tahonermann updated this revision to Diff 420587.
tahonermann added a comment.
Squashed the addition of tests originally made in D122954
<https://reviews.llvm.org/D122954> to this review.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D122958/new/
https://reviews.llvm.org/D122958
Files:
clang/lib/CodeGen/CodeGenModule.cpp
clang/lib/CodeGen/CodeGenModule.h
clang/test/CodeGen/attr-target-clones.c
clang/test/CodeGenCXX/attr-target-clones.cpp
Index: clang/test/CodeGenCXX/attr-target-clones.cpp
===================================================================
--- clang/test/CodeGenCXX/attr-target-clones.cpp
+++ clang/test/CodeGenCXX/attr-target-clones.cpp
@@ -12,29 +12,29 @@
int __attribute__((target_clones("sse4.2", "default"))) overloaded(int) { return 1; }
// LINUX: define {{.*}}i32 @_Z10overloadedi.sse4.2.0(i32{{.+}})
// LINUX: define {{.*}}i32 @_Z10overloadedi.default.1(i32{{.+}})
-// LINUX: define i32 (i32)* @_Z10overloadedi.resolver
+// LINUX: define weak_odr i32 (i32)* @_Z10overloadedi.resolver() comdat
// LINUX: ret i32 (i32)* @_Z10overloadedi.sse4.2.0
// LINUX: ret i32 (i32)* @_Z10overloadedi.default.1
// WINDOWS: define dso_local noundef i32 @"?overloaded@@YAHH@Z.sse4.2.0"(i32{{.+}})
// WINDOWS: define dso_local noundef i32 @"?overloaded@@YAHH@Z.default.1"(i32{{.+}})
-// WINDOWS: define dso_local i32 @"?overloaded@@YAHH@Z"(i32{{.+}})
+// WINDOWS: define weak_odr dso_local i32 @"?overloaded@@YAHH@Z"(i32{{.+}}) comdat
// WINDOWS: call i32 @"?overloaded@@YAHH@Z.sse4.2.0"
// WINDOWS: call i32 @"?overloaded@@YAHH@Z.default.1"
int __attribute__((target_clones("arch=ivybridge", "default"))) overloaded(const char *) { return 2; }
// LINUX: define {{.*}}i32 @_Z10overloadedPKc.arch_ivybridge.0(i8*{{.+}})
// LINUX: define {{.*}}i32 @_Z10overloadedPKc.default.1(i8*{{.+}})
-// LINUX: define i32 (i8*)* @_Z10overloadedPKc.resolver
+// LINUX: define weak_odr i32 (i8*)* @_Z10overloadedPKc.resolver() comdat
// LINUX: ret i32 (i8*)* @_Z10overloadedPKc.arch_ivybridge.0
// LINUX: ret i32 (i8*)* @_Z10overloadedPKc.default.1
// WINDOWS: define dso_local noundef i32 @"?overloaded@@YAHPEBD@Z.arch_ivybridge.0"(i8*{{.+}})
// WINDOWS: define dso_local noundef i32 @"?overloaded@@YAHPEBD@Z.default.1"(i8*{{.+}})
-// WINDOWS: define dso_local i32 @"?overloaded@@YAHPEBD@Z"(i8*{{.+}})
+// WINDOWS: define weak_odr dso_local i32 @"?overloaded@@YAHPEBD@Z"(i8*{{.+}}) comdat
// WINDOWS: call i32 @"?overloaded@@YAHPEBD@Z.arch_ivybridge.0"
// WINDOWS: call i32 @"?overloaded@@YAHPEBD@Z.default.1"
-//
+
void use_overloaded() {
overloaded(1);
// LINUX: call noundef i32 @_Z10overloadedi.ifunc
@@ -81,36 +81,40 @@
// WINDOWS: call noundef i32 @"?foo@?$C@NM@@QEAAHXZ"(%struct.C
}
-// LINUX: define {{.*}}i32 @_ZN1CIssE3fooEv.sse4.2.0(%struct.C{{.+}})
