/// a VarDecl, a FunctionDecl or a BlockDecl.
class GlobalDecl {
llvm::PointerIntPair<const Decl *, 2> Value;
+ unsigned MultiVersionIndex = 0;
void Init(const Decl *D) {
assert(!isa<CXXConstructorDecl>(D) && "Use other ctor with ctor decls!");
public:
GlobalDecl() = default;
GlobalDecl(const VarDecl *D) { Init(D);}
- GlobalDecl(const FunctionDecl *D) { Init(D); }
+ GlobalDecl(const FunctionDecl *D, unsigned MVIndex = 0)
+ : MultiVersionIndex(MVIndex) {
+ Init(D);
+ }
GlobalDecl(const BlockDecl *D) { Init(D); }
GlobalDecl(const CapturedDecl *D) { Init(D); }
GlobalDecl(const ObjCMethodDecl *D) { Init(D); }
GlobalDecl CanonGD;
CanonGD.Value.setPointer(Value.getPointer()->getCanonicalDecl());
CanonGD.Value.setInt(Value.getInt());
+ CanonGD.MultiVersionIndex = MultiVersionIndex;
return CanonGD;
}
return static_cast<CXXDtorType>(Value.getInt());
}
+ unsigned getMultiVersionIndex() const {
+ assert(isa<FunctionDecl>(getDecl()) &&
+ !isa<CXXConstructorDecl>(getDecl()) &&
+ !isa<CXXDestructorDecl>(getDecl()) &&
+ "Decl is not a plain FunctionDecl!");
+ return MultiVersionIndex;
+ }
+
friend bool operator==(const GlobalDecl &LHS, const GlobalDecl &RHS) {
- return LHS.Value == RHS.Value;
+ return LHS.Value == RHS.Value &&
+ LHS.MultiVersionIndex == RHS.MultiVersionIndex;
}
void *getAsOpaquePtr() const { return Value.getOpaqueValue(); }
Result.Value.setPointer(D);
return Result;
}
+
+ GlobalDecl getWithMultiVersionIndex(unsigned Index) {
+ assert(isa<FunctionDecl>(getDecl()) &&
+ !isa<CXXConstructorDecl>(getDecl()) &&
+ !isa<CXXDestructorDecl>(getDecl()) &&
+ "Decl is not a plain FunctionDecl!");
+ GlobalDecl Result(*this);
+ Result.MultiVersionIndex = Index;
+ return Result;
+ }
};
} // namespace clang
let Subjects = SubjectList<[Function]>;
let Documentation = [CPUSpecificCPUDispatchDocs];
let AdditionalMembers = [{
- unsigned ActiveArgIndex = 0;
-
- IdentifierInfo *getCurCPUName() const {
- return *(cpus_begin() + ActiveArgIndex);
+ IdentifierInfo *getCPUName(unsigned Index) const {
+ return *(cpus_begin() + Index);
}
}];
}
} else {
Fn->setVisibility(llvm::GlobalValue::HiddenVisibility);
Fn->setUnnamedAddr(llvm::GlobalValue::UnnamedAddr::Global);
- CGM.SetLLVMFunctionAttributes(nullptr, FI, Fn);
+ CGM.SetLLVMFunctionAttributes(GlobalDecl(), FI, Fn);
CGM.SetLLVMFunctionAttributesForDefinition(nullptr, Fn);
}
}
static RValue emitLibraryCall(CodeGenFunction &CGF, const FunctionDecl *FD,
const CallExpr *E, llvm::Constant *calleeValue) {
- CGCallee callee = CGCallee::forDirect(calleeValue, FD);
+ CGCallee callee = CGCallee::forDirect(calleeValue, GlobalDecl(FD));
return CGF.EmitCall(E->getCallee()->getType(), callee, E, ReturnValueSlot());
}
llvm::Function *Fn = llvm::Function::Create(
FuncTy, llvm::GlobalValue::LinkOnceODRLinkage, Name, &CGM.getModule());
Fn->setVisibility(llvm::GlobalValue::HiddenVisibility);
- CGM.SetLLVMFunctionAttributes(nullptr, FI, Fn);
+ CGM.SetLLVMFunctionAttributes(GlobalDecl(), FI, Fn);
CGM.SetLLVMFunctionAttributesForDefinition(nullptr, Fn);
// Attach 'noinline' at -Oz.
return RValue::get(Builder.CreateCall(F, { Src, Src, ShiftAmt }));
}
-RValue CodeGenFunction::EmitBuiltinExpr(const FunctionDecl *FD,
- unsigned BuiltinID, const CallExpr *E,
+RValue CodeGenFunction::EmitBuiltinExpr(const GlobalDecl GD, unsigned BuiltinID,
+ const CallExpr *E,
ReturnValueSlot ReturnValue) {
+ const FunctionDecl *FD = GD.getDecl()->getAsFunction();
// See if we can constant fold this builtin. If so, don't emit it at all.
