aboutsummaryrefslogtreecommitdiff
path: root/llvm/test/CodeGen/DirectX
diff options
context:
space:
mode:
Diffstat (limited to 'llvm/test/CodeGen/DirectX')
-rw-r--r--llvm/test/CodeGen/DirectX/UAddc.ll8
-rw-r--r--llvm/test/CodeGen/DirectX/bugfix_150050_data_scalarize_const_gep.ll80
-rw-r--r--llvm/test/CodeGen/DirectX/legalize-lifetimes-valver-1.6.ll59
3 files changed, 102 insertions, 45 deletions
diff --git a/llvm/test/CodeGen/DirectX/UAddc.ll b/llvm/test/CodeGen/DirectX/UAddc.ll
index 4b46b56..dd7aa23 100644
--- a/llvm/test/CodeGen/DirectX/UAddc.ll
+++ b/llvm/test/CodeGen/DirectX/UAddc.ll
@@ -35,14 +35,10 @@ define noundef <2 x i32> @test_UAddc_vec2(<2 x i32> noundef %a, <2 x i32> nounde
; CHECK-NEXT: [[UADDC_I1:%.*]] = call [[DX_TYPES_I32C]] @dx.op.binaryWithCarryOrBorrow.i32(i32 44, i32 [[A_I1]], i32 [[B_I1]]) #[[ATTR0]]
; CHECK-NEXT: [[CARRY_ELEM0:%.*]] = extractvalue [[DX_TYPES_I32C]] [[UADDC_I0]], 1
; CHECK-NEXT: [[CARRY_ELEM1:%.*]] = extractvalue [[DX_TYPES_I32C]] [[UADDC_I1]], 1
-; CHECK-NEXT: [[CARRY_UPTO0:%.*]] = insertelement <2 x i1> poison, i1 [[CARRY_ELEM0]], i64 0
-; CHECK-NEXT: [[CARRY:%.*]] = insertelement <2 x i1> [[CARRY_UPTO0]], i1 [[CARRY_ELEM1]], i64 1
-; CHECK-NEXT: [[CARRY_I0:%.*]] = extractelement <2 x i1> [[CARRY]], i64 0
-; CHECK-NEXT: [[CARRY_I1:%.*]] = extractelement <2 x i1> [[CARRY]], i64 1
; CHECK-NEXT: [[SUM_ELEM0:%.*]] = extractvalue [[DX_TYPES_I32C]] [[UADDC_I0]], 0
; CHECK-NEXT: [[SUM_ELEM1:%.*]] = extractvalue [[DX_TYPES_I32C]] [[UADDC_I1]], 0
-; CHECK-NEXT: [[CARRY_ZEXT_I0:%.*]] = zext i1 [[CARRY_I0]] to i32
-; CHECK-NEXT: [[CARRY_ZEXT_I1:%.*]] = zext i1 [[CARRY_I1]] to i32
+; CHECK-NEXT: [[CARRY_ZEXT_I0:%.*]] = zext i1 [[CARRY_ELEM0]] to i32
+; CHECK-NEXT: [[CARRY_ZEXT_I1:%.*]] = zext i1 [[CARRY_ELEM1]] to i32
; CHECK-NEXT: [[RESULT_I0:%.*]] = add i32 [[SUM_ELEM0]], [[CARRY_ZEXT_I0]]
; CHECK-NEXT: [[RESULT_I1:%.*]] = add i32 [[SUM_ELEM1]], [[CARRY_ZEXT_I1]]
; CHECK-NEXT: [[RESULT_UPTO0:%.*]] = insertelement <2 x i32> poison, i32 [[RESULT_I0]], i64 0
diff --git a/llvm/test/CodeGen/DirectX/bugfix_150050_data_scalarize_const_gep.ll b/llvm/test/CodeGen/DirectX/bugfix_150050_data_scalarize_const_gep.ll
new file mode 100644
index 0000000..156a8e7
--- /dev/null
+++ b/llvm/test/CodeGen/DirectX/bugfix_150050_data_scalarize_const_gep.ll
@@ -0,0 +1,80 @@
+; RUN: opt -S -passes='dxil-data-scalarization' -mtriple=dxil-pc-shadermodel6.4-library %s | FileCheck %s --check-prefixes=SCHECK,CHECK
+; RUN: opt -S -passes='dxil-data-scalarization,function(scalarizer<load-store>),dxil-flatten-arrays' -mtriple=dxil-pc-shadermodel6.4-library %s | FileCheck %s --check-prefixes=FCHECK,CHECK
+
+@aTile = hidden addrspace(3) global [10 x [10 x <4 x i32>]] zeroinitializer, align 16
+@bTile = hidden addrspace(3) global [10 x [10 x i32]] zeroinitializer, align 16
+@cTile = internal global [2 x [2 x <2 x i32>]] zeroinitializer, align 16
+@dTile = internal global [2 x [2 x [2 x <2 x i32>]]] zeroinitializer, align 16
+
+define void @CSMain() {
+; CHECK-LABEL: define void @CSMain() {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[AFRAGPACKED_I_SCALARIZE:%.*]] = alloca [4 x i32], align 16