-// WINDOWS: define {{.*}}i32 @"?foo@?$C@FF@@QEAAHXZ.sse4.2.0"(%struct.C{{.+}})
-// LINUX: define i32 (%struct.C*)* @_ZN1CIssE3fooEv.resolver
+// LINUX: define weak_odr i32 (%struct.C*)* @_ZN1CIssE3fooEv.resolver() comdat
// LINUX: ret i32 (%struct.C*)* @_ZN1CIssE3fooEv.sse4.2.0
// LINUX: ret i32 (%struct.C*)* @_ZN1CIssE3fooEv.default.1
+
// WINDOWS: define {{.*}}i32 @"?foo@?$C@FF@@QEAAHXZ"(%struct.C{{.+}})
// WINDOWS: call i32 @"?foo@?$C@FF@@QEAAHXZ.sse4.2.0"
// WINDOWS: call i32 @"?foo@?$C@FF@@QEAAHXZ.default.1"
-// LINUX: define {{.*}}i32 @_ZN1CIisE3fooEv.sse4.2.0(%struct.C{{.+}})
-// WINDOWS: define {{.*}}i32 @"?foo@?$C@HF@@QEAAHXZ.sse4.2.0"(%struct.C{{.+}})
-// LINUX: define i32 (%struct.C{{.+}})* @_ZN1CIisE3fooEv.resolver
+// LINUX: define weak_odr i32 (%struct.C{{.+}})* @_ZN1CIisE3fooEv.resolver() comdat
// LINUX: ret i32 (%struct.C{{.+}})* @_ZN1CIisE3fooEv.sse4.2.0
// LINUX: ret i32 (%struct.C{{.+}})* @_ZN1CIisE3fooEv.default.1
+
// WINDOWS: define {{.*}}i32 @"?foo@?$C@HF@@QEAAHXZ"(%struct.C{{.+}})
// WINDOWS: call i32 @"?foo@?$C@HF@@QEAAHXZ.sse4.2.0"
// WINDOWS: call i32 @"?foo@?$C@HF@@QEAAHXZ.default.1"
-// LINUX: define i32 (%struct.C{{.+}})* @_ZN1CIdfE3fooEv.resolver
+// LINUX: define weak_odr i32 (%struct.C{{.+}})* @_ZN1CIdfE3fooEv.resolver() comdat
// LINUX: ret i32 (%struct.C{{.+}})* @_ZN1CIdfE3fooEv.sse4.2.0
// LINUX: ret i32 (%struct.C{{.+}})* @_ZN1CIdfE3fooEv.default.1
+
// WINDOWS: define {{.*}}i32 @"?foo@?$C@NM@@QEAAHXZ"(%struct.C{{.+}})
// WINDOWS: call i32 @"?foo@?$C@NM@@QEAAHXZ.sse4.2.0"
// WINDOWS: call i32 @"?foo@?$C@NM@@QEAAHXZ.default.1"
+// LINUX: define {{.*}}i32 @_ZN1CIssE3fooEv.sse4.2.0(%struct.C{{.+}})
+// LINUX: define {{.*}}i32 @_ZN1CIssE3fooEv.default.1(%struct.C{{.+}})
+// LINUX: define {{.*}}i32 @_ZN1CIisE3fooEv.sse4.2.0(%struct.C{{.+}})
+// LINUX: define {{.*}}i32 @_ZN1CIisE3fooEv.default.1(%struct.C{{.+}})
// LINUX: define {{.*}}i32 @_ZN1CIdfE3fooEv.sse4.2.0(%struct.C{{.+}})
-// WINDOWS: define {{.*}}i32 @"?foo@?$C@NM@@QEAAHXZ.sse4.2.0"(%struct.C{{.+}})
// LINUX: define {{.*}}i32 @_ZN1CIdfE3fooEv.default.1(%struct.C{{.+}})
-// WINDOWS: define {{.*}}i32 @"?foo@?$C@NM@@QEAAHXZ.default.1"(%struct.C{{.+}})
-// LINUX: define {{.*}}i32 @_ZN1CIssE3fooEv.default.1(%struct.C{{.+}})
+
+// WINDOWS: define {{.*}}i32 @"?foo@?$C@FF@@QEAAHXZ.sse4.2.0"(%struct.C{{.+}})
// WINDOWS: define {{.*}}i32 @"?foo@?$C@FF@@QEAAHXZ.default.1"(%struct.C{{.+}})
-// LINUX: define {{.*}}i32 @_ZN1CIisE3fooEv.default.1(%struct.C{{.+}})
+// WINDOWS: define {{.*}}i32 @"?foo@?$C@HF@@QEAAHXZ.sse4.2.0"(%struct.C{{.+}})
// WINDOWS: define {{.*}}i32 @"?foo@?$C@HF@@QEAAHXZ.default.1"(%struct.C{{.+}})
+// WINDOWS: define {{.*}}i32 @"?foo@?$C@NM@@QEAAHXZ.sse4.2.0"(%struct.C{{.+}})
+// WINDOWS: define {{.*}}i32 @"?foo@?$C@NM@@QEAAHXZ.default.1"(%struct.C{{.+}})
