diff options
Diffstat (limited to 'llvm/test/Transforms/Inline')
5 files changed, 184 insertions, 3 deletions
diff --git a/llvm/test/Transforms/Inline/RISCV/inline-target-features.ll b/llvm/test/Transforms/Inline/RISCV/inline-target-features.ll new file mode 100644 index 0000000..b626a22 --- /dev/null +++ b/llvm/test/Transforms/Inline/RISCV/inline-target-features.ll @@ -0,0 +1,34 @@ +; RUN: opt < %s -mtriple=riscv64-unknown-linux-gnu -S -passes=inline | FileCheck %s +; RUN: opt < %s -mtriple=riscv64-unknown-linux-gnu -S -passes='cgscc(inline)' | FileCheck %s +; Check that we only inline when we have compatible target attributes. + +target datalayout = "e-m:e-p:64:64-i64:64-i128:128-n64-S128" +target triple = "riscv64-unknown-linux-gnu" + +define i32 @foo() #0 { +entry: + %call = call i32 (...) @baz() + ret i32 %call +; CHECK-LABEL: foo +; CHECK: call i32 (...) @baz() +} +declare i32 @baz(...) #0 + +define i32 @bar() #1 { +entry: + %call = call i32 @foo() + ret i32 %call +; CHECK-LABEL: bar +; CHECK: call i32 (...) @baz() +} + +define i32 @qux() #0 { +entry: + %call = call i32 @bar() + ret i32 %call +; CHECK-LABEL: qux +; CHECK: call i32 @bar() +} + +attributes #0 = { "target-cpu"="generic-rv64" "target-features"="+f,+d" } +attributes #1 = { "target-cpu"="generic-rv64" "target-features"="+f,+d,+m,+v" } diff --git a/llvm/test/Transforms/Inline/RISCV/lit.local.cfg b/llvm/test/Transforms/Inline/RISCV/lit.local.cfg new file mode 100644 index 0000000..1735174 --- /dev/null +++ b/llvm/test/Transforms/Inline/RISCV/lit.local.cfg @@ -0,0 +1,2 @@ +if not "RISCV" in config.root.targets: + config.unsupported = True diff --git a/llvm/test/Transforms/Inline/devirtualize-4.ll b/llvm/test/Transforms/Inline/devirtualize-4.ll index fae3643..d29360f 100644 --- a/llvm/test/Transforms/Inline/devirtualize-4.ll +++ b/llvm/test/Transforms/Inline/devirtualize-4.ll @@ -60,7 +60,7 @@ declare void @llvm.lifetime.start.p0(i64 immarg, ptr nocapture) define linkonce_odr dso_local void @_ZN4ImplC2Ev(ptr %this) unnamed_addr align 2 { entry: call void @_ZN9InterfaceC2Ev(ptr %this) - store ptr getelementptr inbounds ({ [3 x ptr] }, ptr @_ZTV4Impl, i64 0, inrange i32 0, i64 2), ptr %this, align 8 + store ptr getelementptr inbounds inrange(-16, 8) ({ [3 x ptr] }, ptr @_ZTV4Impl, i64 0, i32 0, i64 2), ptr %this, align 8 %f = getelementptr inbounds %class.Impl, ptr %this, i64 0, i32 1 store i32 3, ptr %f, align 8 ret void @@ -78,7 +78,7 @@ declare void @llvm.lifetime.end.p0(i64 immarg, ptr nocapture) define linkonce_odr dso_local void @_ZN9InterfaceC2Ev(ptr %this) unnamed_addr align 2 { entry: - store ptr getelementptr inbounds ({ [3 x ptr] }, ptr @_ZTV9Interface, i64 0, inrange i32 0, i64 2), ptr %this, align 8 + store ptr getelementptr inbounds inrange(-16, 8) ({ [3 x ptr] }, ptr @_ZTV9Interface, i64 0, i32 0, i64 2), ptr %this, align 8 ret void } @@ -185,7 +185,7 @@ entry: define linkonce_odr void @_ZN1AC2Ev(ptr %this) align 2 { entry: - store ptr getelementptr inbounds ({ [4 x ptr] }, ptr @_ZTV1A, i64 0, inrange i32 0, i64 2), ptr %this, align 8 + store ptr getelementptr