+;
+; SCHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds [10 x <4 x i32>], ptr addrspace(3) getelementptr inbounds ([10 x [10 x [4 x i32]]], ptr addrspace(3) @aTile.scalarized, i32 0, i32 1), i32 0, i32 2
+; SCHECK-NEXT: [[TMP1:%.*]] = load <4 x i32>, ptr addrspace(3) [[TMP0]], align 16
+; SCHECK-NEXT: store <4 x i32> [[TMP1]], ptr [[AFRAGPACKED_I_SCALARIZE]], align 16
+;
+; FCHECK-NEXT: [[AFRAGPACKED_I_SCALARIZE_I14:%.*]] = getelementptr [4 x i32], ptr [[AFRAGPACKED_I_SCALARIZE]], i32 0, i32 1
+; FCHECK-NEXT: [[AFRAGPACKED_I_SCALARIZE_I25:%.*]] = getelementptr [4 x i32], ptr [[AFRAGPACKED_I_SCALARIZE]], i32 0, i32 2
+; FCHECK-NEXT: [[AFRAGPACKED_I_SCALARIZE_I36:%.*]] = getelementptr [4 x i32], ptr [[AFRAGPACKED_I_SCALARIZE]], i32 0, i32 3
+; FCHECK-NEXT: [[DOTI07:%.*]] = load i32, ptr addrspace(3) getelementptr inbounds ([400 x i32], ptr addrspace(3) @aTile.scalarized.1dim, i32 0, i32 48), align 16
+; FCHECK-NEXT: [[DOTI119:%.*]] = load i32, ptr addrspace(3) getelementptr ([400 x i32], ptr addrspace(3) @aTile.scalarized.1dim, i32 0, i32 49), align 4
+; FCHECK-NEXT: [[DOTI2211:%.*]] = load i32, ptr addrspace(3) getelementptr ([400 x i32], ptr addrspace(3) @aTile.scalarized.1dim, i32 0, i32 50), align 8
+; FCHECK-NEXT: [[DOTI3313:%.*]] = load i32, ptr addrspace(3) getelementptr ([400 x i32], ptr addrspace(3) @aTile.scalarized.1dim, i32 0, i32 51), align 4
+; FCHECK-NEXT: store i32 [[DOTI07]], ptr [[AFRAGPACKED_I_SCALARIZE]], align 16
+; FCHECK-NEXT: store i32 [[DOTI119]], ptr [[AFRAGPACKED_I_SCALARIZE_I14]], align 4
+; FCHECK-NEXT: store i32 [[DOTI2211]], ptr [[AFRAGPACKED_I_SCALARIZE_I25]], align 8
+; FCHECK-NEXT: store i32 [[DOTI3313]], ptr [[AFRAGPACKED_I_SCALARIZE_I36]], align 4
+;
+; CHECK-NEXT: ret void
+entry:
+ %aFragPacked.i = alloca <4 x i32>, align 16
+ %0 = load <4 x i32>, ptr addrspace(3) getelementptr inbounds ([10 x <4 x i32>], ptr addrspace(3) getelementptr inbounds ([10 x [10 x <4 x i32>]], ptr addrspace(3) @aTile, i32 0, i32 1), i32 0, i32 2), align 16
+ store <4 x i32> %0, ptr %aFragPacked.i, align 16
+ ret void
+}
+
+define void @Main() {
+; CHECK-LABEL: define void @Main() {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[BFRAGPACKED_I:%.*]] = alloca i32, align 16
+;
+; SCHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds [10 x i32], ptr addrspace(3) getelementptr inbounds ([10 x [10 x i32]], ptr addrspace(3) @bTile, i32 0, i32 1), i32 0, i32 1
+; SCHECK-NEXT: [[TMP1:%.*]] = load i32, ptr addrspace(3) [[TMP0]], align 16
+; SCHECK-NEXT: store i32 [[TMP1]], ptr [[BFRAGPACKED_I]], align 16
+;
+; FCHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(3) getelementptr inbounds ([100 x i32], ptr addrspace(3) @bTile.1dim, i32 0, i32 11), align 16
+; FCHECK-NEXT: store i32 [[TMP0]], ptr [[BFRAGPACKED_I]], align 16
+;
+; CHECK-NEXT: ret void
+entry:
+ %bFragPacked.i = alloca i32, align 16
+ %0 = load i32, ptr addrspace(3) getelementptr inbounds ([10 x i32], ptr addrspace(3) getelementptr inbounds ([10 x [10 x i32]], ptr addrspace(3) @bTile, i32 0, i32 1), i32 0, i32 1), align 16