Index: clang/test/CodeGen/attr-target-clones.c
===================================================================
--- clang/test/CodeGen/attr-target-clones.c
+++ clang/test/CodeGen/attr-target-clones.c
@@ -18,30 +18,31 @@
// LINUX: @unused.ifunc = weak_odr ifunc void (), void ()* ()* @unused.resolver
// LINUX: @foo_inline.ifunc = weak_odr ifunc i32 (), i32 ()* ()* @foo_inline.resolver
// LINUX: @foo_inline2.ifunc = weak_odr ifunc i32 (), i32 ()* ()* @foo_inline2.resolver
+// LINUX: @foo_used_no_defn.ifunc = weak_odr ifunc i32 (), i32 ()* ()* @foo_used_no_defn.resolver
int __attribute__((target_clones("sse4.2, default"))) foo(void) { return 0; }
// LINUX: define {{.*}}i32 @foo.sse4.2.0()
// LINUX: define {{.*}}i32 @foo.default.1()
-// LINUX: define i32 ()* @foo.resolver() comdat
+// LINUX: define weak_odr i32 ()* @foo.resolver() comdat
// LINUX: ret i32 ()* @foo.sse4.2.0
// LINUX: ret i32 ()* @foo.default.1
// WINDOWS: define dso_local i32 @foo.sse4.2.0()
// WINDOWS: define dso_local i32 @foo.default.1()
-// WINDOWS: define dso_local i32 @foo() comdat
+// WINDOWS: define weak_odr dso_local i32 @foo() comdat
// WINDOWS: musttail call i32 @foo.sse4.2.0
// WINDOWS: musttail call i32 @foo.default.1
__attribute__((target_clones("default,default ,sse4.2"))) void foo_dupes(void) {}
// LINUX: define {{.*}}void @foo_dupes.default.1()
// LINUX: define {{.*}}void @foo_dupes.sse4.2.0()
-// LINUX: define void ()* @foo_dupes.resolver() comdat
+// LINUX: define weak_odr void ()* @foo_dupes.resolver() comdat
// LINUX: ret void ()* @foo_dupes.sse4.2.0
// LINUX: ret void ()* @foo_dupes.default.1
// WINDOWS: define dso_local void @foo_dupes.default.1()
// WINDOWS: define dso_local void @foo_dupes.sse4.2.0()
-// WINDOWS: define dso_local void @foo_dupes() comdat
+// WINDOWS: define weak_odr dso_local void @foo_dupes() comdat
// WINDOWS: musttail call void @foo_dupes.sse4.2.0
// WINDOWS: musttail call void @foo_dupes.default.1
@@ -64,13 +65,13 @@
void __attribute__((target_clones("default, arch=ivybridge"))) unused(void) {}
// LINUX: define {{.*}}void @unused.default.1()
// LINUX: define {{.*}}void @unused.arch_ivybridge.0()
-// LINUX: define void ()* @unused.resolver() comdat
+// LINUX: define weak_odr void ()* @unused.resolver() comdat
// LINUX: ret void ()* @unused.arch_ivybridge.0
// LINUX: ret void ()* @unused.default.1
// WINDOWS: define dso_local void @unused.default.1()
// WINDOWS: define dso_local void @unused.arch_ivybridge.0()
-// WINDOWS: define dso_local void @unused() comdat
+// WINDOWS: define weak_odr dso_local void @unused() comdat
// WINDOWS: musttail call void @unused.arch_ivybridge.0
// WINDOWS: musttail call void @unused.default.1
@@ -90,46 +91,80 @@
// WINDOWS: call i32 @foo_inline2()
}
-// Deferred emission of foo_inline, which got delayed because it is inline.