inbounds inrange(-16, 8) ({ [4 x ptr] }, ptr @_ZTV1A, i64 0, i32 0, i64 2), ptr %this, align 8 ret void } diff --git a/llvm/test/Transforms/Inline/update_invoke_prof.ll b/llvm/test/Transforms/Inline/update_invoke_prof.ll new file mode 100644 index 0000000..5f09c7c --- /dev/null +++ b/llvm/test/Transforms/Inline/update_invoke_prof.ll @@ -0,0 +1,64 @@ +; A pre-commit test to show that branch weights and value profiles associated with invoke are not updated. +; RUN: opt < %s -passes='require<profile-summary>,cgscc(inline)' -S | FileCheck %s + +declare i32 @__gxx_personality_v0(...) + +define void @caller(ptr %func) personality ptr @__gxx_personality_v0 !prof !15 { + call void @callee(ptr %func), !prof !16 + ret void +} + +declare void @inner_callee(ptr %func) + +define void @callee(ptr %func) personality ptr @__gxx_personality_v0 !prof !17 { + invoke void %func() + to label %next unwind label %lpad, !prof !18 + +next: + invoke void @inner_callee(ptr %func) + to label %ret unwind label %lpad, !prof !19 + +lpad: + %exn = landingpad {ptr, i32} + cleanup + unreachable + +ret: + ret void +} + +!llvm.module.flags = !{!1} +!1 = !{i32 1, !"ProfileSummary", !2} +!2 = !{!3, !4, !5, !6, !7, !8, !9, !10} +!3 = !{!"ProfileFormat", !"SampleProfile"} +!4 = !{!"TotalCount", i64 10000} +!5 = !{!"MaxCount", i64 10} +!6 = !{!"MaxInternalCount", i64 1} +!7 = !{!"MaxFunctionCount", i64 2000} +!8 = !{!"NumCounts", i64 2} +!9 = !{!"NumFunctions", i64 2} +!10 = !{!"DetailedSummary", !11} +!11 = !{!12, !13, !14} +!12 = !{i32 10000, i64 100, i32 1} +!13 = !{i32 999000, i64 100, i32 1} +!14 = !{i32 999999, i64 1, i32 2} +!15 = !{!"function_entry_count", i64 1000} +!16 = !{!"branch_weights", i64 1000} +!17 = !{!"function_entry_count", i32 1500} +!18 = !{!"VP", i32 0, i64 1500, i64 123, i64 900, i64 456, i64 600} +!19 = !{!"branch_weights", i32 1500} + +; CHECK-LABEL: @caller( +; CHECK: invoke void %func( +; CHECK-NEXT: {{.*}} !prof ![[PROF1:[0-9]+]] +; CHECK: invoke void @inner_callee( +; CHECK-NEXT: {{.*}} !prof ![[PROF2:[0-9]+]] + +; CHECK-LABL: @callee( +; CHECK: invoke void %func( +; CHECK-NEXT: {{.*}} !prof ![[PROF1]] +; CHECK: invoke void @inner_callee( +; CHECK-NEXT: {{.*}} !prof ![[PROF2]] + +; CHECK: ![[PROF1]] = !{!"VP", i32 0, i64 1500, i64 123, i64 900, i64 456, i64 600} +; CHECK: ![[PROF2]] = !{!"branch_weights", i32 1500} diff --git a/llvm/test/Transforms/Inline/update_value_profile.ll b/llvm/test/Transforms/Inline/update_value_profile.ll new file mode 100644 index 0000000..daa95e9 --- /dev/null +++ b/llvm/test/Transforms/Inline/update_value_profile.ll @@ -0,0 +1,81 @@ +; RUN: opt < %s -passes='require<profile-summary>,cgscc(inline)' -inline-threshold=100 -S | FileCheck %s +target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128" +target triple = "x86_64-unknown-linux-gnu" + +; When 'callee' is inlined into caller1 and caller2, the indirect call value +; profiles of the inlined copy should be scaled based on callers' profiles, +; and the indirect call value profiles in 'callee' should be updated. +define i32 @callee(ptr %0, i32 %1) !prof !20 { +; CHECK-LABEL: define i32 @callee( +; CHECK-SAME: ptr [[TMP0:%.