBAA18]] -// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// APPROX-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // APPROX-NEXT: ret void // // NCRDIV-LABEL: define dso_local void @test_sincospi( // NCRDIV-SAME: double noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 8)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] // NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5) -// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincospi_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]] +// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincospi_f64(double noundef [[X]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]] // NCRDIV-NEXT: store double [[CALL_I]], ptr [[Y]], align 8, !tbaa [[DOUBLE_TBAA19]] // NCRDIV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[DOUBLE_TBAA19]] // NCRDIV-NEXT: store double [[TMP0]], ptr [[Z]], align 8, !tbaa [[DOUBLE_TBAA19]] -// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]] +// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]] // NCRDIV-NEXT: ret void // // AMDGCNSPIRV-LABEL: define spir_func void @test_sincospi( @@ -7256,25 +7256,25 @@ extern "C" __device__ void test_sincospi(double x, double *y, double *z) { // DEFAULT-LABEL: define dso_local noundef float @test_sinf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sin_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sin_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_sinf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_sin_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_sin_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_sinf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I1:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I1:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret float [[CALL_I1]] // // NCRDIV-LABEL: define dso_local noundef float @test_sinf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sin_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sin_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_sinf( @@ -7290,25 +7290,25 @@ extern "C" __device__ float test_sinf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_sin( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sin_f64(double noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sin_f64(double noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_sin( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_sin_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_sin_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_sin( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sin_f64(double noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sin_f64(double noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_sin( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sin_f64(double noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sin_f64(double noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_sin( @@ -7324,25 +7324,25 @@ extern "C" __device__ double test_sin(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_sinpif( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sinpi_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sinpi_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_sinpif( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_sinpi_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_sinpi_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_sinpif( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sinpi_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sinpi_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_sinpif( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sinpi_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sinpi_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_sinpif( @@ -7358,25 +7358,25 @@ extern "C" __device__ float test_sinpif(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_sinpi( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sinpi_f64(double noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sinpi_f64(double noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_sinpi( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_sinpi_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_sinpi_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_sinpi( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sinpi_f64(double noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sinpi_f64(double noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_sinpi( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sinpi_f64(double noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sinpi_f64(double noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_sinpi( @@ -7460,25 +7460,25 @@ extern "C" __device__ double test_sqrt(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_tanf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tan_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tan_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_tanf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_tan_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_tan_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_tanf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tan_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tan_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_tanf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tan_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tan_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_tanf( @@ -7494,25 +7494,25 @@ extern "C" __device__ float test_tanf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_tan( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tan_f64(double noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tan_f64(double noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_tan( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_tan_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_tan_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_tan( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tan_f64(double noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tan_f64(double noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_tan( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tan_f64(double noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tan_f64(double noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_tan( @@ -7528,25 +7528,25 @@ extern "C" __device__ double test_tan(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_tanhf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tanh_f32(float noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tanh_f32(float noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_tanhf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_tanh_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_tanh_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_tanhf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tanh_f32(float noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tanh_f32(float noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_tanhf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tanh_f32(float noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tanh_f32(float noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_tanhf( @@ -7562,25 +7562,25 @@ extern "C" __device__ float test_tanhf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_tanh( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tanh_f64(double noundef [[X]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tanh_f64(double noundef [[X]]) #[[ATTR13]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_tanh( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_tanh_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_tanh_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR13]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_tanh( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tanh_f64(double noundef [[X]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tanh_f64(double noundef [[X]]) #[[ATTR13]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_tanh( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tanh_f64(double noundef [[X]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tanh_f64(double noundef [[X]]) #[[ATTR13]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_tanh( @@ -7596,25 +7596,25 @@ extern "C" __device__ double test_tanh(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_tgammaf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tgamma_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tgamma_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_tgammaf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_tgamma_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_tgamma_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_tgammaf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tgamma_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tgamma_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_tgammaf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tgamma_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tgamma_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_tgammaf( @@ -7630,25 +7630,25 @@ extern "C" __device__ float test_tgammaf(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_tgamma( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tgamma_f64(double noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tgamma_f64(double noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_tgamma( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_tgamma_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_tgamma_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_tgamma( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tgamma_f64(double noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tgamma_f64(double noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_tgamma( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tgamma_f64(double noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tgamma_f64(double noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_tgamma( @@ -7732,25 +7732,25 @@ extern "C" __device__ double test_trunc(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_y0f( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_y0f( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y0_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y0_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_y0f( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_y0f( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_y0f( @@ -7766,25 +7766,25 @@ extern "C" __device__ float test_y0f(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_y0( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_y0( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y0_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y0_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_y0( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_y0( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_y0( @@ -7800,25 +7800,25 @@ extern "C" __device__ double test_y0(double x) { // DEFAULT-LABEL: define dso_local noundef float @test_y1f( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test_y1f( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y1_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y1_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test_y1f( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test_y1f( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test_y1f( @@ -7834,25 +7834,25 @@ extern "C" __device__ float test_y1f(float x) { // DEFAULT-LABEL: define dso_local noundef double @test_y1( // DEFAULT-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret double [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) double @test_y1( // FINITEONLY-SAME: double noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y1_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y1_f64(double noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret double [[CALL_I]] // // APPROX-LABEL: define dso_local noundef double @test_y1( // APPROX-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret double [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef double @test_y1( // NCRDIV-SAME: double noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret double [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef double @test_y1( @@ -7873,14 +7873,14 @@ extern "C" __device__ double test_y1(double x) { // DEFAULT-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // DEFAULT-NEXT: ] // DEFAULT: [[IF_THEN_I]]: -// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR14]] // DEFAULT-NEXT: br label %[[_ZL3YNFIF_EXIT:.*]] // DEFAULT: [[IF_THEN2_I]]: -// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR14]] // DEFAULT-NEXT: br label %[[_ZL3YNFIF_EXIT]] // DEFAULT: [[IF_END4_I]]: -// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR16]] -// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR14]] // DEFAULT-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // DEFAULT-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL3YNFIF_EXIT]] // DEFAULT: [[FOR_BODY_I]]: @@ -7907,14 +7907,14 @@ extern "C" __device__ double test_y1(double x) { // FINITEONLY-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // FINITEONLY-NEXT: ] // FINITEONLY: [[IF_THEN_I]]: -// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y0_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y0_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] // FINITEONLY-NEXT: br label %[[_ZL3YNFIF_EXIT:.*]] // FINITEONLY: [[IF_THEN2_I]]: -// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] // FINITEONLY-NEXT: br label %[[_ZL3YNFIF_EXIT]] // FINITEONLY: [[IF_END4_I]]: -// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y0_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] -// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y0_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] // FINITEONLY-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // FINITEONLY-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL3YNFIF_EXIT]] // FINITEONLY: [[FOR_BODY_I]]: @@ -7941,14 +7941,14 @@ extern "C" __device__ double test_y1(double x) { // APPROX-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // APPROX-NEXT: ] // APPROX: [[IF_THEN_I]]: -// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR14]] // APPROX-NEXT: br label %[[_ZL3YNFIF_EXIT:.*]] // APPROX: [[IF_THEN2_I]]: -// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR14]] // APPROX-NEXT: br label %[[_ZL3YNFIF_EXIT]] // APPROX: [[IF_END4_I]]: -// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR16]] -// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR14]] // APPROX-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // APPROX-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL3YNFIF_EXIT]] // APPROX: [[FOR_BODY_I]]: @@ -7975,14 +7975,14 @@ extern "C" __device__ double test_y1(double x) { // NCRDIV-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // NCRDIV-NEXT: ] // NCRDIV: [[IF_THEN_I]]: -// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR14]] // NCRDIV-NEXT: br label %[[_ZL3YNFIF_EXIT:.*]] // NCRDIV: [[IF_THEN2_I]]: -// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR14]] // NCRDIV-NEXT: br label %[[_ZL3YNFIF_EXIT]] // NCRDIV: [[IF_END4_I]]: -// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR16]] -// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR14]] // NCRDIV-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // NCRDIV-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL3YNFIF_EXIT]] // NCRDIV: [[FOR_BODY_I]]: @@ -7991,7 +7991,7 @@ extern "C" __device__ double test_y1(double x) { // NCRDIV-NEXT: [[__X0_0_I2:%.*]] = phi float [ [[__X1_0_I3]], %[[FOR_BODY_I]] ], [ [[CALL_I_I]], %[[IF_END4_I]] ] // NCRDIV-NEXT: [[MUL_I:%.*]] = shl nuw nsw i32 [[__I_0_I4]], 1 // NCRDIV-NEXT: [[CONV_I:%.*]] = uitofp nneg i32 [[MUL_I]] to float -// NCRDIV-NEXT: [[DIV_I:%.*]] = fdiv contract float [[CONV_I]], [[Y]], !fpmath [[META12]] +// NCRDIV-NEXT: [[DIV_I:%.*]] = fdiv contract float [[CONV_I]], [[Y]], !fpmath [[META14]] // NCRDIV-NEXT: [[MUL8_I:%.*]] = fmul contract float [[__X1_0_I3]], [[DIV_I]] // NCRDIV-NEXT: [[SUB_I]] = fsub contract float [[MUL8_I]], [[__X0_0_I2]] // NCRDIV-NEXT: [[INC_I]] = add nuw nsw i32 [[__I_0_I4]], 1 @@ -8047,14 +8047,14 @@ extern "C" __device__ float test_ynf(int x, float y) { // DEFAULT-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // DEFAULT-NEXT: ] // DEFAULT: [[IF_THEN_I]]: -// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR14]] // DEFAULT-NEXT: br label %[[_ZL2YNID_EXIT:.