anna updated this revision to Diff 251615.
anna added a comment.
use isGuaranteedToTransferExecutionToSuccessor instead of `MayThrow`
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D76140/new/
https://reviews.llvm.org/D76140
Files:
clang/test/CodeGen/builtins-systemz-zvector.c
clang/test/CodeGen/builtins-systemz-zvector2.c
clang/test/CodeGen/movbe-builtins.c
clang/test/CodeGen/rot-intrinsics.c
clang/test/CodeGen/waitpkg.c
llvm/lib/Transforms/Utils/InlineFunction.cpp
llvm/test/Transforms/Inline/ret_attr_update.ll
Index: llvm/test/Transforms/Inline/ret_attr_update.ll
===================================================================
--- /dev/null
+++ llvm/test/Transforms/Inline/ret_attr_update.ll
@@ -0,0 +1,75 @@
+; RUN: opt < %s -inline-threshold=0 -always-inline -S | FileCheck %s
+; RUN: opt < %s -passes=always-inline -S | FileCheck %s
+
+declare i8* @foo(i8*)
+declare i8* @bar(i8*)
+
+define i8* @callee(i8 *%p) alwaysinline {
+; CHECK: @callee(
+; CHECK: call i8* @foo(i8* noalias %p)
+ %r = call i8* @foo(i8* noalias %p)
+ ret i8* %r
+}
+
+define i8* @caller(i8* %ptr, i64 %x) {
+; CHECK-LABEL: @caller
+; CHECK: call nonnull i8* @foo(i8* noalias
+ %gep = getelementptr inbounds i8, i8* %ptr, i64 %x
+ %p = call nonnull i8* @callee(i8* %gep)
+ ret i8* %p
+}
+
+declare void @llvm.experimental.guard(i1,...)
+; Cannot add nonnull attribute to foo.
+define internal i8* @callee_with_throwable(i8* %p) alwaysinline {
+; CHECK-NOT: callee_with_throwable
+ %r = call i8* @foo(i8* %p)
+ %cond = icmp ne i8* %r, null
+ call void (i1, ...) @llvm.experimental.guard(i1 %cond) [ "deopt"() ]
+ ret i8* %r
+}
+
+; Here also we cannot add nonnull attribute to the call bar.
+define internal i8* @callee_with_explicit_control_flow(i8* %p) alwaysinline {
+; CHECK-NOT: callee_with_explicit_control_flow
+ %r = call i8* @bar(i8* %p)
+ %cond = icmp ne i8* %r, null
+ br i1 %cond, label %ret, label %orig
+
+ret:
+ ret i8* %r
+
+orig:
+ ret i8* %p
+}
+
+define i8* @caller2(i8* %ptr, i64 %x, i1 %cond) {
+; CHECK-LABEL: @caller2
+; CHECK: call i8* @foo
+; CHECK: call i8* @bar
+ %gep = getelementptr inbounds i8, i8* %ptr, i64 %x
+ %p = call nonnull i8* @callee_with_throwable(i8* %gep)
+ %q = call nonnull i8* @callee_with_explicit_control_flow(i8* %gep)
+ br i1 %cond, label %pret, label %qret
+
+pret:
+ ret i8* %p
+
+qret:
+ ret i8* %q
+}
+
+define internal i8* @callee3(i8 *%p) alwaysinline {
+; CHECK-NOT: callee3
+ %r = call noalias i8* @foo(i8* %p)
+ ret i8* %r
+}
+
+; add the deref attribute to the existing attributes on foo.