*]], i32 [[TMP1:%.*]]) !prof [[PROF0:![0-9]+]] { +; CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP0]], align 8 +; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP3]], i64 8 +; CHECK-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8 +; CHECK-NEXT: [[TMP6:%.*]] = tail call i32 [[TMP5]](ptr [[TMP0]], i32 [[TMP1]]), !prof [[PROF1:![0-9]+]] +; CHECK-NEXT: ret i32 [[TMP6]] +; + %3 = load ptr, ptr %0 + %5 = getelementptr inbounds i8, ptr %3, i64 8 + %6 = load ptr, ptr %5 + %7 = tail call i32 %6(ptr %0, i32 %1), !prof !17 + ret i32 %7 +} + +define i32 @caller1(i32 %0) !prof !18 { +; CHECK-LABEL: define i32 @caller1( +; CHECK-SAME: i32 [[TMP0:%.*]]) !prof [[PROF2:![0-9]+]] { +; CHECK-NEXT: [[TMP2:%.*]] = tail call ptr @_Z10createTypei(i32 [[TMP0]]) +; CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 8 +; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP3]], i64 8 +; CHECK-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8 +; CHECK-NEXT: [[TMP6:%.*]] = tail call i32 [[TMP5]](ptr [[TMP2]], i32 [[TMP0]]), !prof [[PROF3:![0-9]+]] +; CHECK-NEXT: ret i32 [[TMP6]] +; + %2 = tail call ptr @_Z10createTypei(i32 %0) + %3 = tail call i32 @callee(ptr %2, i32 %0) + ret i32 %3 +} + +define i32 @caller2(i32 %0) !prof !19 { +; CHECK-LABEL: define i32 @caller2( +; CHECK-SAME: i32 [[TMP0:%.*]]) !prof [[PROF4:![0-9]+]] { +; CHECK-NEXT: [[TMP2:%.*]] = tail call ptr @_Z10createTypei(i32 [[TMP0]]) +; CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 8 +; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP3]], i64 8 +; CHECK-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8 +; CHECK-NEXT: [[TMP6:%.*]] = tail call i32 [[TMP5]](ptr [[TMP2]], i32 [[TMP0]]), !prof [[PROF5:![0-9]+]] +; CHECK-NEXT: ret i32 [[TMP6]] +; + %2 = tail call ptr @_Z10createTypei(i32 %0) + %3 = tail call i32 @callee(ptr %2, i32 %0) + ret i32 %3 +} + +declare ptr @_Z10createTypei(i32) + +!1 = !{i32 1, !"ProfileSummary", !2} +!2 = !{!3, !4, !5, !6, !7, !8, !9, !10} +!3 = !{!"ProfileFormat", !"InstrProf"} +!4 = !{!"TotalCount", i64 10000} +!5 = !{!"MaxCount", i64 10} +!6 = !{!"MaxInternalCount", i64 1} +!7 = !{!"MaxFunctionCount", i64 1000} +!8 = !{!"NumCounts", i64 3} +!9 = !{!"NumFunctions", i64 3} +!10 = !{!"DetailedSummary", !11} +!11 = !{!12, !13, !14} +!12 = !{i32 10000, i64 100, i32 1} +!13 = !{i32 999000, i64 100, i32 1} +!14 = !{i32 999999, i64 1, i32 2} +!17 = !{!"VP", i32 0, i64 1600, i64 123, i64 1000, i64 456, i64 600} +!18 = !{!"function_entry_count", i64 1000} +!19 = !{!"function_entry_count", i64 600} +!20 = !{!"function_entry_count", i64 1700} +;. +; CHECK: [[PROF0]] = !{!"function_entry_count", i64 100} +; CHECK: [[PROF1]] = !{!"VP", i32 0, i64 94, i64 123, i64 58, i64 456, i64 35} +; CHECK: [[PROF2]] = !{!"function_entry_count", i64 1000} +; CHECK: [[PROF3]] = !{!"VP", i32 0, i64 941, i64 123, i64 588, i64 456, i64 352} +; CHECK: [[PROF4]] = !{!"function_entry_count", i64 600} +; CHECK: [[PROF5]] = !{!"VP", i32 0, i64 564, i64 123, i64 352, i64 456, i64 211} +;. |