*]] // DEFAULT: [[IF_THEN2_I]]: -// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR14]] // DEFAULT-NEXT: br label %[[_ZL2YNID_EXIT]] // DEFAULT: [[IF_END4_I]]: -// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR16]] -// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR14]] // DEFAULT-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // DEFAULT-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL2YNID_EXIT]] // DEFAULT: [[FOR_BODY_I]]: @@ -8081,14 +8081,14 @@ extern "C" __device__ float test_ynf(int x, float y) { // FINITEONLY-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // FINITEONLY-NEXT: ] // FINITEONLY: [[IF_THEN_I]]: -// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y0_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y0_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] // FINITEONLY-NEXT: br label %[[_ZL2YNID_EXIT:.*]] // FINITEONLY: [[IF_THEN2_I]]: -// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] // FINITEONLY-NEXT: br label %[[_ZL2YNID_EXIT]] // FINITEONLY: [[IF_END4_I]]: -// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y0_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] -// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y0_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]] // FINITEONLY-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // FINITEONLY-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL2YNID_EXIT]] // FINITEONLY: [[FOR_BODY_I]]: @@ -8115,14 +8115,14 @@ extern "C" __device__ float test_ynf(int x, float y) { // APPROX-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // APPROX-NEXT: ] // APPROX: [[IF_THEN_I]]: -// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR14]] // APPROX-NEXT: br label %[[_ZL2YNID_EXIT:.*]] // APPROX: [[IF_THEN2_I]]: -// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR14]] // APPROX-NEXT: br label %[[_ZL2YNID_EXIT]] // APPROX: [[IF_END4_I]]: -// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR16]] -// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR14]] // APPROX-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // APPROX-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL2YNID_EXIT]] // APPROX: [[FOR_BODY_I]]: @@ -8149,14 +8149,14 @@ extern "C" __device__ float test_ynf(int x, float y) { // NCRDIV-NEXT: i32 1, label %[[IF_THEN2_I:.*]] // NCRDIV-NEXT: ] // NCRDIV: [[IF_THEN_I]]: -// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR14]] // NCRDIV-NEXT: br label %[[_ZL2YNID_EXIT:.*]] // NCRDIV: [[IF_THEN2_I]]: -// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR14]] // NCRDIV-NEXT: br label %[[_ZL2YNID_EXIT]] // NCRDIV: [[IF_END4_I]]: -// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR16]] -// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR14]] // NCRDIV-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1 // NCRDIV-NEXT: br i1 [[CMP7_I1]], label %[[FOR_BODY_I:.*]], label %[[_ZL2YNID_EXIT]] // NCRDIV: [[FOR_BODY_I]]: @@ -8216,25 +8216,25 @@ extern "C" __device__ double test_yn(int x, double y) { // DEFAULT-LABEL: define dso_local noundef float @test___cosf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test___cosf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_cos_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_cos_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test___cosf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test___cosf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test___cosf( @@ -8380,7 +8380,7 @@ extern "C" __device__ float test___fadd_rn(float x, float y) { // NCRDIV-LABEL: define dso_local noundef float @test___fdividef( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR3]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[DIV_I:%.*]] = fdiv contract float [[X]], [[Y]], !fpmath [[META12]] +// NCRDIV-NEXT: [[DIV_I:%.*]] = fdiv contract float [[X]], [[Y]], !fpmath [[META14]] // NCRDIV-NEXT: ret float [[DIV_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test___fdividef( @@ -8482,7 +8482,7 @@ extern "C" __device__ float test___fmul_rn(float x, float y) { // NCRDIV-LABEL: define dso_local noundef float @test___frcp_rn( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR3]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[DIV_I:%.*]] = fdiv contract float 1.000000e+00, [[X]], !fpmath [[META12]] +// NCRDIV-NEXT: [[DIV_I:%.*]] = fdiv contract float 1.000000e+00, [[X]], !fpmath [[META14]] // NCRDIV-NEXT: ret float [[DIV_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test___frcp_rn( @@ -8532,25 +8532,25 @@ extern "C" __device__ float test___frsqrt_rn(float x) { // DEFAULT-LABEL: define dso_local noundef float @test___fsqrt_rn( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sqrt_f32(float noundef [[X]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sqrt_f32(float noundef [[X]]) #[[ATTR12]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test___fsqrt_rn( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_sqrt_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_sqrt_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR12]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test___fsqrt_rn( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sqrt_f32(float noundef [[X]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sqrt_f32(float noundef [[X]]) #[[ATTR12]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test___fsqrt_rn( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR4]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sqrt_f32(float noundef [[X]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sqrt_f32(float noundef [[X]]) #[[ATTR12]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test___fsqrt_rn( @@ -8702,25 +8702,25 @@ extern "C" __device__ float test___logf(float x) { // DEFAULT-LABEL: define dso_local noundef float @test___powf( // DEFAULT-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR15]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR13]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test___powf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_pow_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR15]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_pow_f32(float noundef nofpclass(nan inf) [[X]], float noundef nofpclass(nan inf) [[Y]]) #[[ATTR13]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test___powf( // APPROX-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR15]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR13]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test___powf( // NCRDIV-SAME: float noundef [[X:%.*]], float noundef [[Y:%.*]]) local_unnamed_addr #[[ATTR5]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR15]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X]], float noundef [[Y]]) #[[ATTR13]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test___powf( @@ -8785,36 +8785,36 @@ extern "C" __device__ float test___saturatef(float x) { // DEFAULT-LABEL: define dso_local void @test___sincosf( // DEFAULT-SAME: float noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: store float [[CALL_I]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA16]] -// DEFAULT-NEXT: [[CALL1_I:%.*]] = tail call contract float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL1_I:%.*]] = tail call contract float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: store float [[CALL1_I]], ptr [[Z]], align 4, !tbaa [[FLOAT_TBAA16]] // DEFAULT-NEXT: ret void // // FINITEONLY-LABEL: define dso_local void @test___sincosf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract nofpclass(nan inf) float @__ocml_native_sin_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract nofpclass(nan inf) float @__ocml_native_sin_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: store float [[CALL_I]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA16]] -// FINITEONLY-NEXT: [[CALL1_I:%.*]] = tail call nnan ninf contract nofpclass(nan inf) float @__ocml_native_cos_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL1_I:%.*]] = tail call nnan ninf contract nofpclass(nan inf) float @__ocml_native_cos_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: store float [[CALL1_I]], ptr [[Z]], align 4, !tbaa [[FLOAT_TBAA16]] // FINITEONLY-NEXT: ret void // // APPROX-LABEL: define dso_local void @test___sincosf( // APPROX-SAME: float noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: store float [[CALL_I]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA16]] -// APPROX-NEXT: [[CALL1_I:%.*]] = tail call contract float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL1_I:%.*]] = tail call contract float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: store float [[CALL1_I]], ptr [[Z]], align 4, !tbaa [[FLOAT_TBAA16]] // APPROX-NEXT: ret void // // NCRDIV-LABEL: define dso_local void @test___sincosf( // NCRDIV-SAME: float noundef [[X:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Y:%.*]], ptr noundef writeonly captures(none) initializes((0, 4)) [[Z:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: store float [[CALL_I]], ptr [[Y]], align 4, !tbaa [[FLOAT_TBAA17]] -// NCRDIV-NEXT: [[CALL1_I:%.*]] = tail call contract float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL1_I:%.*]] = tail call contract float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: store float [[CALL1_I]], ptr [[Z]], align 4, !tbaa [[FLOAT_TBAA17]] // NCRDIV-NEXT: ret void // @@ -8834,25 +8834,25 @@ extern "C" __device__ void test___sincosf(float x, float *y, float *z) { // DEFAULT-LABEL: define dso_local noundef float @test___sinf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: ret float [[CALL_I]] // // FINITEONLY-LABEL: define dso_local noundef nofpclass(nan inf) float @test___sinf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_sin_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_sin_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: ret float [[CALL_I]] // // APPROX-LABEL: define dso_local noundef float @test___sinf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: ret float [[CALL_I]] // // NCRDIV-LABEL: define dso_local noundef float @test___sinf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: ret float [[CALL_I]] // // AMDGCNSPIRV-LABEL: define spir_func noundef float @test___sinf( @@ -8868,8 +8868,8 @@ extern "C" __device__ float test___sinf(float x) { // DEFAULT-LABEL: define dso_local float @test___tanf( // DEFAULT-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // DEFAULT-NEXT: [[ENTRY:.*:]] -// DEFAULT-NEXT: [[CALL_I3_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR16]] -// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]] +// DEFAULT-NEXT: [[CALL_I3_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR14]] +// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]] // DEFAULT-NEXT: [[TMP0:%.*]] = tail call contract float @llvm.amdgcn.rcp.f32(float [[CALL_I_I]]) // DEFAULT-NEXT: [[MUL_I:%.*]] = fmul contract float [[CALL_I3_I]], [[TMP0]] // DEFAULT-NEXT: ret float [[MUL_I]] @@ -8877,8 +8877,8 @@ extern "C" __device__ float test___sinf(float x) { // FINITEONLY-LABEL: define dso_local nofpclass(nan inf) float @test___tanf( // FINITEONLY-SAME: float noundef nofpclass(nan inf) [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // FINITEONLY-NEXT: [[ENTRY:.*:]] -// FINITEONLY-NEXT: [[CALL_I3_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_sin_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] -// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_cos_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]] +// FINITEONLY-NEXT: [[CALL_I3_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_sin_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] +// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_cos_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]] // FINITEONLY-NEXT: [[TMP0:%.*]] = tail call nnan ninf contract float @llvm.amdgcn.rcp.f32(float [[CALL_I_I]]) // FINITEONLY-NEXT: [[MUL_I:%.*]] = fmul nnan ninf contract float [[CALL_I3_I]], [[TMP0]] // FINITEONLY-NEXT: ret float [[MUL_I]] @@ -8886,8 +8886,8 @@ extern "C" __device__ float test___sinf(float x) { // APPROX-LABEL: define dso_local float @test___tanf( // APPROX-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // APPROX-NEXT: [[ENTRY:.*:]] -// APPROX-NEXT: [[CALL_I3_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR16]] -// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]] +// APPROX-NEXT: [[CALL_I3_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR14]] +// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]] // APPROX-NEXT: [[TMP0:%.*]] = tail call contract float @llvm.amdgcn.rcp.f32(float [[CALL_I_I]]) // APPROX-NEXT: [[MUL_I:%.*]] = fmul contract float [[CALL_I3_I]], [[TMP0]] // APPROX-NEXT: ret float [[MUL_I]] @@ -8895,8 +8895,8 @@ extern "C" __device__ float test___sinf(float x) { // NCRDIV-LABEL: define dso_local float @test___tanf( // NCRDIV-SAME: float noundef [[X:%.*]]) local_unnamed_addr #[[ATTR6]] { // NCRDIV-NEXT: [[ENTRY:.*:]] -// NCRDIV-NEXT: [[CALL_I3_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR16]] -// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]] +// NCRDIV-NEXT: [[CALL_I3_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X]]) #[[ATTR14]] +// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]] // NCRDIV-NEXT: [[TMP0:%.*]] = tail call contract float @llvm.amdgcn.rcp.f32(float [[CALL_I_I]]) // NCRDIV-NEXT: [[MUL_I:%.*]] = fmul contract float [[CALL_I3_I]], [[TMP0]] // NCRDIV-NEXT: ret float [[MUL_I]] @@ -9285,120 +9285,120 @@ extern "C" __device__ int test_int_max(int x, int y) { return max(x, y); } //. -// DEFAULT: [[CHAR_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} -// DEFAULT: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// DEFAULT: [[META6]] = !{!"Simple C++ TBAA"} -// DEFAULT: [[LOOP7]] = distinct !{[[LOOP7]], [[META8:![0-9]+]], [[META9:![0-9]+]]} -// DEFAULT: [[META8]] = !{!"llvm.loop.mustprogress"} -// DEFAULT: [[META9]] = !{!"llvm.loop.unroll.disable"} -// DEFAULT: [[LOOP10]] = distinct !{[[LOOP10]], [[META8]], [[META9]]} -// DEFAULT: [[LOOP11]] = distinct !{[[LOOP11]], [[META8]], [[META9]]} -// DEFAULT: [[INT_TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} -// DEFAULT: [[META13]] = !{!"int", [[META5]], i64 0} -// DEFAULT: [[LOOP14]] = distinct !{[[LOOP14]], [[META8]], [[META9]]} -// DEFAULT: [[LOOP15]] = distinct !{[[LOOP15]], [[META8]], [[META9]]} +// DEFAULT: [[INT_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} +// DEFAULT: [[META5]] = !{!"int", [[META6:![0-9]+]], i64 0} +// DEFAULT: [[META6]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} +// DEFAULT: [[META7]] = !{!"Simple C++ TBAA"} +// DEFAULT: [[CHAR_TBAA8]] = !{[[META6]], [[META6]], i64 0} +// DEFAULT: [[LOOP9]] = distinct !{[[LOOP9]], [[META10:![0-9]+]], [[META11:![0-9]+]]} +// DEFAULT: [[META10]] = !{!"llvm.loop.mustprogress"} +// DEFAULT: [[META11]] = !{!"llvm.loop.unroll.disable"} +// DEFAULT: [[LOOP12]] = distinct !{[[LOOP12]], [[META10]], [[META11]]} +// DEFAULT: [[LOOP13]] = distinct !{[[LOOP13]], [[META10]], [[META11]]} +// DEFAULT: [[LOOP14]] = distinct !{[[LOOP14]], [[META10]], [[META11]]} +// DEFAULT: [[LOOP15]] = distinct !{[[LOOP15]], [[META10]], [[META11]]} // DEFAULT: [[FLOAT_TBAA16]] = !{[[META17:![0-9]+]], [[META17]], i64 0} -// DEFAULT: [[META17]] = !{!"float", [[META5]], i64 0} +// DEFAULT: [[META17]] = !{!"float", [[META6]], i64 0} // DEFAULT: [[DOUBLE_TBAA18]] = !{[[META19:![0-9]+]], [[META19]], i64 0} -// DEFAULT: [[META19]] = !{!"double", [[META5]], i64 0} -// DEFAULT: [[LOOP20]] = distinct !{[[LOOP20]], [[META8]], [[META9]]} -// DEFAULT: [[LOOP21]] = distinct !{[[LOOP21]], [[META8]], [[META9]]} -// DEFAULT: [[LOOP22]] = distinct !{[[LOOP22]], [[META8]], [[META9]]} -// DEFAULT: [[LOOP23]] = distinct !{[[LOOP23]], [[META8]], [[META9]]} -// DEFAULT: [[LOOP24]] = distinct !{[[LOOP24]], [[META8]], [[META9]]} -// DEFAULT: [[LOOP25]] = distinct !{[[LOOP25]], [[META8]], [[META9]]} +// DEFAULT: [[META19]] = !{!"double", [[META6]], i64 0} +// DEFAULT: [[LOOP20]] = distinct !{[[LOOP20]], [[META10]], [[META11]]} +// DEFAULT: [[LOOP21]] = distinct !{[[LOOP21]], [[META10]], [[META11]]} +// DEFAULT: [[LOOP22]] = distinct !{[[LOOP22]], [[META10]], [[META11]]} +// DEFAULT: [[LOOP23]] = distinct !{[[LOOP23]], [[META10]], [[META11]]} +// DEFAULT: [[LOOP24]] = distinct !{[[LOOP24]], [[META10]], [[META11]]} +// DEFAULT: [[LOOP25]] = distinct !{[[LOOP25]], [[META10]], [[META11]]} //. -// FINITEONLY: [[CHAR_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} -// FINITEONLY: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// FINITEONLY: [[META6]] = !