Expr::EvalResult Result;
if (E->EvaluateAsRValue(Result, CGM.getContext()) &&
CGF.Builder.CreateConstInBoundsGEP1_64(VTable, VTableIndex, "vfnkxt");
llvm::Value *VFunc =
CGF.Builder.CreateAlignedLoad(VFuncPtr, CGF.PointerAlignInBytes);
- CGCallee Callee(GD.getDecl()->getCanonicalDecl(), VFunc);
+ CGCallee Callee(GD, VFunc);
return Callee;
}
AddAttributesFromFunctionProtoType(getContext(), FuncAttrs,
CalleeInfo.getCalleeFunctionProtoType());
- const Decl *TargetDecl = CalleeInfo.getCalleeDecl();
+ const Decl *TargetDecl = CalleeInfo.getCalleeDecl().getDecl();
bool HasOptnone = false;
// FIXME: handle sseregparm someday...
FuncAttrs.addAttribute("disable-tail-calls",
llvm::toStringRef(DisableTailCalls));
- GetCPUAndFeaturesAttributes(TargetDecl, FuncAttrs);
+ GetCPUAndFeaturesAttributes(CalleeInfo.getCalleeDecl(), FuncAttrs);
}
ClangToLLVMArgMapping IRFunctionArgs(getContext(), FI);
// Apply always_inline to all calls within flatten functions.
// FIXME: should this really take priority over __try, below?
if (CurCodeDecl && CurCodeDecl->hasAttr<FlattenAttr>() &&
- !(Callee.getAbstractInfo().getCalleeDecl() &&
- Callee.getAbstractInfo().getCalleeDecl()->hasAttr<NoInlineAttr>())) {
+ !(Callee.getAbstractInfo().getCalleeDecl().getDecl() &&
+ Callee.getAbstractInfo()
+ .getCalleeDecl()
+ .getDecl()
+ ->hasAttr<NoInlineAttr>())) {
Attrs =
Attrs.addAttribute(getLLVMContext(), llvm::AttributeList::FunctionIndex,
llvm::Attribute::AlwaysInline);
// Suppress tail calls if requested.
if (llvm::CallInst *Call = dyn_cast<llvm::CallInst>(CI)) {
- const Decl *TargetDecl = Callee.getAbstractInfo().getCalleeDecl();
+ const Decl *TargetDecl = Callee.getAbstractInfo().getCalleeDecl().getDecl();
if (TargetDecl && TargetDecl->hasAttr<NotTailCalledAttr>())
Call->setTailCallKind(llvm::CallInst::TCK_NoTail);
}
} ();
// Emit the assume_aligned check on the return value.
- const Decl *TargetDecl = Callee.getAbstractInfo().getCalleeDecl();
+ const Decl *TargetDecl = Callee.getAbstractInfo().getCalleeDecl().getDecl();
if (Ret.isScalar() && TargetDecl) {
if (const auto *AA = TargetDecl->getAttr<AssumeAlignedAttr>()) {
llvm::Value *OffsetValue = nullptr;
/// The function prototype of the callee.
const FunctionProtoType *CalleeProtoTy;
/// The function declaration of the callee.
- const Decl *CalleeDecl;
+ GlobalDecl CalleeDecl;
public:
- explicit CGCalleeInfo() : CalleeProtoTy(nullptr), CalleeDecl(nullptr) {}
- CGCalleeInfo(const FunctionProtoType *calleeProtoTy, const Decl *calleeDecl)
+ explicit CGCalleeInfo() : CalleeProtoTy(nullptr), CalleeDecl() {}
+ CGCalleeInfo(const FunctionProtoType *calleeProtoTy, GlobalDecl calleeDecl)
: CalleeProtoTy(calleeProtoTy), CalleeDecl(calleeDecl) {}
CGCalleeInfo(const FunctionProtoType *calleeProtoTy)
- : CalleeProtoTy(calleeProtoTy), CalleeDecl(nullptr) {}
- CGCalleeInfo(const Decl *calleeDecl)
+ : CalleeProtoTy(calleeProtoTy), CalleeDecl() {}
+ CGCalleeInfo(GlobalDecl calleeDecl)
: CalleeProtoTy(nullptr), CalleeDecl(calleeDecl) {}
const FunctionProtoType *getCalleeFunctionProtoType() const {
return CalleeProtoTy;
}
- const Decl *getCalleeDecl() const { return CalleeDecl; }
+ const GlobalDecl getCalleeDecl() const { return CalleeDecl; }
};
/// All available information about a concrete callee.
