[clang] [clang][fmv] Drop .ifunc from target_version's entrypoint's mangling (PR #81194)
Jon Roelofs via cfe-commits
cfe-commits at lists.llvm.org
Thu Feb 8 16:32:45 PST 2024
https://github.com/jroelofs updated https://github.com/llvm/llvm-project/pull/81194
>From 92aef5a078583f69cede47374e2b97de865c5c5d Mon Sep 17 00:00:00 2001
From: Jon Roelofs <jonathan_roelofs at apple.com>
Date: Thu, 8 Feb 2024 13:56:50 -0800
Subject: [PATCH 1/4] [clang][fmv] Drop .ifunc from target_version's
entrypoint's mangling
Fixes: https://github.com/llvm/llvm-project/issues/81043
---
clang/include/clang/AST/Decl.h | 4 +
clang/lib/AST/Decl.cpp | 4 +
clang/lib/CodeGen/CodeGenModule.cpp | 4 +-
clang/test/CodeGen/attr-target-version.c | 129 +++++++++---------
clang/test/CodeGenCXX/attr-target-version.cpp | 87 ++++++------
5 files changed, 123 insertions(+), 105 deletions(-)
diff --git a/clang/include/clang/AST/Decl.h b/clang/include/clang/AST/Decl.h
index f26fb5ad5f1331..42fdf2b56dc278 100644
--- a/clang/include/clang/AST/Decl.h
+++ b/clang/include/clang/AST/Decl.h
@@ -2619,6 +2619,10 @@ class FunctionDecl : public DeclaratorDecl,
/// the target-clones functionality.
bool isTargetClonesMultiVersion() const;
+ /// True if this function is a multiversioned dispatch function as a part of
+ /// the target-version functionality.
+ bool isTargetVersionMultiVersion() const;
+
/// \brief Get the associated-constraints of this function declaration.
/// Currently, this will either be a vector of size 1 containing the
/// trailing-requires-clause or an empty vector.
diff --git a/clang/lib/AST/Decl.cpp b/clang/lib/AST/Decl.cpp
index 26fdfa040796ed..40e2903265bf3c 100644
--- a/clang/lib/AST/Decl.cpp
+++ b/clang/lib/AST/Decl.cpp
@@ -3541,6 +3541,10 @@ bool FunctionDecl::isTargetClonesMultiVersion() const {
return isMultiVersion() && hasAttr<TargetClonesAttr>();
}
+bool FunctionDecl::isTargetVersionMultiVersion() const {
+ return isMultiVersion() && hasAttr<TargetVersionAttr>();
+}
+
void
FunctionDecl::setPreviousDeclaration(FunctionDecl *PrevDecl) {
redeclarable_base::setPreviousDecl(PrevDecl);
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp b/clang/lib/CodeGen/CodeGenModule.cpp
index 36b63d78b06f83..364553e2a6f0f7 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -4212,7 +4212,7 @@ void CodeGenModule::emitMultiVersionFunctions() {
llvm::Constant *ResolverConstant = GetOrCreateMultiVersionResolver(GD);
if (auto *IFunc = dyn_cast<llvm::GlobalIFunc>(ResolverConstant)) {
ResolverConstant = IFunc->getResolver();
- if (FD->isTargetClonesMultiVersion()) {
+ if (FD->isTargetClonesMultiVersion() || FD->isTargetVersionMultiVersion()) {
const CGFunctionInfo &FI = getTypes().arrangeGlobalDeclaration(GD);
llvm::FunctionType *DeclTy = getTypes().GetFunctionType(FI);
std::string MangledName = getMangledNameImpl(
@@ -4393,7 +4393,7 @@ llvm::Constant *CodeGenModule::GetOrCreateMultiVersionResolver(GlobalDecl GD) {
// a separate resolver).
std::string ResolverName = MangledName;
if (getTarget().supportsIFunc()) {
- if (!FD->isTargetClonesMultiVersion())
+ if (!FD->isTargetClonesMultiVersion() && !FD->isTargetVersionMultiVersion())
ResolverName += ".ifunc";
} else if (FD->isTargetMultiVersion()) {
ResolverName += ".resolver";
diff --git a/clang/test/CodeGen/attr-target-version.c b/clang/test/CodeGen/attr-target-version.c
index 2a96697e4291b9..c27d48f3ecf681 100644
--- a/clang/test/CodeGen/attr-target-version.c
+++ b/clang/test/CodeGen/attr-target-version.c
@@ -90,13 +90,20 @@ int hoo(void) {
//.