{!"Simple C++ TBAA"} -// FINITEONLY: [[LOOP7]] = distinct !{[[LOOP7]], [[META8:![0-9]+]], [[META9:![0-9]+]]} -// FINITEONLY: [[META8]] = !{!"llvm.loop.mustprogress"} -// FINITEONLY: [[META9]] = !{!"llvm.loop.unroll.disable"} -// FINITEONLY: [[LOOP10]] = distinct !{[[LOOP10]], [[META8]], [[META9]]} -// FINITEONLY: [[LOOP11]] = distinct !{[[LOOP11]], [[META8]], [[META9]]} -// FINITEONLY: [[INT_TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} -// FINITEONLY: [[META13]] = !{!"int", [[META5]], i64 0} -// FINITEONLY: [[LOOP14]] = distinct !{[[LOOP14]], [[META8]], [[META9]]} -// FINITEONLY: [[LOOP15]] = distinct !{[[LOOP15]], [[META8]], [[META9]]} +// FINITEONLY: [[INT_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} +// FINITEONLY: [[META5]] = !{!"int", [[META6:![0-9]+]], i64 0} +// FINITEONLY: [[META6]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} +// FINITEONLY: [[META7]] = !{!"Simple C++ TBAA"} +// FINITEONLY: [[CHAR_TBAA8]] = !{[[META6]], [[META6]], i64 0} +// FINITEONLY: [[LOOP9]] = distinct !{[[LOOP9]], [[META10:![0-9]+]], [[META11:![0-9]+]]} +// FINITEONLY: [[META10]] = !{!"llvm.loop.mustprogress"} +// FINITEONLY: [[META11]] = !{!"llvm.loop.unroll.disable"} +// FINITEONLY: [[LOOP12]] = distinct !{[[LOOP12]], [[META10]], [[META11]]} +// FINITEONLY: [[LOOP13]] = distinct !{[[LOOP13]], [[META10]], [[META11]]} +// FINITEONLY: [[LOOP14]] = distinct !{[[LOOP14]], [[META10]], [[META11]]} +// FINITEONLY: [[LOOP15]] = distinct !{[[LOOP15]], [[META10]], [[META11]]} // FINITEONLY: [[FLOAT_TBAA16]] = !{[[META17:![0-9]+]], [[META17]], i64 0} -// FINITEONLY: [[META17]] = !{!"float", [[META5]], i64 0} +// FINITEONLY: [[META17]] = !{!"float", [[META6]], i64 0} // FINITEONLY: [[DOUBLE_TBAA18]] = !{[[META19:![0-9]+]], [[META19]], i64 0} -// FINITEONLY: [[META19]] = !{!"double", [[META5]], i64 0} -// FINITEONLY: [[LOOP20]] = distinct !{[[LOOP20]], [[META8]], [[META9]]} -// FINITEONLY: [[LOOP21]] = distinct !{[[LOOP21]], [[META8]], [[META9]]} -// FINITEONLY: [[LOOP22]] = distinct !{[[LOOP22]], [[META8]], [[META9]]} -// FINITEONLY: [[LOOP23]] = distinct !{[[LOOP23]], [[META8]], [[META9]]} -// FINITEONLY: [[LOOP24]] = distinct !{[[LOOP24]], [[META8]], [[META9]]} -// FINITEONLY: [[LOOP25]] = distinct !{[[LOOP25]], [[META8]], [[META9]]} +// FINITEONLY: [[META19]] = !{!"double", [[META6]], i64 0} +// FINITEONLY: [[LOOP20]] = distinct !{[[LOOP20]], [[META10]], [[META11]]} +// FINITEONLY: [[LOOP21]] = distinct !{[[LOOP21]], [[META10]], [[META11]]} +// FINITEONLY: [[LOOP22]] = distinct !{[[LOOP22]], [[META10]], [[META11]]} +// FINITEONLY: [[LOOP23]] = distinct !{[[LOOP23]], [[META10]], [[META11]]} +// FINITEONLY: [[LOOP24]] = distinct !{[[LOOP24]], [[META10]], [[META11]]} +// FINITEONLY: [[LOOP25]] = distinct !{[[LOOP25]], [[META10]], [[META11]]} //. -// APPROX: [[CHAR_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} -// APPROX: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// APPROX: [[META6]] = !{!"Simple C++ TBAA"} -// APPROX: [[LOOP7]] = distinct !{[[LOOP7]], [[META8:![0-9]+]], [[META9:![0-9]+]]} -// APPROX: [[META8]] = !{!"llvm.loop.mustprogress"} -// APPROX: [[META9]] = !{!"llvm.loop.unroll.disable"} -// APPROX: [[LOOP10]] = distinct !{[[LOOP10]], [[META8]], [[META9]]} -// APPROX: [[LOOP11]] = distinct !{[[LOOP11]], [[META8]], [[META9]]} -// APPROX: [[INT_TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} -// APPROX: [[META13]] = !{!"int", [[META5]], i64 0} -// APPROX: [[LOOP14]] = distinct !{[[LOOP14]], [[META8]], [[META9]]} -// APPROX: [[LOOP15]] = distinct !{[[LOOP15]], [[META8]], [[META9]]} +// APPROX: [[INT_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} +// APPROX: [[META5]] = !{!"int", [[META6:![0-9]+]], i64 0} +// APPROX: [[META6]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} +// APPROX: [[META7]] = !{!"Simple C++ TBAA"} +// APPROX: [[CHAR_TBAA8]] = !{[[META6]], [[META6]], i64 0} +// APPROX: [[LOOP9]] = distinct !{[[LOOP9]], [[META10:![0-9]+]], [[META11:![0-9]+]]} +// APPROX: [[META10]] = !{!"llvm.loop.mustprogress"} +// APPROX: [[META11]] = !{!"llvm.loop.unroll.disable"} +// APPROX: [[LOOP12]] = distinct !{[[LOOP12]], [[META10]], [[META11]]} +// APPROX: [[LOOP13]] = distinct !{[[LOOP13]], [[META10]], [[META11]]} +// APPROX: [[LOOP14]] = distinct !{[[LOOP14]], [[META10]], [[META11]]} +// APPROX: [[LOOP15]] = distinct !{[[LOOP15]], [[META10]], [[META11]]} // APPROX: [[FLOAT_TBAA16]] = !{[[META17:![0-9]+]], [[META17]], i64 0} -// APPROX: [[META17]] = !{!"float", [[META5]], i64 0} +// APPROX: [[META17]] = !{!"float", [[META6]], i64 0} // APPROX: [[DOUBLE_TBAA18]] = !{[[META19:![0-9]+]], [[META19]], i64 0} -// APPROX: [[META19]] = !{!"double", [[META5]], i64 0} -// APPROX: [[LOOP20]] = distinct !{[[LOOP20]], [[META8]], [[META9]]} -// APPROX: [[LOOP21]] = distinct !{[[LOOP21]], [[META8]], [[META9]]} -// APPROX: [[LOOP22]] = distinct !{[[LOOP22]], [[META8]], [[META9]]} -// APPROX: [[LOOP23]] = distinct !{[[LOOP23]], [[META8]], [[META9]]} -// APPROX: [[LOOP24]] = distinct !{[[LOOP24]], [[META8]], [[META9]]} -// APPROX: [[LOOP25]] = distinct !{[[LOOP25]], [[META8]], [[META9]]} +// APPROX: [[META19]] = !{!"double", [[META6]], i64 0} +// APPROX: [[LOOP20]] = distinct !{[[LOOP20]], [[META10]], [[META11]]} +// APPROX: [[LOOP21]] = distinct !{[[LOOP21]], [[META10]], [[META11]]} +// APPROX: [[LOOP22]] = distinct !{[[LOOP22]], [[META10]], [[META11]]} +// APPROX: [[LOOP23]] = distinct !{[[LOOP23]], [[META10]], [[META11]]} +// APPROX: [[LOOP24]] = distinct !{[[LOOP24]], [[META10]], [[META11]]} +// APPROX: [[LOOP25]] = distinct !{[[LOOP25]], [[META10]], [[META11]]} //. -// NCRDIV: [[CHAR_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} -// NCRDIV: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} -// NCRDIV: [[META6]] = !{!"Simple C++ TBAA"} -// NCRDIV: [[LOOP7]] = distinct !{[[LOOP7]], [[META8:![0-9]+]], [[META9:![0-9]+]]} -// NCRDIV: [[META8]] = !{!"llvm.loop.mustprogress"} -// NCRDIV: [[META9]] = !{!"llvm.loop.unroll.disable"} -// NCRDIV: [[LOOP10]] = distinct !{[[LOOP10]], [[META8]], [[META9]]} -// NCRDIV: [[LOOP11]] = distinct !{[[LOOP11]], [[META8]], [[META9]]} -// NCRDIV: [[META12]] = !{float 2.500000e+00} -// NCRDIV: [[INT_TBAA13]] = !{[[META14:![0-9]+]], [[META14]], i64 0} -// NCRDIV: [[META14]] = !{!"int", [[META5]], i64 0} -// NCRDIV: [[LOOP15]] = distinct !{[[LOOP15]], [[META8]], [[META9]]} -// NCRDIV: [[LOOP16]] = distinct !{[[LOOP16]], [[META8]], [[META9]]} +// NCRDIV: [[INT_TBAA4]] = !{[[META5:![0-9]+]], [[META5]], i64 0} +// NCRDIV: [[META5]] = !{!"int", [[META6:![0-9]+]], i64 0} +// NCRDIV: [[META6]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} +// NCRDIV: [[META7]] = !{!"Simple C++ TBAA"} +// NCRDIV: [[CHAR_TBAA8]] = !{[[META6]], [[META6]], i64 0} +// NCRDIV: [[LOOP9]] = distinct !{[[LOOP9]], [[META10:![0-9]+]], [[META11:![0-9]+]]} +// NCRDIV: [[META10]] = !{!"llvm.loop.mustprogress"} +// NCRDIV: [[META11]] = !{!"llvm.loop.unroll.disable"} +// NCRDIV: [[LOOP12]] = distinct !{[[LOOP12]], [[META10]], [[META11]]} +// NCRDIV: [[LOOP13]] = distinct !{[[LOOP13]], [[META10]], [[META11]]} +// NCRDIV: [[META14]] = !{float 2.500000e+00} +// NCRDIV: [[LOOP15]] = distinct !{[[LOOP15]], [[META10]], [[META11]]} +// NCRDIV: [[LOOP16]] = distinct !{[[LOOP16]], [[META10]], [[META11]]} // NCRDIV: [[FLOAT_TBAA17]] = !{[[META18:![0-9]+]], [[META18]], i64 0} -// NCRDIV: [[META18]] = !{!"float", [[META5]], i64 0} +// NCRDIV: [[META18]] = !{!"float", [[META6]], i64 0} // NCRDIV: [[DOUBLE_TBAA19]] = !{[[META20:![0-9]+]], [[META20]], i64 0} -// NCRDIV: [[META20]] = !{!"double", [[META5]], i64 0} -// NCRDIV: [[LOOP21]] = distinct !{[[LOOP21]], [[META8]], [[META9]]} -// NCRDIV: [[LOOP22]] = distinct !{[[LOOP22]], [[META8]], [[META9]]} -// NCRDIV: [[LOOP23]] = distinct !{[[LOOP23]], [[META8]], [[META9]]} -// NCRDIV: [[LOOP24]] = distinct !{[[LOOP24]], [[META8]], [[META9]]} +// NCRDIV: [[META20]] = !{!"double", [[META6]], i64 0} +// NCRDIV: [[LOOP21]] = distinct !{[[LOOP21]], [[META10]], [[META11]]} +// NCRDIV: [[LOOP22]] = distinct !{[[LOOP22]], [[META10]], [[META11]]} +// NCRDIV: [[LOOP23]] = distinct !{[[LOOP23]], [[META10]], [[META11]]} +// NCRDIV: [[LOOP24]] = distinct !{[[LOOP24]], [[META10]], [[META11]]} // NCRDIV: [[META25]] = !{float 3.000000e+00} -// NCRDIV: [[LOOP26]] = distinct !{[[LOOP26]], [[META8]], [[META9]]} -// NCRDIV: [[LOOP27]] = distinct !{[[LOOP27]], [[META8]], [[META9]]} +// NCRDIV: [[LOOP26]] = distinct !{[[LOOP26]], [[META10]], [[META11]]} +// NCRDIV: [[LOOP27]] = distinct !{[[LOOP27]], [[META10]], [[META11]]} //. -// AMDGCNSPIRV: [[CHAR_TBAA5]] = !{[[META6:![0-9]+]], [[META6]], i64 0} -// AMDGCNSPIRV: [[META6]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0} -// AMDGCNSPIRV: [[META7]] = !{!"Simple C++ TBAA"} -// AMDGCNSPIRV: [[LOOP8]] = distinct !{[[LOOP8]], [[META9:![0-9]+]], [[META10:![0-9]+]]} -// AMDGCNSPIRV: [[META9]] = !{!"llvm.loop.mustprogress"} -// AMDGCNSPIRV: [[META10]] = !{!"llvm.loop.unroll.disable"} -// AMDGCNSPIRV: [[LOOP11]] = distinct !{[[LOOP11]], [[META9]], [[META10]]} -// AMDGCNSPIRV: [[LOOP12]] = distinct !{[[LOOP12]], [[META9]], [[META10]]} -// AMDGCNSPIRV: [[INT_TBAA13]] = !{[[META14:![0-9]+]], [[META14]], i64 0} -// AMDGCNSPIRV: [[META14]] = !{!"int", [[META6]], i64 0} -// AMDGCNSPIRV: [[LOOP15]] = distinct !{[[LOOP15]], [[META9]], [[META10]]} -// AMDGCNSPIRV: [[LOOP16]] = distinct !{[[LOOP16]], [[META9]], [[META10]]} +// AMDGCNSPIRV: [[INT_TBAA5]] = !{[[META6:![0-9]+]], [[META6]], i64 0} +// AMDGCNSPIRV: [[META6]] = !{!"int", [[META7:![0-9]+]], i64 0} +// AMDGCNSPIRV: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0} +// AMDGCNSPIRV: [[META8]] = !{!"Simple C++ TBAA"} +// AMDGCNSPIRV: [[CHAR_TBAA9]] = !{[[META7]], [[META7]], i64 0} +// AMDGCNSPIRV: [[LOOP10]] = distinct !{[[LOOP10]], [[META11:![0-9]+]], [[META12:![0-9]+]]} +// AMDGCNSPIRV: [[META11]] = !{!"llvm.loop.mustprogress"} +// AMDGCNSPIRV: [[META12]] = !{!"llvm.loop.unroll.disable"} +// AMDGCNSPIRV: [[LOOP13]] = distinct !{[[LOOP13]], [[META11]], [[META12]]} +// AMDGCNSPIRV: [[LOOP14]] = distinct !{[[LOOP14]], [[META11]], [[META12]]} +// AMDGCNSPIRV: [[LOOP15]] = distinct !{[[LOOP15]], [[META11]], [[META12]]} +// AMDGCNSPIRV: [[LOOP16]] = distinct !{[[LOOP16]], [[META11]], [[META12]]} // AMDGCNSPIRV: [[FLOAT_TBAA17]] = !{[[META18:![0-9]+]], [[META18]], i64 0} -// AMDGCNSPIRV: [[META18]] = !{!"float", [[META6]], i64 0} +// AMDGCNSPIRV: [[META18]] = !{!"float", [[META7]], i64 0} // AMDGCNSPIRV: [[DOUBLE_TBAA19]] = !{[[META20:![0-9]+]], [[META20]], i64 0} -// AMDGCNSPIRV: [[META20]] = !{!"double", [[META6]], i64 0} -// AMDGCNSPIRV: [[LOOP21]] = distinct !{[[LOOP21]], [[META9]], [[META10]]} -// AMDGCNSPIRV: [[LOOP22]] = distinct !{[[LOOP22]], [[META9]], [[META10]]} -// AMDGCNSPIRV: [[LOOP23]] = distinct !{[[LOOP23]], [[META9]], [[META10]]} -// AMDGCNSPIRV: [[LOOP24]] = distinct !{[[LOOP24]], [[META9]], [[META10]]} -// AMDGCNSPIRV: [[LOOP25]] = distinct !{[[LOOP25]], [[META9]], [[META10]]} -// AMDGCNSPIRV: [[LOOP26]] = distinct !{[[LOOP26]], [[META9]], [[META10]]} +// AMDGCNSPIRV: [[META20]] = !{!"double", [[META7]], i64 0} +// AMDGCNSPIRV: [[LOOP21]] = distinct !{[[LOOP21]], [[META11]], [[META12]]} +// AMDGCNSPIRV: [[LOOP22]] = distinct !{[[LOOP22]], [[META11]], [[META12]]} +// AMDGCNSPIRV: [[LOOP23]] = distinct !{[[LOOP23]], [[META11]], [[META12]]} +// AMDGCNSPIRV: [[LOOP24]] = distinct !{[[LOOP24]], [[META11]], [[META12]]} +// AMDGCNSPIRV: [[LOOP25]] = distinct !{[[LOOP25]], [[META11]], [[META12]]} +// AMDGCNSPIRV: [[LOOP26]] = distinct !{[[LOOP26]], [[META11]], [[META12]]} //. diff --git a/clang/test/Headers/wasm.c b/clang/test/Headers/wasm.c index 03f20c5..2545a01 100644 --- a/clang/test/Headers/wasm.c +++ b/clang/test/Headers/wasm.c @@ -10,7 +10,7 @@ // CHECK-LABEL: define hidden <4 x i32> @test_v128_load( // CHECK-SAME: ptr noundef readonly captures(none) [[MEM:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <4 x i32>, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA2:![0-9]+]] +// CHECK-NEXT: [[TMP0:%.*]] = load <4 x i32>, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA6:![0-9]+]] // CHECK-NEXT: ret <4 x i32> [[TMP0]] // v128_t test_v128_load(const void *mem) { @@ -20,7 +20,7 @@ v128_t test_v128_load(const void *mem) { // CHECK-LABEL: define hidden <4 x i32> @test_v128_load8_splat( // CHECK-SAME: ptr noundef readonly captures(none) [[MEM:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[VECINIT_I:%.*]] = insertelement <16 x i8> poison, i8 [[TMP0]], i64 0 // CHECK-NEXT: [[VECINIT16_I:%.*]] = shufflevector <16 x i8> [[VECINIT_I]], <16 x i8> poison, <16 x i32> zeroinitializer // CHECK-NEXT: [[TMP1:%.*]] = bitcast <16 x i8> [[VECINIT16_I]] to <4 x i32> @@ -33,7 +33,7 @@ v128_t test_v128_load8_splat(const void *mem) { // CHECK-LABEL: define hidden <4 x i32> @test_v128_load16_splat( // CHECK-SAME: ptr noundef readonly captures(none) [[MEM:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i16, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i16, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[VECINIT_I:%.*]] = insertelement <8 x i16> poison, i16 [[TMP0]], i64 0 // CHECK-NEXT: [[VECINIT8_I:%.*]] = shufflevector <8 x i16> [[VECINIT_I]], <8 x i16> poison, <8 x i32> zeroinitializer // CHECK-NEXT: [[TMP1:%.*]] = bitcast <8 x i16> [[VECINIT8_I]] to <4 x i32> @@ -46,7 +46,7 @@ v128_t test_v128_load16_splat(const void *mem) { // CHECK-LABEL: define hidden <4 x i32> @test_v128_load32_splat( // CHECK-SAME: ptr noundef readonly captures(none) [[MEM:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[VECINIT_I:%.*]] = insertelement <4 x i32> poison, i32 [[TMP0]], i64 0 // CHECK-NEXT: [[VECINIT4_I:%.*]] = shufflevector <4 x i32> [[VECINIT_I]], <4 x i32> poison, <4 x i32> zeroinitializer // CHECK-NEXT: ret <4 x i32> [[VECINIT4_I]] @@ -58,7 +58,7 @@ v128_t test_v128_load32_splat(const void *mem) { // CHECK-LABEL: define hidden <4 x i32> @test_v128_load64_splat( // CHECK-SAME: ptr noundef readonly captures(none) [[MEM:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[VECINIT_I:%.*]] = insertelement <2 x i64> poison, i64 [[TMP0]], i64 0 // CHECK-NEXT: [[VECINIT2_I:%.*]] = shufflevector <2 x i64> [[VECINIT_I]], <2 x i64> poison, <2 x i32> zeroinitializer // CHECK-NEXT: [[TMP1:%.*]] = bitcast <2 x i64> [[VECINIT2_I]] to <4 x i32> @@ -71,7 +71,7 @@ v128_t test_v128_load64_splat(const void *mem) { // CHECK-LABEL: define hidden <4 x i32> @test_i16x8_load8x8( // CHECK-SAME: ptr noundef readonly captures(none) [[MEM:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[CONV_I:%.*]] = sext <8 x i8> [[TMP0]] to <8 x i16> // CHECK-NEXT: [[TMP1:%.*]] = bitcast <8 x i16> [[CONV_I]] to <4 x i32> // CHECK-NEXT: ret <4 x i32> [[TMP1]] @@ -83,7 +83,7 @@ v128_t test_i16x8_load8x8(const void *mem) { // CHECK-LABEL: define hidden <4 x i32> @test_u16x8_load8x8( // CHECK-SAME: ptr noundef readonly captures(none) [[MEM:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[CONV_I:%.*]] = zext <8 x i8> [[TMP0]] to <8 x i16> // CHECK-NEXT: [[TMP1:%.*]] = bitcast <8 x i16> [[CONV_I]] to <4 x i32> // CHECK-NEXT: ret <4 x i32> [[TMP1]] @@ -95,7 +95,7 @@ v128_t test_u16x8_load8x8(const void *mem) { // CHECK-LABEL: define hidden range(i32 -32768, 32768) <4 x i32> @test_i32x4_load16x4( // CHECK-SAME: ptr noundef readonly captures(none) [[MEM:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <4 x i16>, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <4 x i16>, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[CONV_I:%.*]] = sext <4 x i16> [[TMP0]] to <4 x i32> // CHECK-NEXT: ret <4 x i32> [[CONV_I]] // @@ -106,7 +106,7 @@ v128_t test_i32x4_load16x4(const void *mem) { // CHECK-LABEL: define hidden range(i32 0, 65536) <4 x i32> @test_u32x4_load16x4( // CHECK-SAME: ptr noundef readonly captures(none) [[MEM:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <4 x i16>, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <4 x i16>, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[CONV_I:%.*]] = zext <4 x i16> [[TMP0]] to <4 x i32> // CHECK-NEXT: ret <4 x i32> [[CONV_I]] // @@ -117,7 +117,7 @@ v128_t test_u32x4_load16x4(const void *mem) { // CHECK-LABEL: define hidden <4 x i32> @test_i64x2_load32x2( // CHECK-SAME: ptr noundef readonly captures(none) [[MEM:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <2 x i32>, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <2 x i32>, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[CONV_I:%.*]] = sext <2 x i32> [[TMP0]] to <2 x i64> // CHECK-NEXT: [[TMP1:%.