+define i8* @caller3(i8* %ptr, i64 %x) {
+; CHECK-LABEL: caller3
+; CHECK: call noalias dereferenceable_or_null(12) i8* @foo
+ %gep = getelementptr inbounds i8, i8* %ptr, i64 %x
+ %p = call dereferenceable_or_null(12) i8* @callee3(i8* %gep)
+ ret i8* %p
+}
Index: llvm/lib/Transforms/Utils/InlineFunction.cpp
===================================================================
--- llvm/lib/Transforms/Utils/InlineFunction.cpp
+++ llvm/lib/Transforms/Utils/InlineFunction.cpp
@@ -80,11 +80,21 @@
cl::Hidden,
cl::desc("Convert noalias attributes to metadata during inlining."));
+static cl::opt<bool> UpdateReturnAttributes(
+ "update-return-attrs", cl::init(true), cl::Hidden,
+ cl::desc("Update return attributes on calls within inlined body"));
+
static cl::opt<bool>
PreserveAlignmentAssumptions("preserve-alignment-assumptions-during-inlining",
cl::init(true), cl::Hidden,
cl::desc("Convert align attributes to assumptions during inlining."));
+static cl::opt<unsigned> MaxInstCheckedForThrow(
+ "max-inst-checked-for-throw-during-inlining", cl::Hidden,
+ cl::desc("the maximum number of instructions analyzed for may throw during "
+ "attribute inference in inlined body"),
+ cl::init(4));
+
llvm::InlineResult llvm::InlineFunction(CallBase *CB, InlineFunctionInfo &IFI,
AAResults *CalleeAAR,
bool InsertLifetime) {
@@ -1136,6 +1146,80 @@
}
}
+static void AddReturnAttributes(CallSite CS, ValueToValueMapTy &VMap) {
+ if (!UpdateReturnAttributes)
+ return;
+ AttrBuilder AB(CS.getAttributes(), AttributeList::ReturnIndex);
+ if (AB.empty())
+ return;
+
+ auto *CalledFunction = CS.getCalledFunction();
+ auto &Context = CalledFunction->getContext();
+
+ auto GetClonedValue = [&](Instruction *I) -> Value * {
+ ValueToValueMapTy::iterator VMI = VMap.find(I);
+ if (VMI == VMap.end())
+ return nullptr;
+ return VMI->second;
+ };
+
+ auto MayContainThrowingOrExitingCall = [&](Instruction *RVal,
+ Instruction *RInst) {
+ unsigned NumInstChecked = 0;
+ for (auto &I :
+ make_range(RVal->getIterator(), std::prev(RInst->getIterator())))
+ if (NumInstChecked++ > MaxInstCheckedForThrow ||
+ isGuaranteedToTransferExecutionToSuccessor(&I))
+ return true;
+ return false;
+ };
+
+ for (auto &BB : *CalledFunction) {
+ auto *RI = dyn_cast<ReturnInst>(BB.getTerminator());
+ if (!RI || !isa<CallBase>(RI->getOperand(0)))
+ continue;
+ // Sanity check that the cloned return instruction exists and is a return
+ // instruction itself.
+ auto *NewRI = dyn_cast_or_null<ReturnInst>(GetClonedValue(RI));
+ if (!NewRI)
+ continue;
+ auto *RetVal = cast<CallBase>(RI->getOperand(0));
+ // Sanity check that the cloned RetVal exists and is a call.
+ // Simplification during inlining could have transformed the cloned
+ // instruction.
+ auto *NewRetVal = dyn_cast_or_null<CallBase>(GetClonedValue(RetVal));
+ if (!NewRetVal)
+ continue;
+ // Backward propagation of attributes to the returned value may be incorrect
+ // if it is control flow dependent.
+ // Consider:
+ // @callee {
+ // %rv = call @foo()
+ // %rv2 = call @bar()
+ // if (%rv2 != null)
+ // return %rv2
+ // if (%rv == null)
+ // exit()
+ // return %rv
+ // }
+ // caller() {
+ // %val = call nonnull @callee()
+ // }
+ // Here we cannot add the nonnull attribute on either foo or bar. So, we
+ // limit the check to both NewRetVal and NewRI are in the same basic block
+ // and there are no throwing/exiting instructions between these
+ // instructions.