// CHECK: @__aarch64_cpu_features = external dso_local global { i64 }
-// CHECK: @fmv.ifunc = weak_odr ifunc i32 (), ptr @fmv.resolver
-// CHECK: @fmv_one.ifunc = weak_odr ifunc i32 (), ptr @fmv_one.resolver
-// CHECK: @fmv_two.ifunc = weak_odr ifunc i32 (), ptr @fmv_two.resolver
-// CHECK: @fmv_e.ifunc = weak_odr ifunc i32 (), ptr @fmv_e.resolver
-// CHECK: @fmv_c.ifunc = weak_odr ifunc void (), ptr @fmv_c.resolver
-// CHECK: @fmv_inline.ifunc = weak_odr ifunc i32 (), ptr @fmv_inline.resolver
-// CHECK: @fmv_d.ifunc = internal ifunc i32 (), ptr @fmv_d.resolver
+// CHECK: @fmv.ifunc = weak_odr alias i32 (), ptr @fmv
+// CHECK: @fmv_one.ifunc = weak_odr alias i32 (), ptr @fmv_one
+// CHECK: @fmv_two.ifunc = weak_odr alias i32 (), ptr @fmv_two
+// CHECK: @fmv_e.ifunc = weak_odr alias i32 (), ptr @fmv_e
+// CHECK: @fmv_inline.ifunc = weak_odr alias i32 (), ptr @fmv_inline
+// CHECK: @fmv_d.ifunc = internal alias i32 (), ptr @fmv_d
+// CHECK: @fmv_c.ifunc = weak_odr alias void (), ptr @fmv_c
+// CHECK: @fmv = weak_odr ifunc i32 (), ptr @fmv.resolver
+// CHECK: @fmv_one = weak_odr ifunc i32 (), ptr @fmv_one.resolver
+// CHECK: @fmv_two = weak_odr ifunc i32 (), ptr @fmv_two.resolver
+// CHECK: @fmv_e = weak_odr ifunc i32 (), ptr @fmv_e.resolver
+// CHECK: @fmv_inline = weak_odr ifunc i32 (), ptr @fmv_inline.resolver
+// CHECK: @fmv_d = internal ifunc i32 (), ptr @fmv_d.resolver
+// CHECK: @fmv_c = weak_odr ifunc void (), ptr @fmv_c.resolver
//.
// CHECK: Function Attrs: noinline nounwind optnone
// CHECK-LABEL: define {{[^@]+}}@fmv._MrngMflagmMfp16fml
@@ -105,6 +112,32 @@ int hoo(void) {
// CHECK-NEXT: ret i32 1
//
//
+// CHECK: Function Attrs: noinline nounwind optnone
+// CHECK-LABEL: define {{[^@]+}}@fmv_one._MsimdMls64
+// CHECK-SAME: () #[[ATTR1:[0-9]+]] {
+// CHECK-NEXT: entry:
+// CHECK-NEXT: ret i32 1
+//
+//
+// CHECK: Function Attrs: noinline nounwind optnone
+// CHECK-LABEL: define {{[^@]+}}@fmv_two._Mfp
+// CHECK-SAME: () #[[ATTR1]] {
+// CHECK-NEXT: entry:
+// CHECK-NEXT: ret i32 1
+//
+//
+// CHECK: Function Attrs: noinline nounwind optnone
+// CHECK-LABEL: define {{[^@]+}}@foo
+// CHECK-SAME: () #[[ATTR2:[0-9]+]] {
+// CHECK-NEXT: entry:
+// CHECK-NEXT: [[CALL:%.*]] = call i32 @fmv()
+// CHECK-NEXT: [[CALL1:%.*]] = call i32 @fmv_one()
+// CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[CALL]], [[CALL1]]
+// CHECK-NEXT: [[CALL2:%.*]] = call i32 @fmv_two()
+// CHECK-NEXT: [[ADD3:%.*]] = add nsw i32 [[ADD]], [[CALL2]]
+// CHECK-NEXT: ret i32 [[ADD3]]
+//
+//
// CHECK-LABEL: define {{[^@]+}}@fmv.resolver() comdat {
// CHECK-NEXT: resolver_entry:
// CHECK-NEXT: call void @__init_cpu_features_resolver()
@@ -183,42 +216,16 @@ int hoo(void) {