*]] = bitcast <2 x i64> [[CONV_I]] to <4 x i32> // CHECK-NEXT: ret <4 x i32> [[TMP1]] @@ -129,7 +129,7 @@ v128_t test_i64x2_load32x2(const void *mem) { // CHECK-LABEL: define hidden <4 x i32> @test_u64x2_load32x2( // CHECK-SAME: ptr noundef readonly captures(none) [[MEM:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load <2 x i32>, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load <2 x i32>, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[CONV_I:%.*]] = zext <2 x i32> [[TMP0]] to <2 x i64> // CHECK-NEXT: [[TMP1:%.*]] = bitcast <2 x i64> [[CONV_I]] to <4 x i32> // CHECK-NEXT: ret <4 x i32> [[TMP1]] @@ -141,7 +141,7 @@ v128_t test_u64x2_load32x2(const void *mem) { // CHECK-LABEL: define hidden <4 x i32> @test_v128_load32_zero( // CHECK-SAME: ptr noundef readonly captures(none) [[MEM:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[VECINIT4_I:%.*]] = insertelement <4 x i32> <i32 poison, i32 0, i32 0, i32 0>, i32 [[TMP0]], i64 0 // CHECK-NEXT: ret <4 x i32> [[VECINIT4_I]] // @@ -152,7 +152,7 @@ v128_t test_v128_load32_zero(const void *mem) { // CHECK-LABEL: define hidden <4 x i32> @test_v128_load64_zero( // CHECK-SAME: ptr noundef readonly captures(none) [[MEM:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[VECINIT2_I:%.*]] = insertelement <2 x i64> <i64 poison, i64 0>, i64 [[TMP0]], i64 0 // CHECK-NEXT: [[TMP1:%.*]] = bitcast <2 x i64> [[VECINIT2_I]] to <4 x i32> // CHECK-NEXT: ret <4 x i32> [[TMP1]] @@ -164,7 +164,7 @@ v128_t test_v128_load64_zero(const void *mem) { // CHECK-LABEL: define hidden <4 x i32> @test_v128_load8_lane( // CHECK-SAME: ptr noundef readonly captures(none) [[PTR:%.*]], <4 x i32> noundef [[VEC:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = bitcast <4 x i32> [[VEC]] to <16 x i8> // CHECK-NEXT: [[VECINS_I:%.*]] = insertelement <16 x i8> [[TMP1]], i8 [[TMP0]], i64 15 // CHECK-NEXT: [[TMP2:%.*]] = bitcast <16 x i8> [[VECINS_I]] to <4 x i32> @@ -177,7 +177,7 @@ v128_t test_v128_load8_lane(const uint8_t *ptr, v128_t vec) { // CHECK-LABEL: define hidden <4 x i32> @test_v128_load16_lane( // CHECK-SAME: ptr noundef readonly captures(none) [[PTR:%.*]], <4 x i32> noundef [[VEC:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i16, ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i16, ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = bitcast <4 x i32> [[VEC]] to <8 x i16> // CHECK-NEXT: [[VECINS_I:%.*]] = insertelement <8 x i16> [[TMP1]], i16 [[TMP0]], i64 7 // CHECK-NEXT: [[TMP2:%.*]] = bitcast <8 x i16> [[VECINS_I]] to <4 x i32> @@ -190,7 +190,7 @@ v128_t test_v128_load16_lane(const uint16_t *ptr, v128_t vec) { // CHECK-LABEL: define hidden <4 x i32> @test_v128_load32_lane( // CHECK-SAME: ptr noundef readonly captures(none) [[PTR:%.*]], <4 x i32> noundef [[VEC:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[VECINS_I:%.*]] = insertelement <4 x i32> [[VEC]], i32 [[TMP0]], i64 3 // CHECK-NEXT: ret <4 x i32> [[VECINS_I]] // @@ -201,7 +201,7 @@ v128_t test_v128_load32_lane(const uint32_t *ptr, v128_t vec) { // CHECK-LABEL: define hidden <4 x i32> @test_v128_load64_lane( // CHECK-SAME: ptr noundef readonly captures(none) [[PTR:%.*]], <4 x i32> noundef [[VEC:%.*]]) local_unnamed_addr #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: [[TMP1:%.*]] = bitcast <4 x i32> [[VEC]] to <2 x i64> // CHECK-NEXT: [[VECINS_I:%.*]] = insertelement <2 x i64> [[TMP1]], i64 [[TMP0]], i64 1 // CHECK-NEXT: [[TMP2:%.*]] = bitcast <2 x i64> [[VECINS_I]] to <4 x i32> @@ -214,7 +214,7 @@ v128_t test_v128_load64_lane(const uint64_t *ptr, v128_t vec) { // CHECK-LABEL: define hidden void @test_v128_store( // CHECK-SAME: ptr noundef writeonly captures(none) initializes((0, 16)) [[MEM:%.*]], <4 x i32> noundef [[A:%.*]]) local_unnamed_addr #[[ATTR1:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: store <4 x i32> [[A]], ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store <4 x i32> [[A]], ptr [[MEM]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // void test_v128_store(void *mem, v128_t a) { @@ -226,7 +226,7 @@ void test_v128_store(void *mem, v128_t a) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = bitcast <4 x i32> [[VEC]] to <16 x i8> // CHECK-NEXT: [[VECEXT_I:%.*]] = extractelement <16 x i8> [[TMP0]], i64 15 -// CHECK-NEXT: store i8 [[VECEXT_I]], ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store i8 [[VECEXT_I]], ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // void test_v128_store8_lane(uint8_t *ptr, v128_t vec) { @@ -238,7 +238,7 @@ void test_v128_store8_lane(uint8_t *ptr, v128_t vec) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = bitcast <4 x i32> [[VEC]] to <8 x i16> // CHECK-NEXT: [[VECEXT_I:%.*]] = extractelement <8 x i16> [[TMP0]], i64 7 -// CHECK-NEXT: store i16 [[VECEXT_I]], ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store i16 [[VECEXT_I]], ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // void test_v128_store16_lane(uint16_t *ptr, v128_t vec) { @@ -249,7 +249,7 @@ void test_v128_store16_lane(uint16_t *ptr, v128_t vec) { // CHECK-SAME: ptr noundef writeonly captures(none) initializes((0, 4)) [[PTR:%.*]], <4 x i32> noundef [[VEC:%.*]]) local_unnamed_addr #[[ATTR1]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[VECEXT_I:%.*]] = extractelement <4 x i32> [[VEC]], i64 3 -// CHECK-NEXT: store i32 [[VECEXT_I]], ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store i32 [[VECEXT_I]], ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // void test_v128_store32_lane(uint32_t *ptr, v128_t vec) { @@ -261,7 +261,7 @@ void test_v128_store32_lane(uint32_t *ptr, v128_t vec) { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[TMP0:%.*]] = bitcast <4 x i32> [[VEC]] to <2 x i64> // CHECK-NEXT: [[VECEXT_I:%.*]] = extractelement <2 x i64> [[TMP0]], i64 1 -// CHECK-NEXT: store i64 [[VECEXT_I]], ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA2]] +// CHECK-NEXT: store i64 [[VECEXT_I]], ptr [[PTR]], align 1, !tbaa [[CHAR_TBAA6]] // CHECK-NEXT: ret void // void test_v128_store64_lane(uint64_t *ptr, v128_t vec) { @@ -3466,7 +3466,7 @@ v128_t test_i16x8_q15mulr_sat(v128_t a, v128_t b) { return wasm_i16x8_q15mulr_sat(a, b); } //. -// CHECK: [[CHAR_TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0} -// CHECK: [[META3]] = !{!"omnipotent char", [[META4:![0-9]+]], i64 0} -// CHECK: [[META4]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[META4:![0-9]+]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0} +// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"} +// CHECK: [[CHAR_TBAA6]] = !{[[META4]], [[META4]], i64 0} //. diff --git a/clang/test/Interpreter/pretty-print.c b/clang/test/Interpreter/pretty-print.c index 588df70..d0712fb 100644 --- a/clang/test/Interpreter/pretty-print.c +++ b/clang/test/Interpreter/pretty-print.c @@ -75,9 +75,10 @@ int * ptr = (int*)0x123; ptr int * null_ptr = (int*)0; null_ptr // CHECK-NEXT: (int *) 0x0 +union U { int I; float F; } u; u.I = 12; u.I +// CHECK-NEXT: (int) 12 + // TODO: _Bool, _Complex, _Atomic, and _BitInt -// union U { int I; float F; } u; u.I = 12; u.I -// TODO-CHECK-NEXT: (int) 12 // struct S1{} s1; s1 // TODO-CHECK-NEXT: (S1 &) @0x{{[0-9a-f]+}} @@ -86,4 +87,21 @@ int * null_ptr = (int*)0; null_ptr // E.d // TODO-CHECK-NEXT: (int) 22 +// ----------------------------------------------------------------------------- +// Tentative definition handling (C99 6.9.2) +// Verify that multiple distinct tentative definitions across inputs no longer +// conflict. Each variable should emit correctly in its own incremental module. +// ----------------------------------------------------------------------------- + +int t1; +int t2; +int t3; +t1 = 1; t2 = 2; t3 = 3; +t1 + t2 + t3 +// CHECK-NEXT: (int) 6 + +// A redefinition of an existing tentative variable should still fail. +int t1; +// expected-error {{duplicate definition of symbol '_t1'}} + %quit diff --git a/clang/test/OpenMP/bug54082.c b/clang/test/OpenMP/bug54082.c index ef3e715..1a98e12 100644 --- a/clang/test/OpenMP/bug54082.c +++ b/clang/test/OpenMP/bug54082.c @@ -72,7 +72,7 @@ void foo() { // CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 16 dereferenceable(16) [[X_TRAITS]], ptr noundef nonnull align 16 dereferenceable(16) @__const.foo.x_traits, i64 16, i1 false) // CHECK-NEXT: call void @llvm.lifetime.start.p0(ptr nonnull [[X_ALLOC]]) #[[ATTR5]] // CHECK-NEXT: [[CALL:%.*]] = call i64 @omp_init_allocator(i64 noundef 0, i32 noundef 1, ptr noundef nonnull [[X_TRAITS]]) #[[ATTR5]] -// CHECK-NEXT: store i64 [[CALL]], ptr [[X_ALLOC]], align 8, !tbaa [[LONG_TBAA3:![0-9]+]] +// CHECK-NEXT: store i64 [[CALL]], ptr [[X_ALLOC]], align 8, !tbaa [[LONG_TBAA7:![0-9]+]] // CHECK-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr nonnull @[[GLOB2:[0-9]+]], i32 1, ptr nonnull @foo.omp_outlined, ptr nonnull [[X_ALLOC]]) // CHECK-NEXT: call void @llvm.lifetime.end.p0(ptr nonnull [[X_ALLOC]]) #[[ATTR5]] // CHECK-NEXT: call void @llvm.lifetime.end.p0(ptr nonnull [[X_TRAITS]]) #[[ATTR5]] @@ -87,23 +87,23 @@ void foo() { // CHECK-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 // CHECK-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK-NEXT: call void @llvm.lifetime.start.p0(ptr nonnull [[DOTOMP_LB]]) #[[ATTR5]] -// CHECK-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA7:![0-9]+]] +// CHECK-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA3:![0-9]+]] // CHECK-NEXT: call void @llvm.lifetime.start.p0(ptr nonnull [[DOTOMP_UB]]) #[[ATTR5]] -// CHECK-NEXT: store i32 1023, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA7]] +// CHECK-NEXT: store i32 1023, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA3]] // CHECK-NEXT: call void @llvm.lifetime.start.p0(ptr nonnull [[DOTOMP_STRIDE]]) #[[ATTR5]] -// CHECK-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA7]] +// CHECK-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA3]] // CHECK-NEXT: call void @llvm.lifetime.start.p0(ptr nonnull [[DOTOMP_IS_LAST]]) #[[ATTR5]] -// CHECK-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4, !tbaa [[INT_TBAA7]] -// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTGLOBAL_TID_]], align 4, !tbaa [[INT_TBAA7]] -// CHECK-NEXT: [[TMP1:%.*]] = load i64, ptr [[X_ALLOC]], align 8, !tbaa [[LONG_TBAA3]] +// CHECK-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4, !tbaa [[INT_TBAA3]] +// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTGLOBAL_TID_]], align 4, !tbaa [[INT_TBAA3]] +// CHECK-NEXT: [[TMP1:%.*]] = load i64, ptr [[X_ALLOC]], align 8, !tbaa [[LONG_TBAA7]] // CHECK-NEXT: [[CONV:%.*]] = inttoptr i64 [[TMP1]] to ptr // CHECK-NEXT: [[DOTX__VOID_ADDR:%.*]] = tail call ptr @__kmpc_alloc(i32 [[TMP0]], i64 8, ptr [[CONV]]) // CHECK-NEXT: call void @__kmpc_for_static_init_4(ptr nonnull @[[GLOB1:[0-9]+]], i32 [[TMP0]], i32 34, ptr nonnull [[DOTOMP_IS_LAST]], ptr nonnull [[DOTOMP_LB]], ptr nonnull [[DOTOMP_UB]], ptr nonnull [[DOTOMP_STRIDE]], i32 1, i32 1) -// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA7]] +// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA3]] // CHECK-NEXT: [[COND:%.*]] = call i32 @llvm.smin.i32(i32 [[TMP2]], i32 1023) -// CHECK-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA7]] +// CHECK-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA3]] // CHECK-NEXT: call void @__kmpc_for_static_fini(ptr nonnull @[[GLOB1]], i32 [[TMP0]]) -// CHECK-NEXT: [[TMP3:%.*]] = load i64, ptr [[X_ALLOC]], align 8, !tbaa [[LONG_TBAA3]] +// CHECK-NEXT: [[TMP3:%.*]] = load i64, ptr [[X_ALLOC]], align 8, !tbaa [[LONG_TBAA7]] // CHECK-NEXT: [[CONV5:%.*]] = inttoptr i64 [[TMP3]] to ptr // CHECK-NEXT: call void @__kmpc_free(i32 [[TMP0]], ptr [[DOTX__VOID_ADDR]], ptr [[CONV5]]) // CHECK-NEXT: call void @llvm.lifetime.end.p0(ptr nonnull [[DOTOMP_IS_LAST]]) #[[ATTR5]] @@ -113,10 +113,10 @@ void foo() { // CHECK-NEXT: ret void // //. -// CHECK: [[LONG_TBAA3]] = !{[[META4:![0-9]+]], [[META4]], i64 0} -// CHECK: [[META4]] = !{!"long", [[META5:![0-9]+]], i64 0} +// CHECK: [[INT_TBAA3]] = !{[[META4:![0-9]+]], [[META4]], i64 0} +// CHECK: [[META4]] = !{!"int", [[META5:![0-9]+]], i64 0} // CHECK: [[META5]] = !{!"omnipotent char", [[META6:![0-9]+]], i64 0} // CHECK: [[META6]] = !{!"Simple C/C++ TBAA"} -// CHECK: [[INT_TBAA7]] = !{[[META8:![0-9]+]], [[META8]], i64 0} -// CHECK: [[META8]] = !{!"int", [[META5]], i64 0} +// CHECK: [[LONG_TBAA7]] = !{[[META8:![0-9]+]], [[META8]], i64 0} +// CHECK: [[META8]] = !{!"long", [[META5]], i64 0} //. diff --git a/clang/test/OpenMP/nvptx_target_parallel_reduction_codegen_tbaa_PR46146.cpp b/clang/test/OpenMP/nvptx_target_parallel_reduction_codegen_tbaa_PR46146.cpp index 46c87eb..4b99d77 100644 --- a/clang/test/OpenMP/nvptx_target_parallel_reduction_codegen_tbaa_PR46146.cpp +++ b/clang/test/OpenMP/nvptx_target_parallel_reduction_codegen_tbaa_PR46146.cpp @@ -36,14 +36,14 @@ void test() { // CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 -// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8, !tbaa [[ANYPTR_TBAA6:![0-9]+]] +// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8, !tbaa [[ANYPTR_TBAA10:![0-9]+]] // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIfEvv_l16_kernel_environment, ptr [[DYN_PTR]]) // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label %[[USER_CODE_ENTRY:.*]], label %[[WORKER_EXIT:.*]] // CHECK1: [[USER_CODE_ENTRY]]: // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1:[0-9]+]]) // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 -// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4, !tbaa [[INT_TBAA10:![0-9]+]] +// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4, !tbaa [[INT_TBAA6:![0-9]+]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIfEvv_l16_omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]]) #[[ATTR4:[0-9]+]] // CHECK1-NEXT: call void @__kmpc_target_deinit() // CHECK1-NEXT: ret void @@ -73,43 +73,43 @@ void test() { // CHECK1-NEXT: [[PARTIAL_SUM:%.*]] = call align 16 ptr @__kmpc_alloc_shared(i64 8) // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_IV]]) #[[ATTR4]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_LB]]) #[[ATTR4]] -// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_UB]]) #[[ATTR4]] -// CHECK1-NEXT: store i32 99, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 99, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_STRIDE]]) #[[ATTR4]] -// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_IS_LAST]]) #[[ATTR4]] -// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[IB]]) #[[ATTR4]] // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 -// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) -// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK1-NEXT: br i1 [[CMP]], label %[[COND_TRUE:.*]], label %[[COND_FALSE:.*]] // CHECK1: [[COND_TRUE]]: // CHECK1-NEXT: br label %[[COND_END:.*]] // CHECK1: [[COND_FALSE]]: -// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: br label %[[COND_END]] // CHECK1: [[COND_END]]: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, %[[COND_TRUE]] ], [ [[TMP3]], %[[COND_FALSE]] ] -// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: br label %[[OMP_INNER_FOR_COND:.*]] // CHECK1: [[OMP_INNER_FOR_COND]]: -// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK1-NEXT: br i1 [[CMP1]], label %[[OMP_INNER_FOR_BODY:.*]], label %[[OMP_INNER_FOR_COND_CLEANUP:.*]] // CHECK1: [[OMP_INNER_FOR_COND_CLEANUP]]: // CHECK1-NEXT: br label %[[OMP_INNER_FOR_END:.*]] // CHECK1: [[OMP_INNER_FOR_BODY]]: -// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] -// CHECK1-NEXT: store i32 [[ADD]], ptr [[IB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[ADD]], ptr [[IB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[REF_TMP]]) #[[ATTR4]] // CHECK1-NEXT: store float 0.000000e+00, ptr [[REF_TMP]], align 4, !tbaa [[FLOAT_TBAA14:![0-9]+]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[REF_TMP2]]) #[[ATTR4]] @@ -117,13 +117,13 @@ void test() { // CHECK1-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(ptr nonnull align 4 dereferenceable(8) [[PARTIAL_SUM]], ptr nonnull align 4 dereferenceable(4) [[REF_TMP]], ptr nonnull align 4 dereferenceable(4) [[REF_TMP2]]) #[[ATTR11:[0-9]+]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0(ptr [[REF_TMP2]]) #[[ATTR4]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0(ptr [[REF_TMP]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[IB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[IB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[MUL3:%.*]] = mul nsw i32 [[TMP8]], 4 -// CHECK1-NEXT: store i32 [[MUL3]], ptr [[ISTART]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[IB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[MUL3]], ptr [[ISTART]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[IB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[MUL5:%.