}
CGCalleeInfo getAbstractInfo() const {
if (isVirtual())
- return VirtualInfo.MD.getDecl();
+ return VirtualInfo.MD;
assert(isOrdinary());
return AbstractInfo;
}
CGM.getAddrOfCXXStructor(D, getFromCtorType(Type));
const CGFunctionInfo &Info = CGM.getTypes().arrangeCXXConstructorCall(
Args, D, Type, ExtraArgs.Prefix, ExtraArgs.Suffix, PassPrototypeArgs);
- CGCallee Callee = CGCallee::forDirect(CalleePtr, D);
+ CGCallee Callee = CGCallee::forDirect(CalleePtr, GlobalDecl(D, Type));
EmitCall(Info, Callee, ReturnValueSlot(), Args);
// Generate vtable assumptions if we're constructing a complete object
// variadic arguments.
// Now emit our call.
- auto callee = CGCallee::forDirect(calleePtr, callOperator);
+ auto callee = CGCallee::forDirect(calleePtr, GlobalDecl(callOperator));
RValue RV = EmitCall(calleeFnInfo, callee, returnSlot, callArgs);
// If necessary, copy the returned value into the slot.
OutlinedStmt->getBeginLoc(), OutlinedStmt->getBeginLoc());
CurSEHParent = ParentCGF.CurSEHParent;
- CGM.SetLLVMFunctionAttributes(nullptr, FnInfo, CurFn);
+ CGM.SetLLVMFunctionAttributes(GlobalDecl(), FnInfo, CurFn);
EmitCapturedLocals(ParentCGF, OutlinedStmt, IsFilter);
}
}
llvm::Constant *calleePtr = EmitFunctionDeclPointer(CGF.CGM, FD);
- return CGCallee::forDirect(calleePtr, FD);
+ return CGCallee::forDirect(calleePtr, GlobalDecl(FD));
}
CGCallee CodeGenFunction::EmitCallee(const Expr *E) {
calleePtr = EmitLValue(E).getPointer();
}
assert(functionType->isFunctionType());
- CGCalleeInfo calleeInfo(functionType->getAs<FunctionProtoType>(),
- E->getReferencedDeclOfCallee());
+
+ GlobalDecl GD;
+ if (const auto *VD =
+ dyn_cast_or_null<VarDecl>(E->getReferencedDeclOfCallee()))
+ GD = GlobalDecl(VD);
+
+ CGCalleeInfo calleeInfo(functionType->getAs<FunctionProtoType>(), GD);
CGCallee callee(calleeInfo, calleePtr);
return callee;
}
assert(CalleeType->isFunctionPointerType() &&
"Call must have function pointer type!");
- const Decl *TargetDecl = OrigCallee.getAbstractInfo().getCalleeDecl();
+ const Decl *TargetDecl =
+ OrigCallee.getAbstractInfo().getCalleeDecl().getDecl();
if (const FunctionDecl *FD = dyn_cast_or_null<FunctionDecl>(TargetDecl))
// We can only guarantee that a function is called from the correct
if (MD->isStatic()) {
// The method is static, emit it as we would a regular call.
- CGCallee callee = CGCallee::forDirect(CGM.GetAddrOfFunction(MD), MD);
+ CGCallee callee =
+ CGCallee::forDirect(CGM.GetAddrOfFunction(MD), GlobalDecl(MD));
return EmitCall(getContext().getPointerType(MD->getType()), callee, CE,
ReturnValue);
}
else if (!DevirtualizedMethod)
Callee = CGCallee::forDirect(
CGM.getAddrOfCXXStructor(Dtor, StructorType::Complete, FInfo, Ty),
- Dtor);
+ GlobalDecl(Dtor, Dtor_Complete));
else {
const CXXDestructorDecl *DDtor =
cast<CXXDestructorDecl>(DevirtualizedMethod);
Callee = CGCallee::forDirect(
- CGM.GetAddrOfFunction(GlobalDecl(DDtor, Dtor_Complete), Ty),
- DDtor);
+ CGM.GetAddrOfFunction(GlobalDecl(DDtor, Dtor_Complete), Ty),
+ GlobalDecl(DDtor, Dtor_Complete));
}
EmitCXXMemberOrOperatorCall(
CalleeDecl, Callee, ReturnValue, This.getPointer(),
CGCallee Callee;
if (const CXXConstructorDecl *Ctor = dyn_cast<CXXConstructorDecl>(MD)) {
Callee = CGCallee::forDirect(
- CGM.GetAddrOfFunction(GlobalDecl(Ctor, Ctor_Complete), Ty),
- Ctor);