// CHECK-NEXT: ret ptr @fmv.default
//
//
-// CHECK: Function Attrs: noinline nounwind optnone
-// CHECK-LABEL: define {{[^@]+}}@fmv_one._MsimdMls64
-// CHECK-SAME: () #[[ATTR1:[0-9]+]] {
-// CHECK-NEXT: entry:
-// CHECK-NEXT: ret i32 1
-//
-//
// CHECK-LABEL: define {{[^@]+}}@fmv_one.resolver() comdat {
// CHECK-NEXT: resolver_entry:
// CHECK-NEXT: ret ptr @fmv_one._MsimdMls64
//
//
-// CHECK: Function Attrs: noinline nounwind optnone
-// CHECK-LABEL: define {{[^@]+}}@fmv_two._Mfp
-// CHECK-SAME: () #[[ATTR1]] {
-// CHECK-NEXT: entry:
-// CHECK-NEXT: ret i32 1
-//
-//
// CHECK-LABEL: define {{[^@]+}}@fmv_two.resolver() comdat {
// CHECK-NEXT: resolver_entry:
// CHECK-NEXT: ret ptr @fmv_two._MsimdMfp16
//
//
-// CHECK: Function Attrs: noinline nounwind optnone
-// CHECK-LABEL: define {{[^@]+}}@foo
-// CHECK-SAME: () #[[ATTR2:[0-9]+]] {
-// CHECK-NEXT: entry:
-// CHECK-NEXT: [[CALL:%.*]] = call i32 @fmv.ifunc()
-// CHECK-NEXT: [[CALL1:%.*]] = call i32 @fmv_one.ifunc()
-// CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[CALL]], [[CALL1]]
-// CHECK-NEXT: [[CALL2:%.*]] = call i32 @fmv_two.ifunc()
-// CHECK-NEXT: [[ADD3:%.*]] = add nsw i32 [[ADD]], [[CALL2]]
-// CHECK-NEXT: ret i32 [[ADD3]]
-//
-//
// CHECK-LABEL: define {{[^@]+}}@fmv_e.resolver() comdat {
// CHECK-NEXT: resolver_entry:
// CHECK-NEXT: ret ptr @fmv_e._Mls64
@@ -238,28 +245,14 @@ int hoo(void) {
// CHECK-NEXT: ret void
//
//
-// CHECK-LABEL: define {{[^@]+}}@fmv_c.resolver() comdat {
-// CHECK-NEXT: resolver_entry:
-// CHECK-NEXT: call void @__init_cpu_features_resolver()
-// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr @__aarch64_cpu_features, align 8
-// CHECK-NEXT: [[TMP1:%.*]] = and i64 [[TMP0]], 281474976710656
-// CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[TMP1]], 281474976710656
-// CHECK-NEXT: [[TMP3:%.*]] = and i1 true, [[TMP2]]
-// CHECK-NEXT: br i1 [[TMP3]], label [[RESOLVER_RETURN:%.*]], label [[RESOLVER_ELSE:%.*]]
-// CHECK: resolver_return:
-// CHECK-NEXT: ret ptr @fmv_c._Mssbs
-// CHECK: resolver_else:
-// CHECK-NEXT: ret ptr @fmv_c.default
-//
-//
// CHECK: Function Attrs: noinline nounwind optnone
// CHECK-LABEL: define {{[^@]+}}@goo
// CHECK-SAME: () #[[ATTR2]] {
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[CALL:%.*]] = call i32 @fmv_inline.ifunc()
-// CHECK-NEXT: [[CALL1:%.*]] = call i32 @fmv_e.ifunc()
-// CHECK-NEXT: [[CALL2:%.*]] = call i32 @fmv_d.ifunc()
-// CHECK-NEXT: call void @fmv_c.ifunc()
+// CHECK-NEXT: [[CALL:%.*]] = call i32 @fmv_inline()
+// CHECK-NEXT: [[CALL1:%.*]] = call i32 @fmv_e()
+// CHECK-NEXT: [[CALL2:%.*]] = call i32 @fmv_d()
+// CHECK-NEXT: call void @fmv_c()
// CHECK-NEXT: [[CALL3:%.*]] = call i32 @fmv_default()