-// LINUX: define i32 ()* @foo_inline.resolver() comdat
+// LINUX: define weak_odr i32 ()* @foo_inline.resolver() comdat
// LINUX: ret i32 ()* @foo_inline.arch_sandybridge.0
// LINUX: ret i32 ()* @foo_inline.sse4.2.1
// LINUX: ret i32 ()* @foo_inline.default.2
-// WINDOWS: define dso_local i32 @foo_inline() comdat
+// WINDOWS: define weak_odr dso_local i32 @foo_inline() comdat
// WINDOWS: musttail call i32 @foo_inline.arch_sandybridge.0
// WINDOWS: musttail call i32 @foo_inline.sse4.2.1
// WINDOWS: musttail call i32 @foo_inline.default.2
inline int __attribute__((target_clones("arch=sandybridge,default,sse4.2")))
foo_inline2(void){ return 0; }
-// LINUX: define linkonce i32 @foo_inline2.arch_sandybridge.0() #[[SB:[0-9]+]]
-// LINUX: define i32 ()* @foo_inline2.resolver() comdat
+// LINUX: define weak_odr i32 ()* @foo_inline2.resolver() comdat
// LINUX: ret i32 ()* @foo_inline2.arch_sandybridge.0
// LINUX: ret i32 ()* @foo_inline2.sse4.2.1
// LINUX: ret i32 ()* @foo_inline2.default.2
-// WINDOWS: define linkonce_odr dso_local i32 @foo_inline2.arch_sandybridge.0() #[[SB:[0-9]+]]
-// WINDOWS: define dso_local i32 @foo_inline2() comdat
+// WINDOWS: define weak_odr dso_local i32 @foo_inline2() comdat
// WINDOWS: musttail call i32 @foo_inline2.arch_sandybridge.0
// WINDOWS: musttail call i32 @foo_inline2.sse4.2.1
// WINDOWS: musttail call i32 @foo_inline2.default.2
-// LINUX: define linkonce i32 @foo_inline.arch_sandybridge.0() #[[SB]]
-// LINUX: define linkonce i32 @foo_inline.default.2() #[[DEF]]
+
+int __attribute__((target_clones("default", "sse4.2")))
+foo_unused_no_defn(void);
+
+int __attribute__((target_clones("default", "sse4.2")))
+foo_used_no_defn(void);
+
+int test_foo_used_no_defn(void) {
+ // LINUX: define {{.*}}i32 @test_foo_used_no_defn()
+ // WINDOWS: define dso_local i32 @test_foo_used_no_defn()
+ return foo_used_no_defn();
+ // LINUX: call i32 @foo_used_no_defn.ifunc()
+ // WINDOWS: call i32 @foo_used_no_defn()
+}
+
+
+// LINUX: define weak_odr i32 ()* @foo_used_no_defn.resolver() comdat
+// LINUX: ret i32 ()* @foo_used_no_defn.sse4.2.0
+// LINUX: ret i32 ()* @foo_used_no_defn.default.1
+
+// WINDOWS: define weak_odr dso_local i32 @foo_used_no_defn() comdat
+// WINDOWS: musttail call i32 @foo_used_no_defn.sse4.2.0
+// WINDOWS: musttail call i32 @foo_used_no_defn.default.1
+
+
+// Deferred emission of inline definitions.