+ CGM.GetAddrOfFunction(GlobalDecl(Ctor, Ctor_Complete), Ty),
+ GlobalDecl(Ctor, Ctor_Complete));
} else if (UseVirtualCall) {
Callee = CGCallee::forVirtual(CE, MD, This.getAddress(), Ty);
} else {
if (getLangOpts().AppleKext && MD->isVirtual() && HasQualifier)
Callee = BuildAppleKextVirtualCall(MD, Qualifier, Ty);
else if (!DevirtualizedMethod)
- Callee = CGCallee::forDirect(CGM.GetAddrOfFunction(MD, Ty), MD);
+ Callee =
+ CGCallee::forDirect(CGM.GetAddrOfFunction(MD, Ty), GlobalDecl(MD));
else {
- Callee = CGCallee::forDirect(
- CGM.GetAddrOfFunction(DevirtualizedMethod, Ty),
- DevirtualizedMethod);
+ Callee =
+ CGCallee::forDirect(CGM.GetAddrOfFunction(DevirtualizedMethod, Ty),
+ GlobalDecl(DevirtualizedMethod));
}
}
const CallArgList &Args) {
llvm::Instruction *CallOrInvoke;
llvm::Constant *CalleePtr = CGF.CGM.GetAddrOfFunction(CalleeDecl);
- CGCallee Callee = CGCallee::forDirect(CalleePtr, CalleeDecl);
+ CGCallee Callee = CGCallee::forDirect(CalleePtr, GlobalDecl(CalleeDecl));
RValue RV =
CGF.EmitCall(CGF.CGM.getTypes().arrangeFreeFunctionCall(
Args, CalleeType, /*chainCall=*/false),
llvm::Function::Create(FuncTy, llvm::GlobalValue::LinkOnceODRLinkage,
FuncName, &CGM.getModule());
F->setVisibility(llvm::GlobalValue::HiddenVisibility);
- CGM.SetLLVMFunctionAttributes(nullptr, FI, F);
+ CGM.SetLLVMFunctionAttributes(GlobalDecl(), FI, F);
CGM.SetLLVMFunctionAttributesForDefinition(nullptr, F);
IdentifierInfo *II = &Ctx.Idents.get(FuncName);
FunctionDecl *FD = FunctionDecl::Create(
CGM.ErrorUnsupported(
MD, "non-trivial argument copy for return-adjusting thunk");
}
- EmitMustTailThunk(MD, AdjustedThisPtr, CalleePtr);
+ EmitMustTailThunk(CurGD, AdjustedThisPtr, CalleePtr);
return;
}
// Now emit our call.
llvm::Instruction *CallOrInvoke;
- CGCallee Callee = CGCallee::forDirect(CalleePtr, MD);
+ CGCallee Callee = CGCallee::forDirect(CalleePtr, CurGD);
RValue RV = EmitCall(*CurFnInfo, Callee, Slot, CallArgs, &CallOrInvoke);
// Consider return adjustment if we have ThunkInfo.
FinishThunk();
}
-void CodeGenFunction::EmitMustTailThunk(const CXXMethodDecl *MD,
+void CodeGenFunction::EmitMustTailThunk(GlobalDecl GD,
llvm::Value *AdjustedThisPtr,
llvm::Value *CalleePtr) {
// Emitting a musttail call thunk doesn't use any of the CGCall.cpp machinery
// Apply the standard set of call attributes.
unsigned CallingConv;
llvm::AttributeList Attrs;
- CGM.ConstructAttributeList(CalleePtr->getName(), *CurFnInfo, MD, Attrs,
+ CGM.ConstructAttributeList(CalleePtr->getName(), *CurFnInfo, GD, Attrs,
CallingConv, /*AttrOnCallSite=*/true);
Call->setAttributes(Attrs);
Call->setCallingConv(static_cast<llvm::CallingConv::ID>(CallingConv));
// Now build up the set of caller features and verify that all the required
// features are there.
llvm::StringMap<bool> CallerFeatureMap;
- CGM.getFunctionFeatureMap(CallerFeatureMap, FD);
+ CGM.getFunctionFeatureMap(CallerFeatureMap, GlobalDecl().getWithDecl(FD));
// If we have at least one of the features in the feature list return
// true, otherwise return false.
void FinishThunk();
/// Emit a musttail call for a thunk with a potentially adjusted this pointer.
- void EmitMustTailThunk(const CXXMethodDecl *MD, llvm::Value *AdjustedThisPtr,
+ void EmitMustTailThunk(GlobalDecl GD, llvm::Value *AdjustedThisPtr,
llvm::Value *Callee);
/// Generate a thunk for the given method.