// CHECK-NEXT: ret i32 [[CALL3]]
//
@@ -412,6 +405,20 @@ int hoo(void) {
// CHECK-NEXT: ret ptr @fmv_d.default
//
//
+// CHECK-LABEL: define {{[^@]+}}@fmv_c.resolver() comdat {
+// CHECK-NEXT: resolver_entry:
+// CHECK-NEXT: call void @__init_cpu_features_resolver()
+// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr @__aarch64_cpu_features, align 8
+// CHECK-NEXT: [[TMP1:%.*]] = and i64 [[TMP0]], 281474976710656
+// CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[TMP1]], 281474976710656
+// CHECK-NEXT: [[TMP3:%.*]] = and i1 true, [[TMP2]]
+// CHECK-NEXT: br i1 [[TMP3]], label [[RESOLVER_RETURN:%.*]], label [[RESOLVER_ELSE:%.*]]
+// CHECK: resolver_return:
+// CHECK-NEXT: ret ptr @fmv_c._Mssbs
+// CHECK: resolver_else:
+// CHECK-NEXT: ret ptr @fmv_c.default
+//
+//
// CHECK: Function Attrs: noinline nounwind optnone
// CHECK-LABEL: define {{[^@]+}}@recur
// CHECK-SAME: () #[[ATTR2]] {
@@ -437,9 +444,9 @@ int hoo(void) {
// CHECK-NEXT: entry:
// CHECK-NEXT: [[FP1:%.*]] = alloca ptr, align 8
// CHECK-NEXT: [[FP2:%.*]] = alloca ptr, align 8
-// CHECK-NEXT: call void @f(ptr noundef @fmv.ifunc)
-// CHECK-NEXT: store ptr @fmv.ifunc, ptr [[FP1]], align 8
-// CHECK-NEXT: store ptr @fmv.ifunc, ptr [[FP2]], align 8
+// CHECK-NEXT: call void @f(ptr noundef @fmv)
+// CHECK-NEXT: store ptr @fmv, ptr [[FP1]], align 8
+// CHECK-NEXT: store ptr @fmv, ptr [[FP2]], align 8
// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[FP1]], align 8
// CHECK-NEXT: [[CALL:%.*]] = call i32 [[TMP0]]()
// CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[FP2]], align 8
@@ -561,13 +568,6 @@ int hoo(void) {
//
//
// CHECK: Function Attrs: noinline nounwind optnone
-// CHECK-LABEL: define {{[^@]+}}@fmv_c.default
-// CHECK-SAME: () #[[ATTR2]] {
-// CHECK-NEXT: entry:
-// CHECK-NEXT: ret void
-//
-//
-// CHECK: Function Attrs: noinline nounwind optnone
// CHECK-LABEL: define {{[^@]+}}@fmv_inline._Msha1MpmullMf64mm
// CHECK-SAME: () #[[ATTR12:[0-9]+]] {
// CHECK-NEXT: entry:
@@ -700,6 +700,13 @@ int hoo(void) {
// CHECK-NEXT: ret i32 1
//
//
+// CHECK: Function Attrs: noinline nounwind optnone
+// CHECK-LABEL: define {{[^@]+}}@fmv_c.default
+// CHECK-SAME: () #[[ATTR2]] {
+// CHECK-NEXT: entry:
+// CHECK-NEXT: ret void
+//
+//
// CHECK-NOFMV: Function Attrs: noinline nounwind optnone
// CHECK-NOFMV-LABEL: define {{[^@]+}}@fmv
// CHECK-NOFMV-SAME: () #[[ATTR0:[0-9]+]] {
diff --git a/clang/test/CodeGenCXX/attr-target-version.cpp b/clang/test/CodeGenCXX/attr-target-version.cpp
index 68dd7be1180b48..b63815db7e40fa 100644
--- a/clang/test/CodeGenCXX/attr-target-version.cpp
+++ b/clang/test/CodeGenCXX/attr-target-version.cpp
@@ -26,9 +26,12 @@ int bar() {
//.