+
+// LINUX: define linkonce i32 @foo_inline.arch_sandybridge.0() #[[SB:[0-9]+]]
+// LINUX: define linkonce i32 @foo_inline.default.2() #[[DEF:[0-9]+]]
// LINUX: define linkonce i32 @foo_inline.sse4.2.1() #[[SSE42:[0-9]+]]
-// WINDOWS: define linkonce_odr dso_local i32 @foo_inline.arch_sandybridge.0() #[[SB]]
+// WINDOWS: define linkonce_odr dso_local i32 @foo_inline.arch_sandybridge.0() #[[SB:[0-9]+]]
// WINDOWS: define linkonce_odr dso_local i32 @foo_inline.default.2() #[[DEF]]
// WINDOWS: define linkonce_odr dso_local i32 @foo_inline.sse4.2.1() #[[SSE42:[0-9]+]]
+// LINUX: define linkonce i32 @foo_inline2.arch_sandybridge.0() #[[SB]]
// LINUX: define linkonce i32 @foo_inline2.default.2() #[[DEF]]
// LINUX: define linkonce i32 @foo_inline2.sse4.2.1() #[[SSE42]]
+// WINDOWS: define linkonce_odr dso_local i32 @foo_inline2.arch_sandybridge.0() #[[SB]]
// WINDOWS: define linkonce_odr dso_local i32 @foo_inline2.default.2() #[[DEF]]
// WINDOWS: define linkonce_odr dso_local i32 @foo_inline2.sse4.2.1() #[[SSE42]]
+
+// LINUX: declare i32 @foo_used_no_defn.default.1()
+// LINUX: declare i32 @foo_used_no_defn.sse4.2.0()
+
+// WINDOWS: declare dso_local i32 @foo_used_no_defn.default.1()
+// WINDOWS: declare dso_local i32 @foo_used_no_defn.sse4.2.0()
+
+
// CHECK: attributes #[[SSE42]] =
// CHECK-SAME: "target-features"="+crc32,+cx8,+mmx,+popcnt,+sse,+sse2,+sse3,+sse4.1,+sse4.2,+ssse3,+x87"
// CHECK: attributes #[[DEF]] =
Index: clang/lib/CodeGen/CodeGenModule.h
===================================================================
--- clang/lib/CodeGen/CodeGenModule.h
+++ clang/lib/CodeGen/CodeGenModule.h
@@ -1505,7 +1505,6 @@
void EmitAliasDefinition(GlobalDecl GD);
void emitIFuncDefinition(GlobalDecl GD);
void emitCPUDispatchDefinition(GlobalDecl GD);
- void EmitTargetClonesResolver(GlobalDecl GD);
void EmitObjCPropertyImplementations(const ObjCImplementationDecl *D);
void EmitObjCIvarInitializations(ObjCImplementationDecl *D);
Index: clang/lib/CodeGen/CodeGenModule.cpp
===================================================================
--- clang/lib/CodeGen/CodeGenModule.cpp
+++ clang/lib/CodeGen/CodeGenModule.cpp
@@ -3320,13 +3320,13 @@
auto *Spec = FD->getAttr<CPUSpecificAttr>();
for (unsigned I = 0; I < Spec->cpus_size(); ++I)
EmitGlobalFunctionDefinition(GD.getWithMultiVersionIndex(I), nullptr);
- // Requires multiple emits.
} else if (FD->isTargetClonesMultiVersion()) {
auto *Clone = FD->getAttr<TargetClonesAttr>();
for (unsigned I = 0; I < Clone->featuresStrs_size(); ++I)
if (Clone->isFirstOfVersion(I))
EmitGlobalFunctionDefinition(GD.getWithMultiVersionIndex(I), nullptr);
- EmitTargetClonesResolver(GD);
+ // Ensure that the resolver function is also emitted.