RValue EmitNVPTXDevicePrintfCallExpr(const CallExpr *E,
ReturnValueSlot ReturnValue);
- RValue EmitBuiltinExpr(const FunctionDecl *FD,
- unsigned BuiltinID, const CallExpr *E,
- ReturnValueSlot ReturnValue);
+ RValue EmitBuiltinExpr(const GlobalDecl GD, unsigned BuiltinID,
+ const CallExpr *E, ReturnValueSlot ReturnValue);
RValue emitRotate(const CallExpr *E, bool IsRotateRight);
static void AppendCPUSpecificCPUDispatchMangling(const CodeGenModule &CGM,
const CPUSpecificAttr *Attr,
+ unsigned CPUIndex,
raw_ostream &Out) {
// cpu_specific gets the current name, dispatch gets the resolver if IFunc is
// supported.
if (Attr)
- Out << getCPUSpecificMangling(CGM, Attr->getCurCPUName()->getName());
+ Out << getCPUSpecificMangling(CGM, Attr->getCPUName(CPUIndex)->getName());
else if (CGM.getTarget().supportsIFunc())
Out << ".resolver";
}
if (const auto *FD = dyn_cast<FunctionDecl>(ND))
if (FD->isMultiVersion() && !OmitMultiVersionMangling) {
if (FD->isCPUDispatchMultiVersion() || FD->isCPUSpecificMultiVersion())
- AppendCPUSpecificCPUDispatchMangling(
- CGM, FD->getAttr<CPUSpecificAttr>(), Out);
+ AppendCPUSpecificCPUDispatchMangling(CGM,
+ FD->getAttr<CPUSpecificAttr>(),
+ GD.getMultiVersionIndex(), Out);
else
AppendTargetMangling(CGM, FD->getAttr<TargetAttr>(), Out);
}
}
}
- const auto *FD = dyn_cast<FunctionDecl>(GD.getDecl());
- // Since CPUSpecific can require multiple emits per decl, store the manglings
- // separately.
- if (FD &&
- (FD->isCPUDispatchMultiVersion() || FD->isCPUSpecificMultiVersion())) {
- const auto *SD = FD->getAttr<CPUSpecificAttr>();
-
- std::pair<GlobalDecl, unsigned> SpecCanonicalGD{
- CanonicalGD,
- SD ? SD->ActiveArgIndex : std::numeric_limits<unsigned>::max()};
-
- auto FoundName = CPUSpecificMangledDeclNames.find(SpecCanonicalGD);
- if (FoundName != CPUSpecificMangledDeclNames.end())
- return FoundName->second;
-
- auto Result = CPUSpecificManglings.insert(
- std::make_pair(getMangledNameImpl(*this, GD, FD), SpecCanonicalGD));
- return CPUSpecificMangledDeclNames[SpecCanonicalGD] = Result.first->first();
- }
-
auto FoundName = MangledDeclNames.find(CanonicalGD);
if (FoundName != MangledDeclNames.end())
return FoundName->second;
return llvm::ConstantInt::get(Int64Ty, llvm::MD5Hash(MDS->getString()));
}
-void CodeGenModule::SetLLVMFunctionAttributes(const Decl *D,
+void CodeGenModule::SetLLVMFunctionAttributes(GlobalDecl GD,
const CGFunctionInfo &Info,
llvm::Function *F) {
unsigned CallingConv;
llvm::AttributeList PAL;
- ConstructAttributeList(F->getName(), Info, D, PAL, CallingConv, false);
+ ConstructAttributeList(F->getName(), Info, GD, PAL, CallingConv, false);
F->setAttributes(PAL);
F->setCallingConv(static_cast<llvm::CallingConv::ID>(CallingConv));
}
}
}
-bool CodeGenModule::GetCPUAndFeaturesAttributes(const Decl *D,
+bool CodeGenModule::GetCPUAndFeaturesAttributes(GlobalDecl GD,
llvm::AttrBuilder &Attrs) {
// Add target-cpu and target-features attributes to functions. If
// we have a decl for the function and it has a target attribute then
// parse that and add it to the feature set.
StringRef TargetCPU = getTarget().getTargetOpts().CPU;
std::vector<std::string> Features;
- const auto *FD = dyn_cast_or_null<FunctionDecl>(D);
+ const auto *FD = dyn_cast_or_null<FunctionDecl>(GD.getDecl());
FD = FD ? FD->getMostRecentDecl() : FD;
const auto *TD = FD ? FD->getAttr<TargetAttr>() : nullptr;
const auto *SD = FD ? FD->getAttr<CPUSpecificAttr>() : nullptr;
bool AddedAttr = false;
if (TD || SD) {
llvm::StringMap<bool> FeatureMap;
- getFunctionFeatureMap(FeatureMap, FD);
+ getFunctionFeatureMap(FeatureMap, GD);
// Produce the canonical string for this set of features.
for (const llvm::StringMap<bool>::value_type &Entry : FeatureMap)
F->addFnAttr("implicit-section-name", SA->getName());
llvm::AttrBuilder Attrs;
- if (GetCPUAndFeaturesAttributes(D, Attrs)) {
+ if (GetCPUAndFeaturesAttributes(GD, Attrs)) {
// We know that GetCPUAndFeaturesAttributes will always have the
// newest set, since it has the newest possible FunctionDecl, so the
// new ones should replace the old.