+ if (NewRI->getParent() != NewRetVal->getParent() ||
+ MayContainThrowingOrExitingCall(NewRetVal, NewRI))
+ continue;
+ // Add to the existing attributes.
+ AttributeList AL = NewRetVal->getAttributes();
+ AttributeList NewAL =
+ AL.addAttributes(Context, AttributeList::ReturnIndex, AB);
+ NewRetVal->setAttributes(NewAL);
+ }
+}
+
/// If the inlined function has non-byval align arguments, then
/// add @llvm.assume-based alignment assumptions to preserve this information.
static void AddAlignmentAssumptions(CallSite CS, InlineFunctionInfo &IFI) {
@@ -1801,6 +1885,10 @@
// Add noalias metadata if necessary.
AddAliasScopeMetadata(CS, VMap, DL, CalleeAAR);
+ // Clone return attributes on the callsite into the calls within the inlined
+ // function which feed into its return value.
+ AddReturnAttributes(CS, VMap);
+
// Propagate llvm.mem.parallel_loop_access if necessary.
PropagateParallelLoopAccessMetadata(CS, VMap);
Index: clang/test/CodeGen/waitpkg.c
===================================================================
--- clang/test/CodeGen/waitpkg.c
+++ clang/test/CodeGen/waitpkg.c
@@ -14,12 +14,12 @@
uint8_t test_umwait(uint32_t control, uint64_t counter) {
//CHECK-LABEL: @test_umwait
- //CHECK: call i8 @llvm.x86.umwait(i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
+ //CHECK: call zeroext i8 @llvm.x86.umwait(i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
return _umwait(control, counter);
}
uint8_t test_tpause(uint32_t control, uint64_t counter) {
//CHECK-LABEL: @test_tpause
- //CHECK: call i8 @llvm.x86.tpause(i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
+ //CHECK: call zeroext i8 @llvm.x86.tpause(i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
return _tpause(control, counter);
}
Index: clang/test/CodeGen/rot-intrinsics.c
===================================================================
--- clang/test/CodeGen/rot-intrinsics.c
+++ clang/test/CodeGen/rot-intrinsics.c
@@ -1,9 +1,9 @@
-// RUN: %clang_cc1 -ffreestanding -triple i686--linux -emit-llvm %s -o - | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
-// RUN: %clang_cc1 -ffreestanding -triple x86_64--linux -emit-llvm %s -o - | FileCheck %s --check-prefixes CHECK,CHECK-64BIT-LONG
-// RUN: %clang_cc1 -fms-extensions -fms-compatibility -ffreestanding %s -triple=i686-windows-msvc -target-feature +sse2 -emit-llvm -o - -Wall -Werror | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
-// RUN: %clang_cc1 -fms-extensions -fms-compatibility -ffreestanding %s -triple=x86_64-windows-msvc -target-feature +sse2 -emit-llvm -o - -Wall -Werror | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
-// RUN: %clang_cc1 -fms-extensions -fms-compatibility -fms-compatibility-version=17.00 -ffreestanding %s -triple=i686-windows-msvc -target-feature +sse2 -emit-llvm -o - -Wall -Werror | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
-// RUN: %clang_cc1 -fms-extensions -fms-compatibility -fms-compatibility-version=17.00 -ffreestanding %s -triple=x86_64-windows-msvc -target-feature +sse2 -emit-llvm -o - -Wall -Werror | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
+// RUN: %clang_cc1 -ffreestanding -triple i686--linux -emit-llvm -mllvm -update-return-attrs=false %s -o - | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
+// RUN: %clang_cc1 -ffreestanding -triple x86_64--linux -emit-llvm -mllvm -update-return-attrs=false %s -o - | FileCheck %s --check-prefixes CHECK,CHECK-64BIT-LONG
+// RUN: %clang_cc1 -fms-extensions -fms-compatibility -ffreestanding %s -triple=i686-windows-msvc -target-feature +sse2 -emit-llvm -mllvm -update-return-attrs=false -o - -Wall -Werror | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
+// RUN: %clang_cc1 -fms-extensions -fms-compatibility -ffreestanding %s -triple=x86_64-windows-msvc -target-feature +sse2 -emit-llvm -mllvm -update-return-attrs=false -o - -Wall -Werror | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