*]] = mul nsw i32 [[ADD4]], 4 -// CHECK1-NEXT: store i32 [[MUL5]], ptr [[IEND]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[MUL5]], ptr [[IEND]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: store ptr [[ISTART]], ptr [[TMP10]], align 8, !tbaa [[ANYPTR_TBAA16:![0-9]+]] // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 @@ -135,9 +135,9 @@ void test() { // CHECK1: [[OMP_BODY_CONTINUE]]: // CHECK1-NEXT: br label %[[OMP_INNER_FOR_INC:.*]] // CHECK1: [[OMP_INNER_FOR_INC]]: -// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP13]], 1 -// CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: br label %[[OMP_INNER_FOR_COND]] // CHECK1: [[OMP_INNER_FOR_END]]: // CHECK1-NEXT: br label %[[OMP_LOOP_EXIT:.*]] @@ -207,38 +207,38 @@ void test() { // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[PARTIAL_SUM_ADDR]], align 8, !tbaa [[_ZTSST7COMPLEXIFEPTR_TBAA18]], !nonnull [[META22]], !align [[META23]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_IV]]) #[[ATTR4]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTCAPTURE_EXPR_]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTCAPTURE_EXPR_1]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_1]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_1]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTCAPTURE_EXPR_2]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[SUB:%.*]] = sub i32 [[TMP5]], [[TMP6]] // CHECK1-NEXT: [[SUB3:%.*]] = sub i32 [[SUB]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add i32 [[SUB3]], 1 // CHECK1-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK1-NEXT: [[SUB4:%.*]] = sub i32 [[DIV]], 1 -// CHECK1-NEXT: store i32 [[SUB4]], ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[SUB4]], ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[I]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: store i32 [[TMP7]], ptr [[I]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: store i32 [[TMP7]], ptr [[I]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0(ptr [[I]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 [[TMP8]], [[TMP9]] // CHECK1-NEXT: br i1 [[CMP]], label %[[OMP_PRECOND_THEN:.*]], label %[[OMP_PRECOND_END:.*]] // CHECK1: [[OMP_PRECOND_THEN]]: // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_LB]]) #[[ATTR4]] -// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_UB]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: store i32 [[TMP10]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: store i32 [[TMP10]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_STRIDE]]) #[[ATTR4]] -// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_IS_LAST]]) #[[ATTR4]] -// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[PARTIAL_SUM5]]) #[[ATTR4]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[REF_TMP]]) #[[ATTR4]] // CHECK1-NEXT: store float 0.000000e+00, ptr [[REF_TMP]], align 4, !tbaa [[FLOAT_TBAA14]] @@ -249,27 +249,27 @@ void test() { // CHECK1-NEXT: call void @llvm.lifetime.end.p0(ptr [[REF_TMP]]) #[[ATTR4]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[I7]]) #[[ATTR4]] // CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 -// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @__kmpc_for_static_init_4u(ptr @[[GLOB3:[0-9]+]], i32 [[TMP12]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: br label %[[OMP_DISPATCH_COND:.*]] // CHECK1: [[OMP_DISPATCH_COND]]: -// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[CMP8:%.*]] = icmp ugt i32 [[TMP13]], [[TMP14]] // CHECK1-NEXT: br i1 [[CMP8]], label %[[COND_TRUE:.*]], label %[[COND_FALSE:.*]] // CHECK1: [[COND_TRUE]]: -// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: br label %[[COND_END:.*]] // CHECK1: [[COND_FALSE]]: -// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: br label %[[COND_END]] // CHECK1: [[COND_END]]: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP15]], %[[COND_TRUE]] ], [ [[TMP16]], %[[COND_FALSE]] ] -// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: store i32 [[TMP17]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: store i32 [[TMP17]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[ADD9:%.*]] = add i32 [[TMP19]], 1 // CHECK1-NEXT: [[CMP10:%.*]] = icmp ult i32 [[TMP18]], [[ADD9]] // CHECK1-NEXT: br i1 [[CMP10]], label %[[OMP_DISPATCH_BODY:.*]], label %[[OMP_DISPATCH_CLEANUP:.*]] @@ -278,26 +278,26 @@ void test() { // CHECK1: [[OMP_DISPATCH_BODY]]: // CHECK1-NEXT: br label %[[OMP_INNER_FOR_COND:.*]] // CHECK1: [[OMP_INNER_FOR_COND]]: -// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[ADD11:%.*]] = add i32 [[TMP21]], 1 // CHECK1-NEXT: [[CMP12:%.*]] = icmp ult i32 [[TMP20]], [[ADD11]] // CHECK1-NEXT: br i1 [[CMP12]], label %[[OMP_INNER_FOR_BODY:.*]], label %[[OMP_INNER_FOR_COND_CLEANUP:.*]] // CHECK1: [[OMP_INNER_FOR_COND_CLEANUP]]: // CHECK1-NEXT: br label %[[OMP_INNER_FOR_END:.*]] // CHECK1: [[OMP_INNER_FOR_BODY]]: -// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[MUL:%.*]] = mul i32 [[TMP23]], 1 // CHECK1-NEXT: [[ADD13:%.*]] = add i32 [[TMP22]], [[MUL]] -// CHECK1-NEXT: store i32 [[ADD13]], ptr [[I7]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[ADD13]], ptr [[I7]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[REF_TMP14]]) #[[ATTR4]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[REF_TMP15]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[I7]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[I7]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP24]] to float // CHECK1-NEXT: store float [[CONV]], ptr [[REF_TMP15]], align 4, !tbaa [[FLOAT_TBAA14]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[REF_TMP16]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP25:%.*]] = load i32, ptr [[I7]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP25:%.*]] = load i32, ptr [[I7]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[CONV17:%.*]] = sitofp i32 [[TMP25]] to float // CHECK1-NEXT: store float [[CONV17]], ptr [[REF_TMP16]], align 4, !tbaa [[FLOAT_TBAA14]] // CHECK1-NEXT: call void @_ZNSt7complexIfEC1ERKfS2_(ptr nonnull align 4 dereferenceable(8) [[REF_TMP14]], ptr nonnull align 4 dereferenceable(4) [[REF_TMP15]], ptr nonnull align 4 dereferenceable(4) [[REF_TMP16]]) #[[ATTR11]] @@ -309,25 +309,25 @@ void test() { // CHECK1: [[OMP_BODY_CONTINUE]]: // CHECK1-NEXT: br label %[[OMP_INNER_FOR_INC:.*]] // CHECK1: [[OMP_INNER_FOR_INC]]: -// CHECK1-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[ADD18:%.*]] = add i32 [[TMP26]], 1 -// CHECK1-NEXT: store i32 [[ADD18]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[ADD18]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: br label %[[OMP_INNER_FOR_COND]] // CHECK1: [[OMP_INNER_FOR_END]]: // CHECK1-NEXT: br label %[[OMP_DISPATCH_INC:.*]] // CHECK1: [[OMP_DISPATCH_INC]]: -// CHECK1-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[ADD19:%.*]] = add i32 [[TMP27]], [[TMP28]] -// CHECK1-NEXT: store i32 [[ADD19]], ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[ADD19]], ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[ADD20:%.*]] = add i32 [[TMP29]], [[TMP30]] -// CHECK1-NEXT: store i32 [[ADD20]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[ADD20]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: br label %[[OMP_DISPATCH_COND]] // CHECK1: [[OMP_DISPATCH_END]]: // CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 -// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP32]]) // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: store ptr [[PARTIAL_SUM5]], ptr [[TMP33]], align 8 @@ -508,16 +508,16 @@ void test() { // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2, !tbaa [[SHORT_TBAA27:![0-9]+]] -// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0 -// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 8, !tbaa [[ANYPTR_TBAA6]] +// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 8, !tbaa [[ANYPTR_TBAA10]] // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 1 -// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8, !tbaa [[ANYPTR_TBAA6]] +// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8, !tbaa [[ANYPTR_TBAA10]] // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 2 -// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[TMP7]], align 8, !tbaa [[ANYPTR_TBAA6]] +// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[TMP7]], align 8, !tbaa [[ANYPTR_TBAA10]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIfEvv_l16_omp_outlined_omp_outlined(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]], ptr [[TMP6]], ptr [[TMP8]]) #[[ATTR4]] // CHECK1-NEXT: ret void // @@ -528,14 +528,14 @@ void test() { // CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 -// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8, !tbaa [[ANYPTR_TBAA6]] +// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8, !tbaa [[ANYPTR_TBAA10]] // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIdEvv_l16_kernel_environment, ptr [[DYN_PTR]]) // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label %[[USER_CODE_ENTRY:.*]], label %[[WORKER_EXIT:.*]] // CHECK1: [[USER_CODE_ENTRY]]: // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 -// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIdEvv_l16_omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]]) #[[ATTR4]] // CHECK1-NEXT: call void @__kmpc_target_deinit() // CHECK1-NEXT: ret void @@ -565,43 +565,43 @@ void test() { // CHECK1-NEXT: [[PARTIAL_SUM:%.*]] = call align 16 ptr @__kmpc_alloc_shared(i64 16) // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_IV]]) #[[ATTR4]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_LB]]) #[[ATTR4]] -// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_UB]]) #[[ATTR4]] -// CHECK1-NEXT: store i32 99, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 99, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_STRIDE]]) #[[ATTR4]] -// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_IS_LAST]]) #[[ATTR4]] -// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[IB]]) #[[ATTR4]] // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 -// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) -// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99 // CHECK1-NEXT: br i1 [[CMP]], label %[[COND_TRUE:.*]], label %[[COND_FALSE:.*]] // CHECK1: [[COND_TRUE]]: // CHECK1-NEXT: br label %[[COND_END:.*]] // CHECK1: [[COND_FALSE]]: -// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: br label %[[COND_END]] // CHECK1: [[COND_END]]: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, %[[COND_TRUE]] ], [ [[TMP3]], %[[COND_FALSE]] ] -// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: br label %[[OMP_INNER_FOR_COND:.*]] // CHECK1: [[OMP_INNER_FOR_COND]]: -// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK1-NEXT: br i1 [[CMP1]], label %[[OMP_INNER_FOR_BODY:.*]], label %[[OMP_INNER_FOR_COND_CLEANUP:.*]] // CHECK1: [[OMP_INNER_FOR_COND_CLEANUP]]: // CHECK1-NEXT: br label %[[OMP_INNER_FOR_END:.*]] // CHECK1: [[OMP_INNER_FOR_BODY]]: -// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] -// CHECK1-NEXT: store i32 [[ADD]], ptr [[IB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[ADD]], ptr [[IB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[REF_TMP]]) #[[ATTR4]] // CHECK1-NEXT: store double 0.000000e+00, ptr [[REF_TMP]], align 8, !tbaa [[DOUBLE_TBAA29:![0-9]+]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[REF_TMP2]]) #[[ATTR4]] @@ -609,13 +609,13 @@ void test() { // CHECK1-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(ptr nonnull align 8 dereferenceable(16) [[PARTIAL_SUM]], ptr nonnull align 8 dereferenceable(8) [[REF_TMP]], ptr nonnull align 8 dereferenceable(8) [[REF_TMP2]]) #[[ATTR11]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0(ptr [[REF_TMP2]]) #[[ATTR4]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0(ptr [[REF_TMP]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[IB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[IB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[MUL3:%.*]] = mul nsw i32 [[TMP8]], 4 -// CHECK1-NEXT: store i32 [[MUL3]], ptr [[ISTART]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[IB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[MUL3]], ptr [[ISTART]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[IB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[MUL5:%.*]] = mul nsw i32 [[ADD4]], 4 -// CHECK1-NEXT: store i32 [[MUL5]], ptr [[IEND]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[MUL5]], ptr [[IEND]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: store ptr [[ISTART]], ptr [[TMP10]], align 8, !tbaa [[ANYPTR_TBAA16]] // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 @@ -627,9 +627,9 @@ void test() { // CHECK1: [[OMP_BODY_CONTINUE]]: // CHECK1-NEXT: br label %[[OMP_INNER_FOR_INC:.*]] // CHECK1: [[OMP_INNER_FOR_INC]]: -// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP13]], 1 -// CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: br label %[[OMP_INNER_FOR_COND]] // CHECK1: [[OMP_INNER_FOR_END]]: // CHECK1-NEXT: br label %[[OMP_LOOP_EXIT:.*]] @@ -699,38 +699,38 @@ void test() { // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[PARTIAL_SUM_ADDR]], align 8, !tbaa [[_ZTSST7COMPLEXIDEPTR_TBAA31]], !nonnull [[META22]], !align [[META35:![0-9]+]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_IV]]) #[[ATTR4]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTCAPTURE_EXPR_]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTCAPTURE_EXPR_1]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_1]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_1]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTCAPTURE_EXPR_2]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[SUB:%.*]] = sub i32 [[TMP5]], [[TMP6]] // CHECK1-NEXT: [[SUB3:%.*]] = sub i32 [[SUB]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add i32 [[SUB3]], 1 // CHECK1-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK1-NEXT: [[SUB4:%.*]] = sub i32 [[DIV]], 1 -// CHECK1-NEXT: store i32 [[SUB4]], ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[SUB4]], ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[I]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: store i32 [[TMP7]], ptr [[I]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: store i32 [[TMP7]], ptr [[I]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.end.p0(ptr [[I]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 [[TMP8]], [[TMP9]] // CHECK1-NEXT: br i1 [[CMP]], label %[[OMP_PRECOND_THEN:.*]], label %[[OMP_PRECOND_END:.*]] // CHECK1: [[OMP_PRECOND_THEN]]: // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_LB]]) #[[ATTR4]] -// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_UB]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: store i32 [[TMP10]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: store i32 [[TMP10]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_STRIDE]]) #[[ATTR4]] -// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[DOTOMP_IS_LAST]]) #[[ATTR4]] -// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[PARTIAL_SUM5]]) #[[ATTR4]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[REF_TMP]]) #[[ATTR4]] // CHECK1-NEXT: store double 0.000000e+00, ptr [[REF_TMP]], align 8, !tbaa [[DOUBLE_TBAA29]] @@ -741,27 +741,27 @@ void test() { // CHECK1-NEXT: call void @llvm.lifetime.end.p0(ptr [[REF_TMP]]) #[[ATTR4]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[I7]]) #[[ATTR4]] // CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 -// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @__kmpc_for_static_init_4u(ptr @[[GLOB3]], i32 [[TMP12]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK1-NEXT: br label %[[OMP_DISPATCH_COND:.*]] // CHECK1: [[OMP_DISPATCH_COND]]: -// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[CMP8:%.*]] = icmp ugt i32 [[TMP13]], [[TMP14]] // CHECK1-NEXT: br i1 [[CMP8]], label %[[COND_TRUE:.*]], label %[[COND_FALSE:.*]] // CHECK1: [[COND_TRUE]]: -// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: br label %[[COND_END:.*]] // CHECK1: [[COND_FALSE]]: -// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: br label %[[COND_END]] // CHECK1: [[COND_END]]: // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP15]], %[[COND_TRUE]] ], [ [[TMP16]], %[[COND_FALSE]] ] -// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: store i32 [[TMP17]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: store i32 [[TMP17]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[ADD9:%.*]] = add i32 [[TMP19]], 1 // CHECK1-NEXT: [[CMP10:%.*]] = icmp ult i32 [[TMP18]], [[ADD9]] // CHECK1-NEXT: br i1 [[CMP10]], label %[[OMP_DISPATCH_BODY:.*]], label %[[OMP_DISPATCH_CLEANUP:.*]] @@ -770,26 +770,26 @@ void test() { // CHECK1: [[OMP_DISPATCH_BODY]]: // CHECK1-NEXT: br label %[[OMP_INNER_FOR_COND:.