llvm::Function *F,
const CGFunctionInfo &FI) {
const Decl *D = GD.getDecl();
- SetLLVMFunctionAttributes(D, FI, F);
+ SetLLVMFunctionAttributes(GD, FI, F);
SetLLVMFunctionAttributesForDefinition(D, F);
F->setLinkage(llvm::Function::InternalLinkage);
const auto *FD = cast<FunctionDecl>(GD.getDecl());
if (!IsIncompleteFunction) {
- SetLLVMFunctionAttributes(FD, getTypes().arrangeGlobalDeclaration(GD), F);
+ SetLLVMFunctionAttributes(GD, getTypes().arrangeGlobalDeclaration(GD), F);
// Setup target-specific attributes.
if (F->isDeclaration())
getTargetCodeGenInfo().setTargetAttributes(FD, F, *this);
return CodeGenOpts.OptimizationLevel > 0;
}
+void CodeGenModule::EmitMultiVersionFunctionDefinition(GlobalDecl GD,
+ llvm::GlobalValue *GV) {
+ const auto *FD = cast<FunctionDecl>(GD.getDecl());
+
+ if (FD->isCPUSpecificMultiVersion()) {
+ auto *Spec = FD->getAttr<CPUSpecificAttr>();
+ for (unsigned I = 0; I < Spec->cpus_size(); ++I)
+ EmitGlobalFunctionDefinition(GD.getWithMultiVersionIndex(I), nullptr);
+ // Requires multiple emits.
+ } else
+ EmitGlobalFunctionDefinition(GD, GV);
+}
+
void CodeGenModule::EmitGlobalDefinition(GlobalDecl GD, llvm::GlobalValue *GV) {
const auto *D = cast<ValueDecl>(GD.getDecl());
Context.getSourceManager(),
"Generating code for declaration");
- if (isa<FunctionDecl>(D)) {
+ if (const auto *FD = dyn_cast<FunctionDecl>(D)) {
// At -O0, don't generate IR for functions with available_externally
// linkage.
if (!shouldEmitFunction(GD))
ABI->emitCXXStructor(CD, getFromCtorType(GD.getCtorType()));
else if (const auto *DD = dyn_cast<CXXDestructorDecl>(Method))
ABI->emitCXXStructor(DD, getFromDtorType(GD.getDtorType()));
+ else if (FD->isMultiVersion())
+ EmitMultiVersionFunctionDefinition(GD, GV);
else
EmitGlobalFunctionDefinition(GD, GV);
return;
}
+ if (FD->isMultiVersion())
+ return EmitMultiVersionFunctionDefinition(GD, GV);
return EmitGlobalFunctionDefinition(GD, GV);
}
SmallVector<CodeGenFunction::MultiVersionResolverOption, 10> Options;
const TargetInfo &Target = getTarget();
+ unsigned Index = 0;
for (const IdentifierInfo *II : DD->cpus()) {
// Get the name of the target function so we can look it up/create it.
std::string MangledName = getMangledNameImpl(*this, GD, FD, true) +
getCPUSpecificMangling(*this, II->getName());
- llvm::Constant *Func = GetOrCreateLLVMFunction(
- MangledName, DeclTy, GD, /*ForVTable=*/false, /*DontDefer=*/true,
- /*IsThunk=*/false, llvm::AttributeList(), ForDefinition);
+
+ llvm::Constant *Func = GetGlobalValue(MangledName);
+
+ if (!Func)
+ Func = GetOrCreateLLVMFunction(
+ MangledName, DeclTy, GD.getWithMultiVersionIndex(Index),
+ /*ForVTable=*/false, /*DontDefer=*/true,
+ /*IsThunk=*/false, llvm::AttributeList(), ForDefinition);
+
llvm::SmallVector<StringRef, 32> Features;
Target.getCPUSpecificCPUDispatchFeatures(II->getName(), Features);
llvm::transform(Features, Features.begin(),
return !Target.validateCpuSupports(Feat);
}), Features.end());
Options.emplace_back(cast<llvm::Function>(Func), StringRef{}, Features);
+ ++Index;
}
llvm::sort(
AddGlobalDtor(Fn, DA->getPriority());
if (D->hasAttr<AnnotateAttr>())
AddGlobalAnnotations(D, Fn);
-
- if (D->isCPUSpecificMultiVersion()) {
- auto *Spec = D->getAttr<CPUSpecificAttr>();
- // If there is another specific version we need to emit, do so here.