// CHECK: @__aarch64_cpu_features = external dso_local global { i64 }
-// CHECK: @_Z3fooi.ifunc = weak_odr ifunc i32 (i32), ptr @_Z3fooi.resolver
-// CHECK: @_Z3foov.ifunc = weak_odr ifunc i32 (), ptr @_Z3foov.resolver
-// CHECK: @_ZN7MyClass3gooEi.ifunc = weak_odr ifunc i32 (ptr, i32), ptr @_ZN7MyClass3gooEi.resolver
+// CHECK: @_ZN7MyClass3gooEi.ifunc = weak_odr alias i32 (ptr, i32), ptr @_ZN7MyClass3gooEi
+// CHECK: @_Z3fooi.ifunc = weak_odr alias i32 (i32), ptr @_Z3fooi
+// CHECK: @_Z3foov.ifunc = weak_odr alias i32 (), ptr @_Z3foov
+// CHECK: @_ZN7MyClass3gooEi = weak_odr ifunc i32 (ptr, i32), ptr @_ZN7MyClass3gooEi.resolver
+// CHECK: @_Z3fooi = weak_odr ifunc i32 (i32), ptr @_Z3fooi.resolver
+// CHECK: @_Z3foov = weak_odr ifunc i32 (), ptr @_Z3foov.resolver
//.
// CHECK-LABEL: @_Z3fooi._Mbf16Msme-f64f64(
// CHECK-NEXT: entry:
@@ -37,39 +40,11 @@ int bar() {
// CHECK-NEXT: ret i32 1
//
//
-// CHECK-LABEL: @_Z3fooi.resolver(
-// CHECK-NEXT: resolver_entry:
-// CHECK-NEXT: call void @__init_cpu_features_resolver()
-// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr @__aarch64_cpu_features, align 8
-// CHECK-NEXT: [[TMP1:%.*]] = and i64 [[TMP0]], 36028797153181696
-// CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[TMP1]], 36028797153181696
-// CHECK-NEXT: [[TMP3:%.*]] = and i1 true, [[TMP2]]
-// CHECK-NEXT: br i1 [[TMP3]], label [[RESOLVER_RETURN:%.*]], label [[RESOLVER_ELSE:%.*]]
-// CHECK: resolver_return:
-// CHECK-NEXT: ret ptr @_Z3fooi._Mbf16Msme-f64f64
-// CHECK: resolver_else:
-// CHECK-NEXT: ret ptr @_Z3fooi.default
-//
-//
// CHECK-LABEL: @_Z3foov._Msm4Mebf16(
// CHECK-NEXT: entry:
// CHECK-NEXT: ret i32 3
//
//
-// CHECK-LABEL: @_Z3foov.resolver(
-// CHECK-NEXT: resolver_entry:
-// CHECK-NEXT: call void @__init_cpu_features_resolver()
-// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr @__aarch64_cpu_features, align 8
-// CHECK-NEXT: [[TMP1:%.*]] = and i64 [[TMP0]], 268435488
-// CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[TMP1]], 268435488
-// CHECK-NEXT: [[TMP3:%.*]] = and i1 true, [[TMP2]]
-// CHECK-NEXT: br i1 [[TMP3]], label [[RESOLVER_RETURN:%.*]], label [[RESOLVER_ELSE:%.*]]
-// CHECK: resolver_return:
-// CHECK-NEXT: ret ptr @_Z3foov._Msm4Mebf16
-// CHECK: resolver_else:
-// CHECK-NEXT: ret ptr @_Z3foov.default
-//
-//
// CHECK-LABEL: @_ZN7MyClass3gooEi.resolver(
// CHECK-NEXT: resolver_entry:
// CHECK-NEXT: call void @__init_cpu_features_resolver()
@@ -95,24 +70,40 @@ int bar() {
// CHECK-LABEL: @_Z3barv(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[M:%.*]] = alloca [[STRUCT_MYCLASS:%.*]], align 1
-// CHECK-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN7MyClass3gooEi.ifunc(ptr noundef nonnull align 1 dereferenceable(1) [[M]], i32 noundef 1)