+// RUN: %clang_cc1 -fms-extensions -fms-compatibility -fms-compatibility-version=17.00 -ffreestanding %s -triple=i686-windows-msvc -target-feature +sse2 -emit-llvm -mllvm -update-return-attrs=false -o - -Wall -Werror | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
+// RUN: %clang_cc1 -fms-extensions -fms-compatibility -fms-compatibility-version=17.00 -ffreestanding %s -triple=x86_64-windows-msvc -target-feature +sse2 -emit-llvm -mllvm -update-return-attrs=false -o - -Wall -Werror | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
#include <x86intrin.h>
Index: clang/test/CodeGen/movbe-builtins.c
===================================================================
--- clang/test/CodeGen/movbe-builtins.c
+++ clang/test/CodeGen/movbe-builtins.c
@@ -7,7 +7,7 @@
short test_loadbe_i16(const short *P) {
// CHECK-LABEL: @test_loadbe_i16
// CHECK: [[LOAD:%.*]] = load i16, i16* %{{.*}}, align 1
- // CHECK: call i16 @llvm.bswap.i16(i16 [[LOAD]])
+ // CHECK: call signext i16 @llvm.bswap.i16(i16 [[LOAD]])
return _loadbe_i16(P);
}
Index: clang/test/CodeGen/builtins-systemz-zvector2.c
===================================================================
--- clang/test/CodeGen/builtins-systemz-zvector2.c
+++ clang/test/CodeGen/builtins-systemz-zvector2.c
@@ -654,10 +654,10 @@
// CHECK-ASM: vsrlb
idx = vec_test_mask(vf, vui);
- // CHECK: call i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
+ // CHECK: call signext i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
// CHECK-ASM: vtm
idx = vec_test_mask(vd, vul);
- // CHECK: call i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
+ // CHECK: call signext i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
// CHECK-ASM: vtm
vuc = vec_msum_u128(vul, vul, vuc, 0);
Index: clang/test/CodeGen/builtins-systemz-zvector.c
===================================================================
--- clang/test/CodeGen/builtins-systemz-zvector.c
+++ clang/test/CodeGen/builtins-systemz-zvector.c
@@ -3665,31 +3665,31 @@
// CHECK-ASM: vsumqg
idx = vec_test_mask(vsc, vuc);
- // CHECK: call i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
+ // CHECK: call signext i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
// CHECK-ASM: vtm
idx = vec_test_mask(vuc, vuc);
- // CHECK: call i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
+ // CHECK: call signext i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
// CHECK-ASM: vtm
idx = vec_test_mask(vss, vus);
- // CHECK: call i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
+ // CHECK: call signext i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
// CHECK-ASM: vtm
idx = vec_test_mask(vus, vus);
- // CHECK: call i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
+ // CHECK: call signext i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
// CHECK-ASM: vtm
idx = vec_test_mask(vsi, vui);
- // CHECK: call i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
+ // CHECK: call signext i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
// CHECK-ASM: vtm
idx = vec_test_mask(vui, vui);
- // CHECK: call i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
+ // CHECK: call signext i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
// CHECK-ASM: vtm
idx = vec_test_mask(vsl, vul);
- // CHECK: call i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
+ // CHECK: call signext i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
// CHECK-ASM: vtm
idx = vec_test_mask(vul, vul);
- // CHECK: call i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
+ // CHECK: call signext i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
// CHECK-ASM: vtm
idx = vec_test_mask(vd, vul);
- // CHECK: call i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
+ // CHECK: call signext i32 @llvm.s390.vtm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
// CHECK-ASM: vtm
}
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