- if (Spec->ActiveArgIndex + 1 < Spec->cpus_size()) {
- ++Spec->ActiveArgIndex;
- EmitGlobalFunctionDefinition(GD, nullptr);
- }
- }
}
void CodeGenModule::EmitAliasDefinition(GlobalDecl GD) {
// Fills in the supplied string map with the set of target features for the
// passed in function.
void CodeGenModule::getFunctionFeatureMap(llvm::StringMap<bool> &FeatureMap,
- const FunctionDecl *FD) {
+ GlobalDecl GD) {
StringRef TargetCPU = Target.getTargetOpts().CPU;
+ const FunctionDecl *FD = GD.getDecl()->getAsFunction();
if (const auto *TD = FD->getAttr<TargetAttr>()) {
TargetAttr::ParsedTargetAttr ParsedAttr = filterFunctionTargetAttrs(TD);
ParsedAttr.Features);
} else if (const auto *SD = FD->getAttr<CPUSpecificAttr>()) {
llvm::SmallVector<StringRef, 32> FeaturesTmp;
- Target.getCPUSpecificCPUDispatchFeatures(SD->getCurCPUName()->getName(),
- FeaturesTmp);
+ Target.getCPUSpecificCPUDispatchFeatures(
+ SD->getCPUName(GD.getMultiVersionIndex())->getName(), FeaturesTmp);
std::vector<std::string> Features(FeaturesTmp.begin(), FeaturesTmp.end());
Target.initFeatureMap(FeatureMap, getDiags(), TargetCPU, Features);
} else {
const CGFunctionInfo &FI);
/// Set the LLVM function attributes (sext, zext, etc).
- void SetLLVMFunctionAttributes(const Decl *D,
- const CGFunctionInfo &Info,
+ void SetLLVMFunctionAttributes(GlobalDecl GD, const CGFunctionInfo &Info,
llvm::Function *F);
/// Set the LLVM function attributes which only apply to a function
// Fills in the supplied string map with the set of target features for the
// passed in function.
- void getFunctionFeatureMap(llvm::StringMap<bool> &FeatureMap,
- const FunctionDecl *FD);
+ void getFunctionFeatureMap(llvm::StringMap<bool> &FeatureMap, GlobalDecl GD);
StringRef getMangledName(GlobalDecl GD);
StringRef getBlockMangledName(GlobalDecl GD, const BlockDecl *BD);
ForDefinition_t IsForDefinition
= NotForDefinition);
- bool GetCPUAndFeaturesAttributes(const Decl *D,
+ bool GetCPUAndFeaturesAttributes(GlobalDecl GD,
llvm::AttrBuilder &AttrBuilder);
void setNonAliasAttributes(GlobalDecl GD, llvm::GlobalObject *GO);
void EmitGlobalDefinition(GlobalDecl D, llvm::GlobalValue *GV = nullptr);
void EmitGlobalFunctionDefinition(GlobalDecl GD, llvm::GlobalValue *GV);
+ void EmitMultiVersionFunctionDefinition(GlobalDecl GD, llvm::GlobalValue *GV);
+
void EmitGlobalVarDefinition(const VarDecl *D, bool IsTentative = false);
void EmitAliasDefinition(GlobalDecl GD);
void emitIFuncDefinition(GlobalDecl GD);
Type != Dtor_Base && DD->isVirtual())
Callee = CGF.BuildAppleKextVirtualDestructorCall(DD, Type, DD->getParent());
else
- Callee =
- CGCallee::forDirect(CGM.getAddrOfCXXStructor(DD, getFromDtorType(Type)),
- DD);
+ Callee = CGCallee::forDirect(
+ CGM.getAddrOfCXXStructor(DD, getFromDtorType(Type)), GD);
CGF.EmitCXXMemberOrOperatorCall(DD, Callee, ReturnValueSlot(),
This.getPointer(), VTT, VTTTy,
VFunc = VFuncLoad;
}
- CGCallee Callee(MethodDecl->getCanonicalDecl(), VFunc);
+ CGCallee Callee(GD, VFunc);
return Callee;
}
llvm::Function::Create(FnTy, getThreadLocalWrapperLinkage(VD, CGM),
WrapperName.str(), &CGM.getModule());
- CGM.SetLLVMFunctionAttributes(nullptr, FI, Wrapper);
+ CGM.SetLLVMFunctionAttributes(GlobalDecl(), FI, Wrapper);
if (VD->hasDefinition())
CGM.SetLLVMFunctionAttributesForDefinition(nullptr, Wrapper);
llvm::GlobalVariable::ExternalWeakLinkage,
InitFnName.str(), &CGM.getModule());
const CGFunctionInfo &FI = CGM.getTypes().arrangeNullaryFunction();
- CGM.SetLLVMFunctionAttributes(nullptr, FI, cast<llvm::Function>(Init));
+ CGM.SetLLVMFunctionAttributes(GlobalDecl(), FI,
+ cast<llvm::Function>(Init));
}
if (Init) {
if (Type == Dtor_Complete && DD->getParent()->getNumVBases() == 0)
Type = Dtor_Base;
- CGCallee Callee = CGCallee::forDirect(
- CGM.getAddrOfCXXStructor(DD, getFromDtorType(Type)),
- DD);
+ CGCallee Callee =
+ CGCallee::forDirect(CGM.getAddrOfCXXStructor(DD, getFromDtorType(Type)),
+ GlobalDecl(DD, Type));
if (DD->isVirtual()) {
assert(Type != CXXDtorType::Dtor_Deleting &&
VFunc = Builder.CreateAlignedLoad(VFuncPtr, CGF.getPointerAlign());
}
- CGCallee Callee(MethodDecl->getCanonicalDecl(), VFunc);
+ CGCallee Callee(GD, VFunc);
return Callee;
}
// Call the destructor with our arguments.