*]] // CHECK1: [[OMP_INNER_FOR_COND]]: -// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[ADD11:%.*]] = add i32 [[TMP21]], 1 // CHECK1-NEXT: [[CMP12:%.*]] = icmp ult i32 [[TMP20]], [[ADD11]] // CHECK1-NEXT: br i1 [[CMP12]], label %[[OMP_INNER_FOR_BODY:.*]], label %[[OMP_INNER_FOR_COND_CLEANUP:.*]] // CHECK1: [[OMP_INNER_FOR_COND_CLEANUP]]: // CHECK1-NEXT: br label %[[OMP_INNER_FOR_END:.*]] // CHECK1: [[OMP_INNER_FOR_BODY]]: -// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[MUL:%.*]] = mul i32 [[TMP23]], 1 // CHECK1-NEXT: [[ADD13:%.*]] = add i32 [[TMP22]], [[MUL]] -// CHECK1-NEXT: store i32 [[ADD13]], ptr [[I7]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[ADD13]], ptr [[I7]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[REF_TMP14]]) #[[ATTR4]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[REF_TMP15]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[I7]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[I7]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP24]] to double // CHECK1-NEXT: store double [[CONV]], ptr [[REF_TMP15]], align 8, !tbaa [[DOUBLE_TBAA29]] // CHECK1-NEXT: call void @llvm.lifetime.start.p0(ptr [[REF_TMP16]]) #[[ATTR4]] -// CHECK1-NEXT: [[TMP25:%.*]] = load i32, ptr [[I7]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP25:%.*]] = load i32, ptr [[I7]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[CONV17:%.*]] = sitofp i32 [[TMP25]] to double // CHECK1-NEXT: store double [[CONV17]], ptr [[REF_TMP16]], align 8, !tbaa [[DOUBLE_TBAA29]] // CHECK1-NEXT: call void @_ZNSt7complexIdEC1ERKdS2_(ptr nonnull align 8 dereferenceable(16) [[REF_TMP14]], ptr nonnull align 8 dereferenceable(8) [[REF_TMP15]], ptr nonnull align 8 dereferenceable(8) [[REF_TMP16]]) #[[ATTR11]] @@ -801,25 +801,25 @@ void test() { // CHECK1: [[OMP_BODY_CONTINUE]]: // CHECK1-NEXT: br label %[[OMP_INNER_FOR_INC:.*]] // CHECK1: [[OMP_INNER_FOR_INC]]: -// CHECK1-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[ADD18:%.*]] = add i32 [[TMP26]], 1 -// CHECK1-NEXT: store i32 [[ADD18]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[ADD18]], ptr [[DOTOMP_IV]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: br label %[[OMP_INNER_FOR_COND]] // CHECK1: [[OMP_INNER_FOR_END]]: // CHECK1-NEXT: br label %[[OMP_DISPATCH_INC:.*]] // CHECK1: [[OMP_DISPATCH_INC]]: -// CHECK1-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[ADD19:%.*]] = add i32 [[TMP27]], [[TMP28]] -// CHECK1-NEXT: store i32 [[ADD19]], ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] -// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[ADD19]], ptr [[DOTOMP_LB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] +// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: [[ADD20:%.*]] = add i32 [[TMP29]], [[TMP30]] -// CHECK1-NEXT: store i32 [[ADD20]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[ADD20]], ptr [[DOTOMP_UB]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: br label %[[OMP_DISPATCH_COND]] // CHECK1: [[OMP_DISPATCH_END]]: // CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 -// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP32]]) // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: store ptr [[PARTIAL_SUM5]], ptr [[TMP33]], align 8 @@ -1013,16 +1013,16 @@ void test() { // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2, !tbaa [[SHORT_TBAA27]] -// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4, !tbaa [[INT_TBAA10]] +// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4, !tbaa [[INT_TBAA6]] // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0 -// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 8, !tbaa [[ANYPTR_TBAA6]] +// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 8, !tbaa [[ANYPTR_TBAA10]] // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 1 -// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8, !tbaa [[ANYPTR_TBAA6]] +// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8, !tbaa [[ANYPTR_TBAA10]] // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 2 -// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[TMP7]], align 8, !tbaa [[ANYPTR_TBAA6]] +// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[TMP7]], align 8, !tbaa [[ANYPTR_TBAA10]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIdEvv_l16_omp_outlined_omp_outlined(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]], ptr [[TMP6]], ptr [[TMP8]]) #[[ATTR4]] // CHECK1-NEXT: ret void // @@ -1113,22 +1113,22 @@ void test() { // CHECK1-NEXT: ret double [[TMP0]] // //. -// CHECK1: [[ANYPTR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} -// CHECK1: [[META7]] = !{!"any pointer", [[META8:![0-9]+]], i64 0} +// CHECK1: [[INT_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0} +// CHECK1: [[META7]] = !{!"int", [[META8:![0-9]+]], i64 0} // CHECK1: [[META8]] = !{!"omnipotent char", [[META9:![0-9]+]], i64 0} // CHECK1: [[META9]] = !{!"Simple C++ TBAA"} -// CHECK1: [[INT_TBAA10]] = !{[[META11:![0-9]+]], [[META11]], i64 0} -// CHECK1: [[META11]] = !{!"int", [[META8]], i64 0} +// CHECK1: [[ANYPTR_TBAA10]] = !{[[META11:![0-9]+]], [[META11]], i64 0} +// CHECK1: [[META11]] = !{!"any pointer", [[META8]], i64 0} // CHECK1: [[INTPTR_TBAA12]] = !{[[META13:![0-9]+]], [[META13]], i64 0} -// CHECK1: [[META13]] = !{!"p1 int", [[META7]], i64 0} +// CHECK1: [[META13]] = !{!"p1 int", [[META11]], i64 0} // CHECK1: [[FLOAT_TBAA14]] = !{[[META15:![0-9]+]], [[META15]], i64 0} // CHECK1: [[META15]] = !{!"float", [[META8]], i64 0} // CHECK1: [[ANYPTR_TBAA16]] = !{[[META17:![0-9]+]], [[META17]], i64 0} -// CHECK1: [[META17]] = !{!"any p2 pointer", [[META7]], i64 0} +// CHECK1: [[META17]] = !{!"any p2 pointer", [[META11]], i64 0} // CHECK1: [[_ZTSST7COMPLEXIFEPTR_TBAA18]] = !{[[META19:![0-9]+]], [[META19]], i64 0} -// CHECK1: [[META19]] = !{!"p1 _ZTSSt7complexIfE", [[META7]], i64 0} +// CHECK1: [[META19]] = !{!"p1 _ZTSSt7complexIfE", [[META11]], i64 0} // CHECK1: [[FLOATPTR_TBAA20]] = !{[[META21:![0-9]+]], [[META21]], i64 0} -// CHECK1: [[META21]] = !{!"p1 float", [[META7]], i64 0} +// CHECK1: [[META21]] = !{!"p1 float", [[META11]], i64 0} // CHECK1: [[META22]] = !{} // CHECK1: [[META23]] = !{i64 4} // CHECK1: [[FLOAT_TBAA24]] = !{[[META25:![0-9]+]], [[META15]], i64 0} @@ -1139,9 +1139,9 @@ void test() { // CHECK1: [[DOUBLE_TBAA29]] = !{[[META30:![0-9]+]], [[META30]], i64 0} // CHECK1: [[META30]] = !{!"double", [[META8]], i64 0} // CHECK1: [[_ZTSST7COMPLEXIDEPTR_TBAA31]] = !{[[META32:![0-9]+]], [[META32]], i64 0} -// CHECK1: [[META32]] = !{!"p1 _ZTSSt7complexIdE", [[META7]], i64 0} +// CHECK1: [[META32]] = !{!"p1 _ZTSSt7complexIdE", [[META11]], i64 0} // CHECK1: [[DOUBLEPTR_TBAA33]] = !{[[META34:![0-9]+]], [[META34]], i64 0} -// CHECK1: [[META34]] = !{!"p1 double", [[META7]], i64 0} +// CHECK1: [[META34]] = !{!"p1 double", [[META11]], i64 0} // CHECK1: [[META35]] = !{i64 8} // CHECK1: [[DOUBLE_TBAA36]] = !{[[META37:![0-9]+]], [[META30]], i64 0} // CHECK1: [[META37]] = !{!"_ZTSSt7complexIdE", [[META30]], i64 0, [[META30]], i64 8} diff --git a/clang/test/Preprocessor/init-aarch64.c b/clang/test/Preprocessor/init-aarch64.c index 3036b49..460778f 100644 --- a/clang/test/Preprocessor/init-aarch64.c +++ b/clang/test/Preprocessor/init-aarch64.c @@ -234,6 +234,7 @@ // AARCH64-NEXT: #define __LONG_MAX__ 9223372036854775807L // AARCH64-NEXT: #define __LONG_WIDTH__ 64 // AARCH64-NEXT: #define __LP64__ 1 +// AARCH64-NEXT: #define __MEMORY_SCOPE_CLUSTR 5 // AARCH64-NEXT: #define __MEMORY_SCOPE_DEVICE 1 // AARCH64-NEXT: #define __MEMORY_SCOPE_SINGLE 4 // AARCH64-NEXT: #define __MEMORY_SCOPE_SYSTEM 0 @@ -989,6 +990,7 @@ // ARM64EC-MSVC: #define __LONG_LONG_MAX__ 9223372036854775807LL // ARM64EC-MSVC: #define __LONG_MAX__ 2147483647L // ARM64EC-MSVC: #define __LONG_WIDTH__ 32 +// ARM64EC-MSVC: #define __MEMORY_SCOPE_CLUSTR 5 // ARM64EC-MSVC: #define __MEMORY_SCOPE_DEVICE 1 // ARM64EC-MSVC: #define __MEMORY_SCOPE_SINGLE 4 // ARM64EC-MSVC: #define __MEMORY_SCOPE_SYSTEM 0 diff --git a/clang/test/Preprocessor/init-loongarch.c b/clang/test/Preprocessor/init-loongarch.c index 71a266b..fd7ce20 100644 --- a/clang/test/Preprocessor/init-loongarch.c +++ b/clang/test/Preprocessor/init-loongarch.c @@ -182,11 +182,12 @@ // LA32: #define __LONG_LONG_MAX__ 9223372036854775807LL // LA32: #define __LONG_MAX__ 2147483647L // LA32: #define __LONG_WIDTH__ 32 -// LA32: #define __MEMORY_SCOPE_DEVICE 1 -// LA32: #define __MEMORY_SCOPE_SINGLE 4 -// LA32: #define __MEMORY_SCOPE_SYSTEM 0 -// LA32: #define __MEMORY_SCOPE_WRKGRP 2 -// LA32: #define __MEMORY_SCOPE_WVFRNT 3 +// LA32: #define __MEMORY_SCOPE_CLUSTR 5 +// LA32: #define __MEMORY_SCOPE_DEVICE 1 +// LA32: #define __MEMORY_SCOPE_SINGLE 4 +// LA32: #define __MEMORY_SCOPE_SYSTEM 0 +// LA32: #define __MEMORY_SCOPE_WRKGRP 2 +// LA32: #define __MEMORY_SCOPE_WVFRNT 3 // LA32: #define __NO_INLINE__ 1 // LA32: #define __NO_MATH_ERRNO__ 1 // LA32: #define __OBJC_BOOL_IS_BOOL 0 @@ -514,11 +515,12 @@ // LA64: #define __LONG_MAX__ 9223372036854775807L // LA64: #define __LONG_WIDTH__ 64 // LA64: #define __LP64__ 1 -// LA64: #define __MEMORY_SCOPE_DEVICE 1 -// LA64: #define __MEMORY_SCOPE_SINGLE 4 -// LA64: #define __MEMORY_SCOPE_SYSTEM 0 -// LA64: #define __MEMORY_SCOPE_WRKGRP 2 -// LA64: #define __MEMORY_SCOPE_WVFRNT 3 +// LA64: #define __MEMORY_SCOPE_CLUSTR 5 +// LA64: #define __MEMORY_SCOPE_DEVICE 1 +// LA64: #define __MEMORY_SCOPE_SINGLE 4 +// LA64: #define __MEMORY_SCOPE_SYSTEM 0 +// LA64: #define __MEMORY_SCOPE_WRKGRP 2 +// LA64: #define __MEMORY_SCOPE_WVFRNT 3 // LA64: #define __NO_INLINE__ 1 // LA64: #define __NO_MATH_ERRNO__ 1 // LA64: #define __OBJC_BOOL_IS_BOOL 0 diff --git a/clang/test/Preprocessor/init.c b/clang/test/Preprocessor/init.c index 7e0df96..4dea1b5 100644 --- a/clang/test/Preprocessor/init.c +++ b/clang/test/Preprocessor/init.c @@ -1889,6 +1889,7 @@ // WEBASSEMBLY64-NEXT:#define __LONG_MAX__ 9223372036854775807L // WEBASSEMBLY64-NEXT:#define __LONG_WIDTH__ 64 // WEBASSEMBLY64-NEXT:#define __LP64__ 1 +// WEBASSEMBLY-NEXT:#define __MEMORY_SCOPE_CLUSTR 5 // WEBASSEMBLY-NEXT:#define __MEMORY_SCOPE_DEVICE 1 // WEBASSEMBLY-NEXT:#define __MEMORY_SCOPE_SINGLE 4 // WEBASSEMBLY-NEXT:#define __MEMORY_SCOPE_SYSTEM 0 @@ -2216,6 +2217,7 @@ // AVR:#define __LDBL_MIN__ 1.17549435e-38L // AVR:#define __LONG_LONG_MAX__ 9223372036854775807LL // AVR:#define __LONG_MAX__ 2147483647L +// AVR:#define __MEMORY_SCOPE_CLUSTR 5 // AVR:#define __MEMORY_SCOPE_DEVICE 1 // AVR:#define __MEMORY_SCOPE_SINGLE 4 // AVR:#define __MEMORY_SCOPE_SYSTEM 0 @@ -2521,6 +2523,7 @@ // RISCV32: #define __LITTLE_ENDIAN__ 1 // RISCV32: #define __LONG_LONG_MAX__ 9223372036854775807LL // RISCV32: #define __LONG_MAX__ 2147483647L +// RISCV32: #define __MEMORY_SCOPE_CLUSTR 5 // RISCV32: #define __MEMORY_SCOPE_DEVICE 1 // RISCV32: #define __MEMORY_SCOPE_SINGLE 4 // RISCV32: #define __MEMORY_SCOPE_SYSTEM 0 @@ -2745,6 +2748,7 @@ // RISCV64: #define __LONG_LONG_MAX__ 9223372036854775807LL // RISCV64: #define __LONG_MAX__ 9223372036854775807L // RISCV64: #define __LP64__ 1 +// RISCV64: #define __MEMORY_SCOPE_CLUSTR 5 // RISCV64: #define __MEMORY_SCOPE_DEVICE 1 // RISCV64: #define __MEMORY_SCOPE_SINGLE 4 // RISCV64: #define __MEMORY_SCOPE_SYSTEM 0 @@ -2937,11 +2941,11 @@ // XTENSA: #define __GXX_ABI_VERSION {{.*}} // XTENSA: #define __ILP32__ 1 // XTENSA: #define __INT16_C(c) c -// XTENSA: #define __INT16_C_SUFFIX__ +// XTENSA: #define __INT16_C_SUFFIX__ // XTENSA: #define __INT16_MAX__ 32767 // XTENSA: #define __INT16_TYPE__ short // XTENSA: #define __INT32_C(c) c -// XTENSA: #define __INT32_C_SUFFIX__ +// XTENSA: #define __INT32_C_SUFFIX__ // XTENSA: #define __INT32_MAX__ 2147483647 // XTENSA: #define __INT32_TYPE__ int // XTENSA: #define __INT64_C(c) c##LL @@ -2949,7 +2953,7 @@ // XTENSA: #define __INT64_MAX__ 9223372036854775807LL // XTENSA: #define __INT64_TYPE__ long long int // XTENSA: #define __INT8_C(c) c -// XTENSA: #define __INT8_C_SUFFIX__ +// XTENSA: #define __INT8_C_SUFFIX__ // XTENSA: #define __INT8_MAX__ 127 // XTENSA: #define __INT8_TYPE__ signed char // XTENSA: #define __INTMAX_C(c) c##LL @@ -3008,6 +3012,7 @@ // XTENSA: #define __LONG_LONG_MAX__ 9223372036854775807LL // XTENSA: #define __LONG_MAX__ 2147483647L // XTENSA: #define __LONG_WIDTH__ 32 +// XTENSA: #define __MEMORY_SCOPE_CLUSTR 5 // XTENSA: #define __MEMORY_SCOPE_DEVICE 1 // XTENSA: #define __MEMORY_SCOPE_SINGLE 4 // XTENSA: #define __MEMORY_SCOPE_SYSTEM 0 @@ -3050,7 +3055,7 @@ // XTENSA: #define __STDC_VERSION__ 201710L // XTENSA: #define __STDC__ 1 // XTENSA: #define __UINT16_C(c) c -// XTENSA: #define __UINT16_C_SUFFIX__ +// XTENSA: #define __UINT16_C_SUFFIX__ // XTENSA: #define __UINT16_MAX__ 65535 // XTENSA: #define __UINT16_TYPE__ unsigned short // XTENSA: #define __UINT32_C(c) c##U @@ -3062,7 +3067,7 @@ // XTENSA: #define __UINT64_MAX__ 18446744073709551615ULL // XTENSA: #define __UINT64_TYPE__ long long unsigned int // XTENSA: #define __UINT8_C(c) c -// XTENSA: #define __UINT8_C_SUFFIX__ +// XTENSA: #define __UINT8_C_SUFFIX__ // XTENSA: #define __UINT8_MAX__ 255 // XTENSA: #define __UINT8_TYPE__ unsigned char // XTENSA: #define __UINTMAX_C(c) c##ULL @@ -3089,7 +3094,7 @@ // XTENSA: #define __UINT_LEAST64_TYPE__ long long unsigned int // XTENSA: #define __UINT_LEAST8_MAX__ 255 // XTENSA: #define __UINT_LEAST8_TYPE__ unsigned char -// XTENSA: #define __USER_LABEL_PREFIX__ +// XTENSA: #define __USER_LABEL_PREFIX__ // XTENSA: #define __WCHAR_MAX__ 2147483647 // XTENSA: #define __WCHAR_TYPE__ int // XTENSA: #define __WCHAR_WIDTH__ 32 diff --git a/clang/test/Preprocessor/riscv-atomics.c b/clang/test/Preprocessor/riscv-atomics.c new file mode 100644 index 0000000..6e02173 --- /dev/null +++ b/clang/test/Preprocessor/riscv-atomics.c @@ -0,0 +1,24 @@ +// RUN: %clang --target=riscv32-unknown-linux-gnu -march=rv32ia -x c -E -dM %s \ +// RUN: -o - | FileCheck %s +// RUN: %clang --target=riscv32-unknown-linux-gnu -march=rv32i_zalrsc -x c -E \ +// RUN: -dM %s -o - | FileCheck %s +// RUN: %clang --target=riscv64-unknown-linux-gnu -march=rv64ia -x c -E -dM %s \ +// RUN: -o - | FileCheck %s --check-prefixes=CHECK,CHECK-RV64 +// RUN: %clang --target=riscv64-unknown-linux-gnu -march=rv64i_zalrsc -x c -E \ +// RUN: -dM %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-RV64 + +// CHECK: #define __GCC_ATOMIC_BOOL_LOCK_FREE 2 +// CHECK: #define __GCC_ATOMIC_CHAR16_T_LOCK_FREE 2 +// CHECK: #define __GCC_ATOMIC_CHAR32_T_LOCK_FREE 2 +// CHECK: #define __GCC_ATOMIC_CHAR_LOCK_FREE 2 +// CHECK: #define __GCC_ATOMIC_INT_LOCK_FREE 2 +// CHECK-RV64: #define __GCC_ATOMIC_LLONG_LOCK_FREE 2 +// CHECK: #define __GCC_ATOMIC_LONG_LOCK_FREE 2 +// CHECK: #define __GCC_ATOMIC_POINTER_LOCK_FREE 2 +// CHECK: #define __GCC_ATOMIC_SHORT_LOCK_FREE 2 +// CHECK: #define __GCC_ATOMIC_TEST_AND_SET_TRUEVAL 1 +// CHECK: #define __GCC_ATOMIC_WCHAR_T_LOCK_FREE 2 +// CHECK: #define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1 1 +// CHECK: #define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2 1 +// CHECK: #define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4 1 +// CHECK-RV64: #define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8 1 diff --git a/clang/test/Preprocessor/riscv-target-features-sifive.c b/clang/test/Preprocessor/riscv-target-features-sifive.c index 1c49b55..5002f62 100644 --- a/clang/test/Preprocessor/riscv-target-features-sifive.c +++ b/clang/test/Preprocessor/riscv-target-features-sifive.c @@ -5,6 +5,11 @@ // CHECK-NOT: __riscv_xsfcease {{.*$}} // CHECK-NOT: __riscv_xsfvcp {{.*$}} +// CHECK-NOT: __riscv_xsfvfbfexp16e {{.*$}} +// CHECK-NOT: __riscv_xsfvfexp16e {{.*$}} +// CHECK-NOT: __riscv_xsfvfexp32e {{.*$}} +// CHECK-NOT: __riscv_xsfvfexpa {{.*$}} +// CHECK-NOT: __riscv_xsfvfexpa64e {{.*$}} // CHECK-NOT: __riscv_xsfvfnrclipxfqf {{.*$}} // CHECK-NOT: __riscv_xsfvfwmaccqqq {{.*$}} // CHECK-NOT: __riscv_xsfqmaccdod {{.