+ store i32 %0, ptr %bFragPacked.i, align 16
+ ret void
+}
+
+define void @global_nested_geps_3d() {
+; CHECK-LABEL: define void @global_nested_geps_3d() {
+; SCHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds <2 x i32>, ptr getelementptr inbounds ([2 x <2 x i32>], ptr getelementptr inbounds ([2 x [2 x [2 x i32]]], ptr @cTile.scalarized, i32 0, i32 1), i32 0, i32 1), i32 0, i32 1
+; SCHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
+;
+; FCHECK-NEXT: [[TMP1:%.*]] = load i32, ptr getelementptr inbounds ([8 x i32], ptr @cTile.scalarized.1dim, i32 0, i32 7), align 4
+;
+; CHECK-NEXT: ret void
+ %1 = load i32, i32* getelementptr inbounds (<2 x i32>, <2 x i32>* getelementptr inbounds ([2 x <2 x i32>], [2 x <2 x i32>]* getelementptr inbounds ([2 x [2 x <2 x i32>]], [2 x [2 x <2 x i32>]]* @cTile, i32 0, i32 1), i32 0, i32 1), i32 0, i32 1), align 4
+ ret void
+}
+
+define void @global_nested_geps_4d() {
+; CHECK-LABEL: define void @global_nested_geps_4d() {
+; SCHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds <2 x i32>, ptr getelementptr inbounds ([2 x <2 x i32>], ptr getelementptr inbounds ([2 x [2 x <2 x i32>]], ptr getelementptr inbounds ([2 x [2 x [2 x [2 x i32]]]], ptr @dTile.scalarized, i32 0, i32 1), i32 0, i32 1), i32 0, i32 1), i32 0, i32 1
+; SCHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
+;
+; FCHECK-NEXT: [[TMP1:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @dTile.scalarized.1dim, i32 0, i32 15), align 4
+;
+; CHECK-NEXT: ret void
+ %1 = load i32, i32* getelementptr inbounds (<2 x i32>, <2 x i32>* getelementptr inbounds ([2 x <2 x i32>], [2 x <2 x i32>]* getelementptr inbounds ([2 x [2 x <2 x i32>]], [2 x [2 x <2 x i32>]]* getelementptr inbounds ([2 x [2 x [2 x <2 x i32>]]], [2 x [2 x [2 x <2 x i32>]]]* @dTile, i32 0, i32 1), i32 0, i32 1), i32 0, i32 1), i32 0, i32 1), align 4
+ ret void
+}
diff --git a/llvm/test/CodeGen/DirectX/legalize-lifetimes-valver-1.6.ll b/llvm/test/CodeGen/DirectX/legalize-lifetimes-valver-1.6.ll
index f77df2d..77133eb 100644
--- a/llvm/test/CodeGen/DirectX/legalize-lifetimes-valver-1.6.ll
+++ b/llvm/test/CodeGen/DirectX/legalize-lifetimes-valver-1.6.ll
@@ -1,30 +1,27 @@
; RUN: opt -S -passes='dxil-op-lower' -mtriple=dxil-pc-shadermodel6.3-library %s | FileCheck %s --check-prefixes=CHECK,CHECK-SM63
; RUN: opt -S -passes='dxil-op-lower' -mtriple=dxil-pc-shadermodel6.6-library %s | FileCheck %s --check-prefixes=CHECK,CHECK-SM66
-; RUN: opt -S -dxil-op-lower -dxil-prepare -mtriple=dxil-pc-shadermodel6.6-library %s | FileCheck %s --check-prefixes=CHECK,CHECK-PREPARE
+; RUN: opt -S -dxil-prepare -dxil-embed -mtriple=dxil-pc-shadermodel6.6-library %s | FileCheck %s --check-prefixes=CHECK,CHECK-EMBED
+
+; Lifetime intrinsics are not valid prior to shader model 6.6 and are instead
+; replaced with undef stores, provided the validator version is 1.6 or greater
+
+; The dxil-embed pass will remove lifetime intrinsics because they transformed
+; in a way that is illegal in modern LLVM IR before serializing to DXIL bitcode.