+ GetOrCreateMultiVersionResolver(GD);
} else
EmitGlobalFunctionDefinition(GD, GV);
}
@@ -3408,57 +3408,6 @@
return llvm::GlobalValue::WeakODRLinkage;
}
-void CodeGenModule::EmitTargetClonesResolver(GlobalDecl GD) {
- const auto *FD = cast<FunctionDecl>(GD.getDecl());
- assert(FD && "Not a FunctionDecl?");
- const auto *TC = FD->getAttr<TargetClonesAttr>();
- assert(TC && "Not a target_clones Function?");
-
- llvm::Function *ResolverFunc;
- if (getTarget().supportsIFunc()) {
- auto *IFunc = cast<llvm::GlobalIFunc>(GetOrCreateMultiVersionResolver(GD));
- ResolverFunc = cast<llvm::Function>(IFunc->getResolver());
- } else
- ResolverFunc = cast<llvm::Function>(GetOrCreateMultiVersionResolver(GD));
-
- SmallVector<CodeGenFunction::MultiVersionResolverOption, 10> Options;
- for (unsigned VersionIndex = 0; VersionIndex < TC->featuresStrs_size();
- ++VersionIndex) {
- if (!TC->isFirstOfVersion(VersionIndex))
- continue;
- StringRef Version = TC->getFeatureStr(VersionIndex);
- StringRef MangledName =
- getMangledName(GD.getWithMultiVersionIndex(VersionIndex));
- llvm::Constant *Func = GetGlobalValue(MangledName);
- assert(Func &&
- "Should have already been created before calling resolver emit");
-
- StringRef Architecture;
- llvm::SmallVector<StringRef, 1> Feature;
-
- if (Version.startswith("arch="))
- Architecture = Version.drop_front(sizeof("arch=") - 1);
- else if (Version != "default")
- Feature.push_back(Version);
-
- Options.emplace_back(cast<llvm::Function>(Func), Architecture, Feature);
- }
-
- if (supportsCOMDAT())
- ResolverFunc->setComdat(
- getModule().getOrInsertComdat(ResolverFunc->getName()));
-
- const TargetInfo &TI = getTarget();
- std::stable_sort(
- Options.begin(), Options.end(),
- [&TI](const CodeGenFunction::MultiVersionResolverOption &LHS,
- const CodeGenFunction::MultiVersionResolverOption &RHS) {
- return TargetMVPriority(TI, LHS) > TargetMVPriority(TI, RHS);
- });
- CodeGenFunction CGF(*this);
- CGF.EmitMultiVersionResolver(ResolverFunc, Options);
-}
-
void CodeGenModule::emitMultiVersionFunctions() {
std::vector<GlobalDecl> MVFuncsToEmit;
MultiVersionFuncs.swap(MVFuncsToEmit);
@@ -3495,26 +3444,58 @@
Options.emplace_back(cast<llvm::Function>(Func),
TA->getArchitecture(), Feats);
});
+ } else if (FD->isTargetClonesMultiVersion()) {
+ const auto *TC = FD->getAttr<TargetClonesAttr>();
+ for (unsigned VersionIndex = 0; VersionIndex < TC->featuresStrs_size();
+ ++VersionIndex) {
+ if (!TC->isFirstOfVersion(VersionIndex))
+ continue;
+ GlobalDecl CurGD{(FD->isDefined() ? FD->getDefinition() : FD),
+ VersionIndex};
+ StringRef Version = TC->getFeatureStr(VersionIndex);
+ StringRef MangledName = getMangledName(CurGD);
+ llvm::Constant *Func = GetGlobalValue(MangledName);
+ if (!Func) {
+ if (FD->isDefined()) {
+ EmitGlobalFunctionDefinition(CurGD, nullptr);
+ Func = GetGlobalValue(MangledName);
+ } else {
+ const CGFunctionInfo &FI =
+ getTypes().arrangeGlobalDeclaration(CurGD);
+ llvm::FunctionType *Ty = getTypes().GetFunctionType(FI);
+ Func = GetAddrOfFunction(CurGD, Ty, /*ForVTable=*/false,
+ /*DontDefer=*/false, ForDefinition);
+ }
+ assert(Func && "This should have just been created");
+ }
+
+ StringRef Architecture;