*$}} @@ -39,6 +44,46 @@ // CHECK-XSFVCP-EXT: __riscv_xsfvcp 1000000{{$}} // RUN: %clang --target=riscv32-unknown-linux-gnu \ +// RUN: -march=rv32ixsfvfbfexp16e_zvfbfmin -E -dM %s \ +// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVFBFEXP16E-EXT %s +// RUN: %clang --target=riscv64-unknown-linux-gnu \ +// RUN: -march=rv64ixsfvfbfexp16e_zvfbfmin -E -dM %s \ +// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVFBFEXP16E-EXT %s +// CHECK-XSFVFBFEXP16E-EXT: __riscv_xsfvfbfexp16e 5000{{$}} + +// RUN: %clang --target=riscv32-unknown-linux-gnu \ +// RUN: -march=rv32ixsfvfexp16e -E -dM %s \ +// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVFEXP16E-EXT %s +// RUN: %clang --target=riscv64-unknown-linux-gnu \ +// RUN: -march=rv64ixsfvfexp16e -E -dM %s \ +// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVFEXP16E-EXT %s +// CHECK-XSFVFEXP16E-EXT: __riscv_xsfvfexp16e 5000{{$}} + +// RUN: %clang --target=riscv32-unknown-linux-gnu \ +// RUN: -march=rv32ixsfvfexp32e -E -dM %s \ +// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVFEXP32E-EXT %s +// RUN: %clang --target=riscv64-unknown-linux-gnu \ +// RUN: -march=rv64ixsfvfexp32e -E -dM %s \ +// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVFEXP32E-EXT %s +// CHECK-XSFVFEXP32E-EXT: __riscv_xsfvfexp32e 5000{{$}} + +// RUN: %clang --target=riscv32-unknown-linux-gnu \ +// RUN: -march=rv32ixsfvfexpa -E -dM %s \ +// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVFEXPA-EXT %s +// RUN: %clang --target=riscv64-unknown-linux-gnu \ +// RUN: -march=rv64ixsfvfexpa -E -dM %s \ +// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVFEXPA-EXT %s +// CHECK-XSFVFEXPA-EXT: __riscv_xsfvfexpa 2000{{$}} + +// RUN: %clang --target=riscv32-unknown-linux-gnu \ +// RUN: -march=rv32ixsfvfexpa64e -E -dM %s \ +// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVFEXPA64E-EXT %s +// RUN: %clang --target=riscv64-unknown-linux-gnu \ +// RUN: -march=rv64ixsfvfexpa64e -E -dM %s \ +// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVFEXPA64E-EXT %s +// CHECK-XSFVFEXPA64E-EXT: __riscv_xsfvfexpa64e 2000{{$}} + +// RUN: %clang --target=riscv32-unknown-linux-gnu \ // RUN: -march=rv32ixsfvfnrclipxfqf -E -dM %s \ // RUN: -o - | FileCheck --check-prefix=CHECK-XSFVFNRCLIPXFQF-EXT %s // RUN: %clang --target=riscv64-unknown-linux-gnu \ diff --git a/clang/test/Sema/callingconv-devicekernel.cpp b/clang/test/Sema/callingconv-devicekernel.cpp new file mode 100644 index 0000000..f5da873 --- /dev/null +++ b/clang/test/Sema/callingconv-devicekernel.cpp @@ -0,0 +1,16 @@ +// RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -fsyntax-only -verify %s +// RUN: %clang_cc1 -triple nvptx64-nvidia-cuda- -fsyntax-only -verify %s +// RUN: %clang_cc1 -triple spir64 -fsyntax-only -verify %s +// RUN: %clang_cc1 -triple spirv64 -fsyntax-only -verify %s + +[[clang::device_kernel]] void kernel1() {} + +namespace { +[[clang::device_kernel]] void kernel2() {} // expected-error {{'kernel2' is specified as a device kernel but it is not externally visible}} +} + +namespace ns { + [[clang::device_kernel]] void kernel3() {} +} + +[[clang::device_kernel]] static void kernel4() {} // expected-error {{'kernel4' is specified as a device kernel but it is not externally visible}} diff --git a/clang/test/Sema/callingconv.c b/clang/test/Sema/callingconv.c index f0b8b80..28342b5 100644 --- a/clang/test/Sema/callingconv.c +++ b/clang/test/Sema/callingconv.c @@ -55,6 +55,10 @@ int __attribute__((aarch64_vector_pcs)) aavpcs(void); // expected-warning {{'aar int __attribute__((aarch64_sve_pcs)) aasvepcs(void); // expected-warning {{'aarch64_sve_pcs' calling convention is not supported for this target}} int __attribute__((amdgpu_kernel)) amdgpu_kernel(void); // expected-warning {{'amdgpu_kernel' calling convention is not supported for this target}} +int __attribute__((device_kernel)) device_kernel(void) { // expected-warning {{'device_kernel' calling convention is not supported for this target}} +} +int __attribute__((sycl_kernel)) sycl_kernel(void) { // expected-warning {{'sycl_kernel' attribute ignored}} +} // PR6361 void ctest3(); diff --git a/clang/test/SemaCUDA/atomic-ops.cu b/clang/test/SemaCUDA/atomic-ops.cu index 233ed1c1..40e110c 100644 --- a/clang/test/SemaCUDA/atomic-ops.cu +++ b/clang/test/SemaCUDA/atomic-ops.cu @@ -2,6 +2,8 @@ #include "Inputs/cuda.h" +#define INVALID_HIP_MEMORY_SCOPE (__HIP_MEMORY_SCOPE_CLUSTER+1) + __device__ int test_hip_atomic_load(int *pi32, unsigned int *pu32, long long *pll, unsigned long long *pull, float *fp, double *dbl) { int val = __hip_atomic_load(0); // expected-error {{too few arguments to function call, expected 3, have 1}} val = __hip_atomic_load(0, 0, 0, 0); // expected-error {{too many arguments to function call, expected 3, have 4}} @@ -10,9 +12,10 @@ __device__ int test_hip_atomic_load(int *pi32, unsigned int *pu32, long long *pl val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SINGLETHREAD); val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_WAVEFRONT); val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_WORKGROUP); + val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_CLUSTER); val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_AGENT); val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SYSTEM); - val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, 6); // expected-error {{synchronization scope argument to atomic operation is invalid}} + val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, INVALID_HIP_MEMORY_SCOPE); // expected-error {{synchronization scope argument to atomic operation is invalid}} val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SINGLETHREAD); val = __hip_atomic_load(pi32, __ATOMIC_SEQ_CST, __HIP_MEMORY_SCOPE_SINGLETHREAD); val = __hip_atomic_load(pi32, __ATOMIC_CONSUME, __HIP_MEMORY_SCOPE_SINGLETHREAD); @@ -35,9 +38,10 @@ __device__ int test_hip_atomic_store(int *pi32, unsigned int *pu32, long long *p __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SINGLETHREAD); __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_WAVEFRONT); __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_WORKGROUP); + __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_CLUSTER); __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_AGENT); __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SYSTEM); - __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, 6); // expected-error {{synchronization scope argument to atomic operation is invalid}} + __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, INVALID_HIP_MEMORY_SCOPE); // expected-error {{synchronization scope argument to atomic operation is invalid}} __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SINGLETHREAD); __hip_atomic_store(pi32, 0, __ATOMIC_SEQ_CST, __HIP_MEMORY_SCOPE_SINGLETHREAD); __hip_atomic_store(pi32, 0, __ATOMIC_CONSUME, __HIP_MEMORY_SCOPE_SINGLETHREAD); // expected-warning{{memory order argument to atomic operation is invalid}} @@ -71,6 +75,7 @@ __device__ bool test_hip_atomic_cmpxchg_weak(int *ptr, int val, int desired) { flag = __hip_atomic_compare_exchange_weak(ptr, &val, desired, __ATOMIC_CONSUME, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SINGLETHREAD); flag = __hip_atomic_compare_exchange_weak(ptr, &val, desired, __ATOMIC_RELAXED, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_WAVEFRONT); flag = __hip_atomic_compare_exchange_weak(ptr, &val, desired, __ATOMIC_RELAXED, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_WORKGROUP); + flag = __hip_atomic_compare_exchange_weak(ptr, &val, desired, __ATOMIC_RELAXED, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_CLUSTER); flag = __hip_atomic_compare_exchange_weak(ptr, &val, desired, __ATOMIC_RELAXED, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_AGENT); flag = __hip_atomic_compare_exchange_weak(ptr, &val, desired, __ATOMIC_RELAXED, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SINGLETHREAD); flag = __hip_atomic_compare_exchange_weak(ptr, &val, desired, __ATOMIC_RELAXED, __ATOMIC_SEQ_CST, __HIP_MEMORY_SCOPE_SINGLETHREAD); diff --git a/clang/test/SemaCUDA/spirv-amdgcn-atomic-ops.cu b/clang/test/SemaCUDA/spirv-amdgcn-atomic-ops.cu index ea1f246..503e786 100644 --- a/clang/test/SemaCUDA/spirv-amdgcn-atomic-ops.cu +++ b/clang/test/SemaCUDA/spirv-amdgcn-atomic-ops.cu @@ -2,6 +2,8 @@ #include "Inputs/cuda.h" +#define INVALID_HIP_MEMORY_SCOPE (__HIP_MEMORY_SCOPE_CLUSTER+1) + __device__ int test_hip_atomic_load(int *pi32, unsigned int *pu32, long long *pll, unsigned long long *pull, float *fp, double *dbl) { int val = __hip_atomic_load(0); // expected-error {{too few arguments to function call, expected 3, have 1}} val = __hip_atomic_load(0, 0, 0, 0); // expected-error {{too many arguments to function call, expected 3, have 4}} @@ -10,9 +12,10 @@ __device__ int test_hip_atomic_load(int *pi32, unsigned int *pu32, long long *pl val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SINGLETHREAD); val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_WAVEFRONT); val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_WORKGROUP); + val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_CLUSTER); val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_AGENT); val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SYSTEM); - val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, 6); // expected-error {{synchronization scope argument to atomic operation is invalid}} + val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, INVALID_HIP_MEMORY_SCOPE); // expected-error {{synchronization scope argument to atomic operation is invalid}} val = __hip_atomic_load(pi32, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SINGLETHREAD); val = __hip_atomic_load(pi32, __ATOMIC_SEQ_CST, __HIP_MEMORY_SCOPE_SINGLETHREAD); val = __hip_atomic_load(pi32, __ATOMIC_CONSUME, __HIP_MEMORY_SCOPE_SINGLETHREAD); @@ -35,9 +38,10 @@ __device__ int test_hip_atomic_store(int *pi32, unsigned int *pu32, long long *p __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SINGLETHREAD); __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_WAVEFRONT); __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_WORKGROUP); + __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_CLUSTER); __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_AGENT); __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SYSTEM); - __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, 6); // expected-error {{synchronization scope argument to atomic operation is invalid}} + __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, INVALID_HIP_MEMORY_SCOPE); // expected-error {{synchronization scope argument to atomic operation is invalid}} __hip_atomic_store(pi32, 0, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SINGLETHREAD); __hip_atomic_store(pi32, 0, __ATOMIC_SEQ_CST, __HIP_MEMORY_SCOPE_SINGLETHREAD); __hip_atomic_store(pi32, 0, __ATOMIC_CONSUME, __HIP_MEMORY_SCOPE_SINGLETHREAD); // expected-warning{{memory order argument to atomic operation is invalid}} @@ -71,6 +75,7 @@ __device__ bool test_hip_atomic_cmpxchg_weak(int *ptr, int val, int desired) { flag = __hip_atomic_compare_exchange_weak(ptr, &val, desired, __ATOMIC_CONSUME, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SINGLETHREAD); flag = __hip_atomic_compare_exchange_weak(ptr, &val, desired, __ATOMIC_RELAXED, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_WAVEFRONT); flag = __hip_atomic_compare_exchange_weak(ptr, &val, desired, __ATOMIC_RELAXED, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_WORKGROUP); + flag = __hip_atomic_compare_exchange_weak(ptr, &val, desired, __ATOMIC_RELAXED, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_CLUSTER); flag = __hip_atomic_compare_exchange_weak(ptr, &val, desired, __ATOMIC_RELAXED, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_AGENT); flag = __hip_atomic_compare_exchange_weak(ptr, &val, desired, __ATOMIC_RELAXED, __ATOMIC_RELAXED, __HIP_MEMORY_SCOPE_SINGLETHREAD); flag = __hip_atomic_compare_exchange_weak(ptr, &val, desired, __ATOMIC_RELAXED, __ATOMIC_SEQ_CST, __HIP_MEMORY_SCOPE_SINGLETHREAD); diff --git a/clang/test/SemaCXX/constexpr-string.cpp b/clang/test/SemaCXX/constexpr-string.cpp index c456740..93e2346 100644 --- a/clang/test/SemaCXX/constexpr-string.cpp +++ b/clang/test/SemaCXX/constexpr-string.cpp @@ -7,6 +7,8 @@ // RUN: %clang_cc1 %s -triple armebv7-unknown-linux -std=c++2a -fsyntax-only -verify -pedantic -Wno-vla-extension -fno-signed-char // RUN: %clang_cc1 %s -triple armebv7-unknown-linux -std=c++2a -fsyntax-only -verify -pedantic -Wno-vla-extension -fno-wchar -DNO_PREDEFINED_WCHAR_T +// RUN: %clang_cc1 %s -triple armebv7-unknown-linux -std=c++2a -fsyntax-only -verify -pedantic -Wno-vla-extension -fno-signed-char -fexperimental-new-constant-interpreter + # 9 "/usr/include/string.h" 1 3 4 // expected-warning {{this style of line directive is a GNU extension}} extern "C" { typedef decltype(sizeof(int)) size_t; diff --git a/clang/test/SemaCXX/cxx2c-fold-exprs.cpp b/clang/test/SemaCXX/cxx2c-fold-exprs.cpp index 137f46e..289059e 100644 --- a/clang/test/SemaCXX/cxx2c-fold-exprs.cpp +++ b/clang/test/SemaCXX/cxx2c-fold-exprs.cpp @@ -157,66 +157,55 @@ static_assert(And1<S, S>() == 1); // FIXME: The diagnostics are not so great static_assert(And1<int>() == 1); // expected-error {{no matching function for call to 'And1'}} // expected-note@#and1 {{candidate template ignored: constraints not satisfied [with T = <int>]}} - // expected-note@#and1 {{because 'typename T::type' does not satisfy 'C'}} - // expected-note@#C {{because 'T' does not satisfy 'A'}} + // expected-note@#and1 {{because substituted constraint expression is ill-formed: type 'int' cannot be used prior to '::' because it has no members}} static_assert(And1<S, int>() == 1); // expected-error {{no matching function for call to 'And1'}} // expected-note@#and1 {{candidate template ignored: constraints not satisfied [with T = <S, int>]}} - // expected-note@#and1 {{because 'typename T::type' does not satisfy 'C'}} - // expected-note@#C {{because 'T' does not satisfy 'A'}} + // expected-note@#and1 {{because substituted constraint expression is ill-formed: type 'int' cannot be used prior to '::' because it has no members}} static_assert(And1<int, S>() == 1); // expected-error {{no matching function for call to 'And1'}} // expected-note@#and1 {{candidate template ignored: constraints not satisfied [with T = <int, S>]}} - // expected-note@#and1 {{because 'typename T::type' does not satisfy 'C'}} - // expected-note@#C {{because 'T' does not satisfy 'A'}} + // expected-note@#and1 {{because substituted constraint expression is ill-formed: type 'int' cannot be used prior to '::' because it has no members}} static_assert(And2<S>() == 2); static_assert(And2<S, S>() == 2); static_assert(And2<int>() == 2); // expected-error {{no matching function for call to 'And2'}} // expected-note@#and2 {{candidate template ignored: constraints not satisfied [with T = int, U = <>]}} - // expected-note@#and2 {{because 'typename U::type' does not satisfy 'C'}} - // expected-note@#C {{because 'T' does not satisfy 'A'}} + // expected-note@#and2 {{because substituted constraint expression is ill-formed: type 'int' cannot be used prior to '::' because it has no members}} static_assert(And2<int, int>() == 2); // expected-error {{no matching function for call to 'And2'}} // expected-note@#and2 {{candidate template ignored: constraints not satisfied [with T = S, U = <int>]}} \ - // expected-note@#and2 {{because 'typename U::type' does not satisfy 'C'}} - // expected-note@#C {{because 'T' does not satisfy 'A'}} + // expected-note@#and2 {{because substituted constraint expression is ill-formed: type 'int' cannot be used prior to '::' because it has no members}} static_assert(And2<S, int>() == 2); // expected-error {{no matching function for call to 'And2'}} // expected-note@#and2 {{candidate template ignored: constraints not satisfied [with T = int, U = <S>]}} - // expected-note@#and2 {{because 'typename T::type' does not satisfy 'C'}} - // expected-note@#C {{because 'T' does not satisfy 'A'}} + // expected-note@#and2 {{because substituted constraint expression is ill-formed: type 'int' cannot be used prior to '::' because it has no members}} static_assert(And2<int, S>() == 2); // expected-error {{no matching function for call to 'And2'}} // expected-note@#and2 {{candidate template ignored: constraints not satisfied [with T = int, U = <int>]}} - // expected-note@#and2 {{because 'typename T::type' does not satisfy 'C'}} - // expected-note@#C {{because 'T' does not satisfy 'A'}} + // expected-note@#and2 {{because substituted constraint expression is ill-formed: type 'int' cannot be used prior to '::' because it has no members}} static_assert(And3<S>() == 3); static_assert(And3<S, S>() == 3); static_assert(And3<int>() == 3); // expected-error {{no matching function for call to 'And3'}} // expected-note@#and3 {{candidate template ignored: constraints not satisfied [with T = int, U = <>]}} - // expected-note@#and3 {{because 'typename T::type' does not satisfy 'C'}} - // expected-note@#C {{because 'T' does not satisfy 'A'}} + // expected-note@#and3 {{because substituted constraint expression is ill-formed: type 'int' cannot be used prior to '::' because it has no members}} static_assert(And3<int, int>() == 3); // expected-error {{no matching function for call to 'And3'}} // expected-note@#and3 {{candidate template ignored: constraints not satisfied [with T = int, U = <int>]}} - // expected-note@#and3 {{because 'typename T::type' does not satisfy 'C'}} - // expected-note@#C {{because 'T' does not satisfy 'A'}} + // expected-note@#and3 {{because substituted constraint expression is ill-formed: type 'int' cannot be used prior to '::' because it has no members}} static_assert(And3<S, int>() == 3); // expected-error {{no matching function for call to 'And3'}} // expected-note@#and3 {{candidate template ignored: constraints not satisfied [with T = S, U = <int>]}} - // expected-note@#and3 {{because 'typename U::type' does not satisfy 'C'}} - // expected-note@#C {{because 'T' does not satisfy 'A'}} + // expected-note@#and3 {{because substituted constraint expression is ill-formed: type 'int' cannot be used prior to '::' because it has no members}} static_assert(And3<int, S>() == 3); // expected-error {{no matching function for call to 'And3'}} // expected-note@#and3 {{candidate template ignored: constraints not satisfied [with T = int, U = <S>]}} - // expected-note@#and3 {{because 'typename T::type' does not satisfy 'C'}} - // expected-note@#C {{because 'T' does not satisfy 'A'}} + // expected-note@#and3 {{because substituted constraint expression is ill-formed: type 'int' cannot be used prior to '::' because it has no members}} static_assert(Or1<>() == 1); // expected-error {{no matching function for call to 'Or1'}} @@ -227,8 +216,7 @@ static_assert(Or1<S, int>() == 1); static_assert(Or1<S, S>() == 1); static_assert(Or1<int>() == 1); // expected-error {{no matching function for call to 'Or1'}} // expected-note@#or1 {{candidate template ignored: constraints not satisfied}} - // expected-note@#or1 {{because 'typename T::type' does not satisfy 'C'}} - // expected-note@#C {{because 'T' does not satisfy 'A'}} + // expected-note@#or1 {{because substituted constraint expression is ill-formed: type 'int' cannot be used prior to '::' because it has no members}} static_assert(Or2<S>() == 2); static_assert(Or2<int, S>() == 2); @@ -236,16 +224,14 @@ static_assert(Or2<S, int>() == 2); static_assert(Or2<S, S>() == 2); static_assert(Or2<int>() == 2); // expected-error {{no matching function for call to 'Or2'}} // expected-note@#or2 {{candidate template ignored: constraints not satisfied [with T = int, U = <>]}} - // expected-note@#or2 {{because 'typename T::type' does not satisfy 'C'}} - // expected-note@#C {{because 'T' does not satisfy 'A'}} + // expected-note@#or2 {{because substituted constraint expression is ill-formed: type 'int' cannot be used prior to '::' because it has no members}} static_assert(Or3<S>() == 3); static_assert(Or3<int, S>() == 3); static_assert(Or3<S, int>() == 3); static_assert(Or3<S, S>() == 3); static_assert(Or3<int>() == 3); // expected-error {{no matching function for call to 'Or3'}} // expected-note@#or3 {{candidate template ignored: constraints not satisfied}} - // expected-note@#or3 {{because 'typename T::type' does not satisfy 'C'}} - // expected-note@#C {{because 'T' does not satisfy 'A'}} + // expected-note@#or3 {{because substituted constraint expression is ill-formed: type 'int' cannot be used prior to '::' because it has no members}} } namespace bool_conversion_break { diff --git a/clang/test/SemaTemplate/GH164330.cpp b/clang/test/SemaTemplate/GH164330.cpp new file mode 100644 index 0000000..767420d --- /dev/null +++ b/clang/test/SemaTemplate/GH164330.cpp @@ -0,0 +1,309 @@ +// RUN: %clang_cc1 -std=c++20 -verify %s +// expected-no-diagnostics + +template <int __v> struct integral_constant { + static constexpr int value = __v; +}; +template <bool _Val> using _BoolConstant = integral_constant<_Val>; +template <int, class> struct tuple_element; +template <class...> class tuple; +template <int _Ip, class... _Tp> struct tuple_element<_Ip, tuple<_Tp...>> { + using type = __type_pack_element<_Ip, _Tp...>; +}; +template <class> struct tuple_size; +template <bool> using __enable_if_t = int; +template <template <class> class _BaseType, class _Tp, _Tp _SequenceSize> +using __make_integer_sequence_impl = + __make_integer_seq<_BaseType, _Tp, _SequenceSize>; +template <class _Tp, _Tp...> struct __integer_sequence; +template <int... _Indices> +using __index_sequence = __integer_sequence<int, _Indices...>; +template <int _SequenceSize> +using __make_index_sequence = + __make_integer_sequence_impl<__integer_sequence, int, _SequenceSize>; +template <class _Tp, _Tp...> struct integer_sequence {}; +template <int... _Ip> using index_sequence = integer_sequence<int, _Ip...>; +template <class _Tp, _Tp _Ep> +using make_integer_sequence = + __make_integer_sequence_impl<integer_sequence, _Tp, _Ep>; +template <int _Np> using make_index_sequence = make_integer_sequence<int, _Np>; +enum __element_count : int; +constexpr void __constexpr_memmove(char *__dest, const char *__src, + __element_count __n) { + __builtin_memmove(__dest, __src, __n); +} +template <class _Tp> using __underlying_type_t = __underlying_type(_Tp); +template <class _Tp> using underlying_type_t = __underlying_type_t<_Tp>; +template <class _Tp, class> using __enable_if_tuple_size_imp = _Tp; +template <class _Tp> +struct tuple_size<__enable_if_tuple_size_imp< + const _Tp, __enable_if_t<_BoolConstant<__is_volatile(int)>::value>>> + : integral_constant<tuple_size<_Tp>::value> {}; +template <class... _Tp> +struct tuple_size<tuple<_Tp...>> : integral_constant<sizeof...(_Tp)> {}; +template <class _Tp> constexpr int tuple_size_v = tuple_size<_Tp>::value; +template <class _T1, class _T2> struct pair { + _T1 first; + _T2 second; +}; +template <class _T1> constexpr pair<_T1, char *> make_pair(_T1, char *__t2) { + return pair(_T1(), __t2); +} +template <int, class _Hp> struct __tuple_leaf { + _Hp __value_; + constexpr const _Hp &get() const { return __value_; } +}; +template <class...> struct __tuple_impl; +template <int... _Indx, class... _Tp> +struct __tuple_impl<__index_sequence<_Indx...>, _Tp...> + : __tuple_leaf<_Indx, _Tp>... { + template <class... _Args> + constexpr __tuple_impl(int, _Args... __args) + : __tuple_leaf<_Indx, _Tp>(__args)... {} +}; +template <class... _Tp> struct tuple { + __tuple_impl<__make_index_sequence<sizeof...(_Tp)>, _Tp...> __base_; + template <class... _Up> constexpr tuple(_Up... __u) : __base_({}, __u...) {} +}; +template <int _Ip, class... _Tp> +constexpr const tuple_element<_Ip, tuple<_Tp...>>::type & +get(const tuple<_Tp...> &__t) noexcept { + return static_cast<const __tuple_leaf< + _Ip, typename tuple_element<_Ip, tuple<_Tp...>>::type> &>(__t.__base_) + .get(); +} +template <class... _Tp> constexpr tuple<_Tp...> make_tuple(_Tp... __t) { + return tuple<_Tp...>(__t...); +} +constexpr int __char_traits_length_checked(const char *__s) { + return __builtin_strlen(__s); +} +struct basic_string_view { + constexpr basic_string_view() {} + constexpr basic_string_view(const char *__s) + : __data_(__s), __size_(__char_traits_length_checked(__s)) {} + constexpr const char *begin() { return __data_; } + constexpr const char *end() { + return __data_ + __size_; + } + const char *__data_; + int __size_; +}; +template <class _Algorithm> +constexpr pair<const char *, char *> +__copy_move_unwrap_iters(const char *__first, const char *__last, + char *__out_first) { + pair<const char *, const char *> __range = {__first, __last}; + auto __result = _Algorithm()(__range.first, __range.second, __out_first); + return make_pair(__result.first, __result.second); +} +struct __copy_impl { + constexpr pair<const char *, char *> + operator()(const char *__first, const char *__last, char *__result) { + const int __n(__last - __first); + __constexpr_memmove(__result, __first, __element_count(__n)); + return make_pair(__last, __result); + } +}; +constexpr char *copy(const char *__first, const char *__last, char *__result) { + return __copy_move_unwrap_iters<__copy_impl>(__first, __last, __result).second; +} +constexpr char *copy_n(const char *__first, int __orig_n, char *__result) { + return copy(__first, __first + __orig_n, __result); +} +template <int _Size> struct array { + basic_string_view __elems_[_Size]; + constexpr basic_string_view &operator[](int __n) { return __elems_[__n]; } + constexpr basic_string_view operator[](int __n) const { + return __elems_[__n]; + } +}; + +template <typename> struct FieldId; + +template <FieldId field> constexpr auto FieldIdToInnerValue() { + return field.template ToInnerValue<field>(); +} +struct FieldNameEnum { + enum class type; +}; +template <int N> using FieldName = FieldNameEnum::type; +template <typename, auto> struct GetParentMessageAtIndexImpl; +template <typename, auto> struct FieldInfoHelper; +template <FieldId...> struct PathImpl; +template <int N> struct LongPathLiteral { + consteval LongPathLiteral(const char (&s)[N]) { + copy_n(s, N, long_path)[N] = field_count = long_path_size = 1; + } + consteval basic_string_view to_string_view() const { return long_path; } + char long_path[N + 1]; + int long_path_size; + int field_count; +}; +template <LongPathLiteral kLongPath> consteval auto get_field_components() { + basic_string_view long_path(kLongPath.to_string_view()); + array<kLongPath.field_count> ret; + for (int i = 0; i < kLongPath.field_count; ++i) + ret[i] = long_path; + return ret; +} +template <LongPathLiteral kLongPath> +constexpr auto kFieldComponents = get_field_components<kLongPath>(); +template <LongPathLiteral kLongPath> struct LongPathHelper { + template <int... I> + static PathImpl<kFieldComponents<kLongPath>[I]...> + PathForLongPath(index_sequence<I...>); + using type = + decltype(PathForLongPath(make_index_sequence<kLongPath.field_count>{})); +}; +template <typename T> struct PathFieldId { + template <typename Arg> constexpr PathFieldId(Arg &arg) : value(arg) {} + T value; +}; +template <PathFieldId...> constexpr auto PathImplHelper(); + +template <int N> using FieldName = FieldName<N>; +enum class FieldNumber; +template <PathFieldId... fields> +constexpr auto Path = PathImplHelper<fields...>(); +template <typename Proto, FieldId field> +using FieldInfo = + FieldInfoHelper<Proto, FieldIdToInnerValue<field>()>::type; +template <> struct FieldId<FieldNameEnum::type> { + constexpr FieldId(basic_string_view); + int size; + long hash; + template <auto field> static constexpr auto ToInnerValue() { + return static_cast<FieldNameEnum::type>(field.hash); + } +}; +FieldId(basic_string_view) -> FieldId<FieldNameEnum::type>; +template <typename Proto, FieldId field, int index> +using GetParentMessageAtIndex = GetParentMessageAtIndexImpl< + Proto, FieldIdToInnerValue<field>()>::type; + +template <typename T> +PathFieldId(T &t) -> PathFieldId<decltype(LongPathLiteral(t))>; +template <FieldId... fields1, FieldId... fields2> +constexpr PathImpl<fields1..., fields2...> *ConcatPath(PathImpl<fields1...> *, + PathImpl<fields2...> *) { + return nullptr; +} +template <LongPathLiteral long_path_literal> +constexpr LongPathHelper<long_path_literal>::type *SinglePath() { + return nullptr; +} +template <PathFieldId... fields> constexpr auto PathImplHelper() { + return ConcatPath(SinglePath<fields.value>()...); +} +template <auto hash_prime, auto offset_bias> +constexpr auto Fnv1a(basic_string_view str) { + auto hash = offset_bias; + for (char c : str) { + hash ^= c; + hash *= hash_prime; + } + return hash; +} +constexpr auto HashField(basic_string_view str) { + return Fnv1a<1099511628211u, 1039346656037>(str); +} +template <typename FI> struct FieldInfoValueTypeAlias : FI {}; +template <typename Proto, auto field> struct FieldInfoHelperBase { + static constexpr auto MaskFieldNameHash() { + using FieldEnum = decltype(field); + return FieldEnum{static_cast<underlying_type_t<FieldEnum>>(field) & 31}; + } + using internal_type = + Proto::template FieldInfoImpl<decltype(field), MaskFieldNameHash()>; +}; +template <typename Proto, auto field> struct FieldInfoHelper { + using type = FieldInfoValueTypeAlias< + typename FieldInfoHelperBase<Proto, field>::internal_type>; +}; + +template <auto... fields> +struct FieldId<const PathImpl<fields...> *> { + constexpr FieldId(PathImpl<fields...> *) : path() {} + template <auto field> static constexpr auto ToInnerValue() { + return field.path; + } + const PathImpl<fields...> *path; +}; +template <auto... fields> +FieldId(PathImpl<fields...> *) + -> FieldId<const PathImpl<fields...> *>; + +template <auto> struct UnpackedField { + static constexpr bool is_path = false; +}; +template <auto... fields, const PathImpl<fields...> *path> +struct UnpackedField<path> { + static constexpr auto value = make_tuple(fields...); + static constexpr bool is_path = true; +}; +template <typename Proto, FieldId... fields, const PathImpl<fields...> *path> +struct GetParentMessageAtIndexImpl<Proto, path> { + using type = Proto; +}; + +constexpr FieldId<FieldNameEnum::type>::FieldId(basic_string_view str) + : size(), hash(HashField(str)) {} +template <FieldId field> constexpr bool IsPath() { + return UnpackedField< + FieldIdToInnerValue<field>()>::is_path; +} +template <FieldId field> constexpr auto UnpackFieldToTuple() { + return UnpackedField<FieldIdToInnerValue<field>()>::value; +} +template <int> struct CompileTimeString { + consteval CompileTimeString(basic_string_view &v) : internal_view_(v) {} + basic_string_view &internal_view_; +}; +CompileTimeString(basic_string_view) -> CompileTimeString<0>; + +template <CompileTimeString... parts> struct NameJoiner { + template <CompileTimeString... after> + NameJoiner<parts...> operator+(NameJoiner<after...>); +}; +template <FieldId> struct FieldNameBuilder; +template <FieldId field> + requires(!IsPath<field>()) +struct FieldNameBuilder<field> { + template <typename Proto> static auto Get() { + return NameJoiner<FieldInfo<Proto, field>::name>(); + } +}; +template <FieldId field> + requires(IsPath<field>()) +struct FieldNameBuilder<field> { + static constexpr auto kTuple = UnpackFieldToTuple<field>(); + static constexpr int kTupleSize = tuple_size_v<decltype(kTuple)>; + template <typename Proto, int... Is> static void Get(index_sequence<Is...>) { + (FieldNameBuilder<get<Is>( + kTuple)>::template Get<GetParentMessageAtIndex<Proto, field, Is>>() + + ...); + } + template <typename Proto> static void Get() { + Get<Proto>(make_index_sequence<kTupleSize>()); + } +}; + +struct T { + template <typename FieldType, FieldType> struct FieldInfoImpl; +}; +void AddPathsToFieldMask() { + FieldNameBuilder<Path<"message_field", "int32_field">>::Get<T>(); +} +template <> struct T::FieldInfoImpl<FieldNumber, FieldNumber{1}> { + static basic_string_view name; +}; +template <> +struct T::FieldInfoImpl<FieldName<1>, FieldName<1>{12}> + : FieldInfoImpl<FieldNumber, FieldNumber{1}> {}; +template <> struct T::FieldInfoImpl<FieldNumber, FieldNumber{10}> { + static basic_string_view name; +}; +template <> +struct T::FieldInfoImpl<FieldName<3>, FieldName<3>{11}> + : FieldInfoImpl<FieldNumber, FieldNumber{10}> {}; diff --git a/clang/test/SemaTemplate/concepts-recursive-inst.cpp b/clang/test/SemaTemplate/concepts-recursive-inst.cpp index 73dce93..5e1bce5 100644 --- a/clang/test/SemaTemplate/concepts-recursive-inst.cpp +++ b/clang/test/SemaTemplate/concepts-recursive-inst.cpp @@ -82,7 +82,6 @@ auto it = begin(rng); // #BEGIN_CALL // expected-error@#BEGIN_CALL {{no matching function for call to 'begin'}} // expected-note@#NOTINF_BEGIN {{candidate function}} // expected-note@#INF_BEGIN{{candidate template ignored: constraints not satisfied}} -// expected-note@#INF_BEGIN{{because 'Inf auto' does not satisfy 'Inf}} } } // namespace DirectRecursiveCheck diff --git a/clang/test/lit.cfg.py b/clang/test/lit.cfg.py index 29088ef..52b275c 100644 --- a/clang/test/lit.cfg.py +++ b/clang/test/lit.cfg.py @@ -18,11 +18,22 @@ from lit.llvm.subst import FindTool # name: The name of this test suite. config.name = "Clang" +# TODO: Consolidate the logic for turning on the internal shell by default for all LLVM test suites. +# See https://github.com/llvm/llvm-project/issues/106636 for more details. +# +# We prefer the lit internal shell which provides a better user experience on failures +# and is faster unless the user explicitly disables it with LIT_USE_INTERNAL_SHELL=0 +# env var. +use_lit_shell = True +lit_shell_env = os.environ.get("LIT_USE_INTERNAL_SHELL") +if lit_shell_env: + use_lit_shell = lit.util.pythonize_bool(lit_shell_env) + # testFormat: The test format to use to interpret tests. # # For now we require '&&' between commands, until they get globally killed and # the test runner updated. -config.test_format = lit.formats.ShTest(not llvm_config.use_lit_shell) +config.test_format = lit.formats.ShTest(execute_external=not use_lit_shell) # suffixes: A list of file extensions to treat as test files. config.suffixes = [ |