-// CHECK-NEXT: [[CALL1:%.*]] = call noundef i32 @_Z3fooi.ifunc(i32 noundef 1)
+// CHECK-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN7MyClass3gooEi(ptr noundef nonnull align 1 dereferenceable(1) [[M]], i32 noundef 1)
+// CHECK-NEXT: [[CALL1:%.*]] = call noundef i32 @_Z3fooi(i32 noundef 1)
// CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[CALL]], [[CALL1]]
-// CHECK-NEXT: [[CALL2:%.*]] = call noundef i32 @_Z3foov.ifunc()
+// CHECK-NEXT: [[CALL2:%.*]] = call noundef i32 @_Z3foov()
// CHECK-NEXT: [[ADD3:%.*]] = add nsw i32 [[ADD]], [[CALL2]]
// CHECK-NEXT: ret i32 [[ADD3]]
//
//
-// CHECK-LABEL: @_Z3fooi.default(
-// CHECK-NEXT: entry:
-// CHECK-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
-// CHECK-NEXT: store i32 [[TMP0:%.*]], ptr [[DOTADDR]], align 4
-// CHECK-NEXT: ret i32 2
+// CHECK-LABEL: @_Z3fooi.resolver(
+// CHECK-NEXT: resolver_entry:
+// CHECK-NEXT: call void @__init_cpu_features_resolver()
+// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr @__aarch64_cpu_features, align 8
+// CHECK-NEXT: [[TMP1:%.*]] = and i64 [[TMP0]], 36028797153181696
+// CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[TMP1]], 36028797153181696
+// CHECK-NEXT: [[TMP3:%.*]] = and i1 true, [[TMP2]]
+// CHECK-NEXT: br i1 [[TMP3]], label [[RESOLVER_RETURN:%.*]], label [[RESOLVER_ELSE:%.*]]
+// CHECK: resolver_return:
+// CHECK-NEXT: ret ptr @_Z3fooi._Mbf16Msme-f64f64
+// CHECK: resolver_else:
+// CHECK-NEXT: ret ptr @_Z3fooi.default
//
//
-// CHECK-LABEL: @_Z3foov.default(
-// CHECK-NEXT: entry:
-// CHECK-NEXT: ret i32 4
+// CHECK-LABEL: @_Z3foov.resolver(
+// CHECK-NEXT: resolver_entry:
+// CHECK-NEXT: call void @__init_cpu_features_resolver()
+// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr @__aarch64_cpu_features, align 8
+// CHECK-NEXT: [[TMP1:%.*]] = and i64 [[TMP0]], 268435488
+// CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[TMP1]], 268435488
+// CHECK-NEXT: [[TMP3:%.*]] = and i1 true, [[TMP2]]
+// CHECK-NEXT: br i1 [[TMP3]], label [[RESOLVER_RETURN:%.*]], label [[RESOLVER_ELSE:%.*]]
+// CHECK: resolver_return:
+// CHECK-NEXT: ret ptr @_Z3foov._Msm4Mebf16
+// CHECK: resolver_else:
+// CHECK-NEXT: ret ptr @_Z3foov.default
//
//
// CHECK-LABEL: @_ZN7MyClass3gooEi._Mdotprod(
@@ -144,6 +135,18 @@ int bar() {
// CHECK-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK-NEXT: ret i32 1
//
+//
+// CHECK-LABEL: @_Z3fooi.default(
+// CHECK-NEXT: entry:
+// CHECK-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK-NEXT: store i32 [[TMP0:%.*]], ptr [[DOTADDR]], align 4
+// CHECK-NEXT: ret i32 2
+//
+//
+// CHECK-LABEL: @_Z3foov.default(
+// CHECK-NEXT: entry:
+// CHECK-NEXT: ret i32 4
+//
//.