+ llvm::SmallVector<StringRef, 1> Feature;
+
+ if (Version.startswith("arch="))
+ Architecture = Version.drop_front(sizeof("arch=") - 1);
+ else if (Version != "default")
+ Feature.push_back(Version);
+
+ Options.emplace_back(cast<llvm::Function>(Func), Architecture, Feature);
+ }
} else {
- assert(0 && "Expected a target multiversion function");
+ assert(0 && "Expected a target or target_clones multiversion function");
continue;
}
- llvm::Function *ResolverFunc;
- const TargetInfo &TI = getTarget();
+ llvm::Constant *ResolverConstant = GetOrCreateMultiVersionResolver(GD);
+ if (auto *IFunc = dyn_cast<llvm::GlobalIFunc>(ResolverConstant))
+ ResolverConstant = IFunc->getResolver();
+ llvm::Function *ResolverFunc = cast<llvm::Function>(ResolverConstant);
- if (TI.supportsIFunc() || FD->isTargetMultiVersion()) {
- ResolverFunc = cast<llvm::Function>(
- GetGlobalValue((getMangledName(GD) + ".resolver").str()));
- ResolverFunc->setLinkage(getMultiversionLinkage(*this, GD));
- } else {
- ResolverFunc = cast<llvm::Function>(GetGlobalValue(getMangledName(GD)));
- }
+ ResolverFunc->setLinkage(getMultiversionLinkage(*this, GD));
if (supportsCOMDAT())
ResolverFunc->setComdat(
getModule().getOrInsertComdat(ResolverFunc->getName()));
+ const TargetInfo &TI = getTarget();
llvm::stable_sort(
Options, [&TI](const CodeGenFunction::MultiVersionResolverOption &LHS,
const CodeGenFunction::MultiVersionResolverOption &RHS) {
@@ -3676,35 +3657,17 @@
else if (FD->isTargetMultiVersion())
ResolverName += ".resolver";
- // If this already exists, just return that one.
+ // If the resolver has already been created, just return it.
if (llvm::GlobalValue *ResolverGV = GetGlobalValue(ResolverName))
return ResolverGV;
const CGFunctionInfo &FI = getTypes().arrangeGlobalDeclaration(GD);
llvm::FunctionType *DeclTy = getTypes().GetFunctionType(FI);
- // Since this is the first time we've created this IFunc, make sure
- // that we put this multiversioned function into the list to be
- // replaced later if necessary (target multiversioning only).
- if (FD->isTargetMultiVersion())
+ // The resolver needs to be created. For target and target_clones, defer
+ // creation until the end of the TU.
+ if (FD->isTargetMultiVersion() || FD->isTargetClonesMultiVersion())
MultiVersionFuncs.push_back(GD);
- else if (FD->isTargetClonesMultiVersion()) {
- // In target_clones multiversioning, make sure we emit this if used.
- auto DDI =
- DeferredDecls.find(getMangledName(GD.getWithMultiVersionIndex(0)));
- if (DDI != DeferredDecls.end()) {
- addDeferredDeclToEmit(GD);
- DeferredDecls.erase(DDI);
- } else {
- // Emit the symbol of the 1st variant, so that the deferred decls know we
- // need it, otherwise the only global value will be the resolver/ifunc,
- // which end up getting broken if we search for them with GetGlobalValue'.
- GetOrCreateLLVMFunction(
- getMangledName(GD.getWithMultiVersionIndex(0)), DeclTy, FD,
- /*ForVTable=*/false, /*DontDefer=*/true,
- /*IsThunk=*/false, llvm::AttributeList(), ForDefinition);
- }
- }
// For cpu_specific, don't create an ifunc yet because we don't know if the
// cpu_dispatch will be emitted in this translation unit.
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