llvm::Constant *CalleePtr =
CGM.getAddrOfCXXStructor(CD, StructorType::Complete);
- CGCallee Callee = CGCallee::forDirect(CalleePtr, CD);
+ CGCallee Callee =
+ CGCallee::forDirect(CalleePtr, GlobalDecl(CD, Ctor_Complete));
const CGFunctionInfo &CalleeInfo = CGM.getTypes().arrangeCXXConstructorCall(
Args, CD, Ctor_Complete, ExtraArgs.Prefix, ExtraArgs.Suffix);
CGF.EmitCall(CalleeInfo, Callee, ReturnValueSlot(), Args);
/// nested-namespace-definition
///
/// named-namespace-definition:
-/// 'inline'[opt] 'namespace' attributes[opt] identifier '{' namespace-body '}'
+/// 'inline'[opt] 'namespace' attributes[opt] identifier '{'
+/// namespace-body '}'
///
/// unnamed-namespace-definition:
/// 'inline'[opt] 'namespace' attributes[opt] '{' namespace-body '}'
///
/// nested-namespace-definition:
-/// 'namespace' enclosing-namespace-specifier '::' 'inline'[opt] identifier '{' namespace-body '}'
+/// 'namespace' enclosing-namespace-specifier '::' 'inline'[opt]
+/// identifier '{' namespace-body '}'
///
/// enclosing-namespace-specifier:
/// identifier
assert(!ImplicitUsingDirectiveDecl &&
"nested namespace definition cannot define anonymous namespace");
- ParseInnerNamespace(InnerNSs, ++index, InlineLoc,
- attrs, Tracker);
+ ParseInnerNamespace(InnerNSs, ++index, InlineLoc, attrs, Tracker);
NamespaceScope.Exit();
Actions.ActOnFinishNamespaceDef(NamespcDecl, Tracker.getCloseLocation());
// WINDOWS-NOT: call i32 @GenericAndPentium.A
// WINDOWS-NOT: call void @llvm.trap
+ATTR(cpu_dispatch(atom, pentium))
+int DispatchFirst(void);
+// LINUX: define i32 ()* @DispatchFirst.resolver
+// LINUX: ret i32 ()* @DispatchFirst.O
+// LINUX: ret i32 ()* @DispatchFirst.B
+
+// WINDOWS: define dso_local i32 @DispatchFirst()
+// WINDOWS: %[[RET:.+]] = musttail call i32 @DispatchFirst.O()
+// WINDOWS-NEXT: ret i32 %[[RET]]
+// WINDOWS: %[[RET:.+]] = musttail call i32 @DispatchFirst.B()
+// WINDOWS-NEXT: ret i32 %[[RET]]
+
+ATTR(cpu_specific(atom))
+int DispatchFirst(void) {return 0;}
+// LINUX: define i32 @DispatchFirst.O
+// LINUX: ret i32 0
+
+// WINDOWS: define dso_local i32 @DispatchFirst.O()
+// WINDOWS: ret i32 0
+
+ATTR(cpu_specific(pentium))
+int DispatchFirst(void) {return 1;}
+// LINUX: define i32 @DispatchFirst.B
+// LINUX: ret i32 1
+
+// WINDOWS: define dso_local i32 @DispatchFirst.B
+// WINDOWS: ret i32 1
+
// CHECK: attributes #[[S]] = {{.*}}"target-features"="+avx,+cmov,+f16c,+mmx,+popcnt,+sse,+sse2,+sse3,+sse4.1,+sse4.2,+ssse3,+x87,+xsave"
// CHECK: attributes #[[K]] = {{.*}}"target-features"="+adx,+avx,+avx2,+avx512cd,+avx512er,+avx512f,+avx512pf,+bmi,+cmov,+f16c,+fma,+lzcnt,+mmx,+movbe,+popcnt,+sse,+sse2,+sse3,+sse4.1,+sse4.2,+ssse3,+x87,+xsave"
// CHECK: attributes #[[O]] = {{.*}}"target-features"="+cmov,+mmx,+movbe,+sse,+sse2,+sse3,+ssse3,+x87"