Index: llvm/trunk/lib/Target/TargetMachine.cpp =================================================================== --- llvm/trunk/lib/Target/TargetMachine.cpp +++ llvm/trunk/lib/Target/TargetMachine.cpp @@ -167,6 +167,13 @@ if (GV && !GV->isDeclarationForLinker()) return true; + // A symbol marked nonlazybind should not be accessed with a plt. If the + // symbol turns out to be external, the linker will convert a direct + // access to an access via the plt, so don't assume it is local. + const Function *F = dyn_cast_or_null(GV); + if (F && F->hasFnAttribute(Attribute::NonLazyBind)) + return false; + bool IsTLS = GV && GV->isThreadLocal(); bool IsAccessViaCopyRelocs = Options.MCOptions.MCPIECopyRelocations && GV && isa(GV); Index: llvm/trunk/lib/Target/X86/X86Subtarget.cpp =================================================================== --- llvm/trunk/lib/Target/X86/X86Subtarget.cpp +++ llvm/trunk/lib/Target/X86/X86Subtarget.cpp @@ -144,15 +144,6 @@ unsigned char X86Subtarget::classifyGlobalFunctionReference(const GlobalValue *GV, const Module &M) const { - const Function *F = dyn_cast_or_null(GV); - - // Do not use the PLT when explicitly told to do so for ELF 64-bit - // target. - if (isTargetELF() && is64Bit() && F && - F->hasFnAttribute(Attribute::NonLazyBind) && - GV->isDeclarationForLinker()) - return X86II::MO_GOTPCREL; - if (TM.shouldAssumeDSOLocal(M, GV)) return X86II::MO_NO_FLAG; @@ -162,12 +153,16 @@ return X86II::MO_DLLIMPORT; } + const Function *F = dyn_cast_or_null(GV); + if (isTargetELF()) { if (is64Bit() && F && (CallingConv::X86_RegCall == F->getCallingConv())) // According to psABI, PLT stub clobbers XMM8-XMM15. // In Regcall calling convention those registers are used for passing // parameters. Thus we need to prevent lazy binding in Regcall. return X86II::MO_GOTPCREL; + if (F && F->hasFnAttribute(Attribute::NonLazyBind) && is64Bit()) + return X86II::MO_GOTPCREL; return X86II::MO_PLT; } Index: llvm/trunk/test/CodeGen/X86/no-plt.ll =================================================================== --- llvm/trunk/test/CodeGen/X86/no-plt.ll +++ llvm/trunk/test/CodeGen/X86/no-plt.ll @@ -6,12 +6,14 @@ define i32 @main() #0 { ; X64: callq *_Z3foov@GOTPCREL(%rip) ; X64: callq _Z3barv +; X64: callq _Z3bazv entry: %retval = alloca i32, align 4 store i32 0, i32* %retval, align 4 %call1 = call i32 @_Z3foov() %call2 = call i32 @_Z3barv() + %call3 = call i32 @_Z3bazv() ret i32 0 } @@ -20,4 +22,9 @@ declare i32 @_Z3barv() #2 +; Function Attrs: nonlazybind +declare hidden i32 @_Z3bazv() #3 + + attributes #1 = { nonlazybind } +attributes #3 = { nonlazybind }