+; So we check that no bitcast or lifetime intrinsics remain after dxil-embed
; CHECK-LABEL: define void @test_legal_lifetime() {
-;
-; CHECK-SM63-NEXT: [[ACCUM_I_FLAT:%.*]] = alloca [1 x i32], align 4
-; CHECK-SM63-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[ACCUM_I_FLAT]], i32 0
-; CHECK-SM63-NEXT: store [1 x i32] undef, ptr [[ACCUM_I_FLAT]], align 4
-; CHECK-SM63-NEXT: store i32 0, ptr [[GEP]], align 4
-; CHECK-SM63-NEXT: store [1 x i32] undef, ptr [[ACCUM_I_FLAT]], align 4
-;
-; CHECK-SM66-NEXT: [[ACCUM_I_FLAT:%.*]] = alloca [1 x i32], align 4
-; CHECK-SM66-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[ACCUM_I_FLAT]], i32 0
-; CHECK-SM66-NEXT: call void @llvm.lifetime.start.p0(i64 4, ptr nonnull [[ACCUM_I_FLAT]])
-; CHECK-SM66-NEXT: store i32 0, ptr [[GEP]], align 4
-; CHECK-SM66-NEXT: call void @llvm.lifetime.end.p0(i64 4, ptr nonnull [[ACCUM_I_FLAT]])
-;
-; CHECK-PREPARE-NEXT: [[ACCUM_I_FLAT:%.*]] = alloca [1 x i32], align 4
-; CHECK-PREPARE-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[ACCUM_I_FLAT]], i32 0
-; CHECK-PREPARE-NEXT: [[BITCAST:%.*]] = bitcast ptr [[ACCUM_I_FLAT]] to ptr
-; CHECK-PREPARE-NEXT: call void @llvm.lifetime.start.p0(i64 4, ptr nonnull [[BITCAST]])
-; CHECK-PREPARE-NEXT: store i32 0, ptr [[GEP]], align 4
-; CHECK-PREPARE-NEXT: [[BITCAST:%.*]] = bitcast ptr [[ACCUM_I_FLAT]] to ptr
-; CHECK-PREPARE-NEXT: call void @llvm.lifetime.end.p0(i64 4, ptr nonnull [[BITCAST]])
-;
-; CHECK-NEXT: ret void
+; CHECK-NEXT: [[ACCUM_I_FLAT:%.*]] = alloca [1 x i32], align 4
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[ACCUM_I_FLAT]], i32 0
+; CHECK-SM63-NEXT: store [1 x i32] undef, ptr [[ACCUM_I_FLAT]], align 4
+; CHECK-SM66-NEXT: call void @llvm.lifetime.start.p0(i64 4, ptr nonnull [[ACCUM_I_FLAT]])
+; CHECK-EMBED-NOT: bitcast
+; CHECK-EMBED-NOT: lifetime
+; CHECK-NEXT: store i32 0, ptr [[GEP]], align 4
+; CHECK-SM63-NEXT: store [1 x i32] undef, ptr [[ACCUM_I_FLAT]], align 4
+; CHECK-SM66-NEXT: call void @llvm.lifetime.end.p0(i64 4, ptr nonnull [[ACCUM_I_FLAT]])
+; CHECK-EMBED-NOT: bitcast
+; CHECK-EMBED-NOT: lifetime
+; CHECK-NEXT: ret void
;
define void @test_legal_lifetime() {
%accum.i.flat = alloca [1 x i32], align 4
@@ -35,22 +32,6 @@ define void @test_legal_lifetime() {
ret void
}
-; CHECK-PREPARE-DAG: attributes [[LIFETIME_ATTRS:#.*]] = { nounwind }
-
-; CHECK-PREPARE-DAG: ; Function Attrs: nounwind
-; CHECK-PREPARE-DAG: declare void @llvm.lifetime.start.p0(i64, ptr) [[LIFETIME_ATTRS]]
-
-; CHECK-PREPARE-DAG: ; Function Attrs: nounwind
-; CHECK-PREPARE-DAG: declare void @llvm.lifetime.end.p0(i64, ptr) [[LIFETIME_ATTRS]]
-
-; Function Attrs: nounwind memory(argmem: readwrite)
-declare void @llvm.lifetime.end.p0(i64, ptr) #0
-
-; Function Attrs: nounwind memory(argmem: readwrite)
-declare void @llvm.lifetime.start.p0(i64, ptr) #0
-
-attributes #0 = { nounwind memory(argmem: readwrite) }
-
; Set the validator version to 1.6
!dx.valver = !{!0}
!0 = !{i32 1, i32 6}
generated by cgit v1.1 at 2025-07-24 04:20:56 +0000