// CHECK: attributes #[[ATTR0:[0-9]+]] = { mustprogress noinline nounwind optnone "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-features"="+bf16,+sme,+sme-f64f64" }
// CHECK: attributes #[[ATTR1:[0-9]+]] = { mustprogress noinline nounwind optnone "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-features"="+bf16,+fp-armv8,+neon,+sm4" }
>From e59dfe3ffaced168bcd491b5ecc0f4cdeb5319b5 Mon Sep 17 00:00:00 2001
From: Jon Roelofs <jonathan_roelofs at apple.com>
Date: Thu, 8 Feb 2024 14:21:08 -0800
Subject: [PATCH 2/4] clang-format
---
clang/lib/CodeGen/CodeGenModule.cpp | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp b/clang/lib/CodeGen/CodeGenModule.cpp
index 364553e2a6f0f7..872bd630caeeb5 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -4212,7 +4212,8 @@ void CodeGenModule::emitMultiVersionFunctions() {
llvm::Constant *ResolverConstant = GetOrCreateMultiVersionResolver(GD);
if (auto *IFunc = dyn_cast<llvm::GlobalIFunc>(ResolverConstant)) {
ResolverConstant = IFunc->getResolver();
- if (FD->isTargetClonesMultiVersion() || FD->isTargetVersionMultiVersion()) {
+ if (FD->isTargetClonesMultiVersion() ||
+ FD->isTargetVersionMultiVersion()) {
const CGFunctionInfo &FI = getTypes().arrangeGlobalDeclaration(GD);
llvm::FunctionType *DeclTy = getTypes().GetFunctionType(FI);
std::string MangledName = getMangledNameImpl(
>From 24692b074e6ac24ae42027f365fe3541542f7581 Mon Sep 17 00:00:00 2001
From: Jon Roelofs <jonathan_roelofs at apple.com>
Date: Thu, 8 Feb 2024 14:40:22 -0800
Subject: [PATCH 3/4] copy the note about .ifunc aliases from target_clones's
docs
---
clang/include/clang/Basic/AttrDocs.td | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/clang/include/clang/Basic/AttrDocs.td b/clang/include/clang/Basic/AttrDocs.td
index 041786f37fb8a7..19a98a0ccb9262 100644
--- a/clang/include/clang/Basic/AttrDocs.td
+++ b/clang/include/clang/Basic/AttrDocs.td
@@ -2517,6 +2517,14 @@ function it instructs compiler to emit multiple function versions based on
priority and target features availability. One of the versions is always
( implicitly or explicitly ) the ``default`` (fallback). Attribute strings can
contain dependent features names joined by the "+" sign.
+
+For targets that support the GNU indirect function (IFUNC) feature, dispatch
+is performed by emitting an indirect function that is resolved to the appropriate
+target clone at load time. The indirect function is given the name the
+multiversioned function would have if it had been declared without the attribute.
+For backward compatibility with earlier Clang releases, a function alias with an
+``.ifunc`` suffix is also emitted. The ``.ifunc`` suffixed symbol is a deprecated
+feature and support for it may be removed in the future.
}];
}
>From 3115821af5aca9cc586d1d2482ee31cf3369cc87 Mon Sep 17 00:00:00 2001
From: Jon Roelofs <jonathan_roelofs at apple.com>
Date: Thu, 8 Feb 2024 16:28:50 -0800
Subject: [PATCH 4/4] if -> switch
---
clang/lib/CodeGen/CodeGenModule.cpp | 13 ++++++++++++-
1 file changed, 12 insertions(+), 1 deletion(-)
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp b/clang/lib/CodeGen/CodeGenModule.cpp
index 872bd630caeeb5..2f923d5457f9cf 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -30,6 +30,7 @@
#include "clang/AST/ASTContext.h"
#include "clang/AST/ASTLambda.h"
#include "clang/AST/CharUnits.h"
+#include "clang/AST/Decl.h"
#include "clang/AST/DeclCXX.h"
#include "clang/AST/DeclObjC.h"
#include "clang/AST/DeclTemplate.h"
@@ -4394,8 +4395,18 @@ llvm::Constant *CodeGenModule::GetOrCreateMultiVersionResolver(GlobalDecl GD) {
// a separate resolver).
std::string ResolverName = MangledName;
if (getTarget().supportsIFunc()) {
- if (!FD->isTargetClonesMultiVersion() && !FD->isTargetVersionMultiVersion())
+ switch (FD->getMultiVersionKind()) {
+ case MultiVersionKind::None:
+ llvm_unreachable("unexpected MultiVersionKind::None for resolver");
+ case MultiVersionKind::Target:
+ case MultiVersionKind::CPUSpecific:
+ case MultiVersionKind::CPUDispatch:
ResolverName += ".ifunc";
+ break;
+ case MultiVersionKind::TargetClones:
+ case MultiVersionKind::TargetVersion:
+ break;
+ }
} else if (FD->isTargetMultiVersion()) {
ResolverName += ".resolver";
}
More information about the cfe-commits
mailing list