Author: yaxunl Date: Thu Mar 29 08:02:08 2018 New Revision: 328795 URL: http://llvm.org/viewvc/llvm-project?rev=328795&view=rev Log: Set calling convention for CUDA kernel
This patch sets target specific calling convention for CUDA kernels in IR. Patch by Greg Rodgers. Revised and lit test added by Yaxun Liu. Differential Revision: https://reviews.llvm.org/D44747 Added: cfe/trunk/test/CodeGenCUDA/kernel-amdgcn.cu Modified: cfe/trunk/include/clang/Basic/Specifiers.h cfe/trunk/lib/AST/ItaniumMangle.cpp cfe/trunk/lib/AST/Type.cpp cfe/trunk/lib/AST/TypePrinter.cpp cfe/trunk/lib/CodeGen/CGCall.cpp cfe/trunk/lib/CodeGen/CGDebugInfo.cpp cfe/trunk/lib/CodeGen/TargetInfo.cpp cfe/trunk/lib/CodeGen/TargetInfo.h cfe/trunk/lib/Sema/SemaExpr.cpp cfe/trunk/lib/Sema/SemaOverload.cpp cfe/trunk/lib/Sema/SemaType.cpp cfe/trunk/tools/libclang/CXType.cpp Modified: cfe/trunk/include/clang/Basic/Specifiers.h URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Basic/Specifiers.h?rev=328795&r1=328794&r2=328795&view=diff ============================================================================== --- cfe/trunk/include/clang/Basic/Specifiers.h (original) +++ cfe/trunk/include/clang/Basic/Specifiers.h Thu Mar 29 08:02:08 2018 @@ -231,23 +231,24 @@ namespace clang { /// \brief CallingConv - Specifies the calling convention that a function uses. enum CallingConv { - CC_C, // __attribute__((cdecl)) - CC_X86StdCall, // __attribute__((stdcall)) - CC_X86FastCall, // __attribute__((fastcall)) - CC_X86ThisCall, // __attribute__((thiscall)) + CC_C, // __attribute__((cdecl)) + CC_X86StdCall, // __attribute__((stdcall)) + CC_X86FastCall, // __attribute__((fastcall)) + CC_X86ThisCall, // __attribute__((thiscall)) CC_X86VectorCall, // __attribute__((vectorcall)) - CC_X86Pascal, // __attribute__((pascal)) - CC_Win64, // __attribute__((ms_abi)) - CC_X86_64SysV, // __attribute__((sysv_abi)) - CC_X86RegCall, // __attribute__((regcall)) - CC_AAPCS, // __attribute__((pcs("aapcs"))) - CC_AAPCS_VFP, // __attribute__((pcs("aapcs-vfp"))) - CC_IntelOclBicc, // __attribute__((intel_ocl_bicc)) - CC_SpirFunction, // default for OpenCL functions on SPIR target - CC_OpenCLKernel, // inferred for OpenCL kernels - CC_Swift, // __attribute__((swiftcall)) - CC_PreserveMost, // __attribute__((preserve_most)) - CC_PreserveAll, // __attribute__((preserve_all)) + CC_X86Pascal, // __attribute__((pascal)) + CC_Win64, // __attribute__((ms_abi)) + CC_X86_64SysV, // __attribute__((sysv_abi)) + CC_X86RegCall, // __attribute__((regcall)) + CC_AAPCS, // __attribute__((pcs("aapcs"))) + CC_AAPCS_VFP, // __attribute__((pcs("aapcs-vfp"))) + CC_IntelOclBicc, // __attribute__((intel_ocl_bicc)) + CC_SpirFunction, // default for OpenCL functions on SPIR target + CC_OpenCLKernel, // inferred for OpenCL kernels + CC_Swift, // __attribute__((swiftcall)) + CC_PreserveMost, // __attribute__((preserve_most)) + CC_PreserveAll, // __attribute__((preserve_all)) + CC_CUDAKernel, // inferred for CUDA kernels }; /// \brief Checks whether the given calling convention supports variadic Modified: cfe/trunk/lib/AST/ItaniumMangle.cpp URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/AST/ItaniumMangle.cpp?rev=328795&r1=328794&r2=328795&view=diff ============================================================================== --- cfe/trunk/lib/AST/ItaniumMangle.cpp (original) +++ cfe/trunk/lib/AST/ItaniumMangle.cpp Thu Mar 29 08:02:08 2018 @@ -2628,6 +2628,7 @@ StringRef CXXNameMangler::getCallingConv case CC_OpenCLKernel: case CC_PreserveMost: case CC_PreserveAll: + case CC_CUDAKernel: // FIXME: we should be mangling all of the above. return ""; Modified: cfe/trunk/lib/AST/Type.cpp URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/AST/Type.cpp?rev=328795&r1=328794&r2=328795&view=diff ============================================================================== --- cfe/trunk/lib/AST/Type.cpp (original) +++ cfe/trunk/lib/AST/Type.cpp Thu Mar 29 08:02:08 2018 @@ -2752,6 +2752,7 @@ StringRef FunctionType::getNameForCallCo case CC_Swift: return "swiftcall"; case CC_PreserveMost: return "preserve_most"; case CC_PreserveAll: return "preserve_all"; + case CC_CUDAKernel: return "cuda_kernel"; } llvm_unreachable("Invalid calling convention."); Modified: cfe/trunk/lib/AST/TypePrinter.cpp URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/AST/TypePrinter.cpp?rev=328795&r1=328794&r2=328795&view=diff ============================================================================== --- cfe/trunk/lib/AST/TypePrinter.cpp (original) +++ cfe/trunk/lib/AST/TypePrinter.cpp Thu Mar 29 08:02:08 2018 @@ -780,6 +780,10 @@ void TypePrinter::printFunctionAfter(con case CC_OpenCLKernel: // Do nothing. These CCs are not available as attributes. break; + case CC_CUDAKernel: + // ToDo: print this before the function. + OS << " __global__"; + break; case CC_Swift: OS << " __attribute__((swiftcall))"; break; Modified: cfe/trunk/lib/CodeGen/CGCall.cpp URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CGCall.cpp?rev=328795&r1=328794&r2=328795&view=diff ============================================================================== --- cfe/trunk/lib/CodeGen/CGCall.cpp (original) +++ cfe/trunk/lib/CodeGen/CGCall.cpp Thu Mar 29 08:02:08 2018 @@ -64,6 +64,7 @@ unsigned CodeGenTypes::ClangCallConvToLL case CC_PreserveMost: return llvm::CallingConv::PreserveMost; case CC_PreserveAll: return llvm::CallingConv::PreserveAll; case CC_Swift: return llvm::CallingConv::Swift; + case CC_CUDAKernel: return CGM.getTargetCodeGenInfo().getCUDAKernelCallingConv(); } } Modified: cfe/trunk/lib/CodeGen/CGDebugInfo.cpp URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CGDebugInfo.cpp?rev=328795&r1=328794&r2=328795&view=diff ============================================================================== --- cfe/trunk/lib/CodeGen/CGDebugInfo.cpp (original) +++ cfe/trunk/lib/CodeGen/CGDebugInfo.cpp Thu Mar 29 08:02:08 2018 @@ -1022,6 +1022,9 @@ static unsigned getDwarfCC(CallingConv C return llvm::dwarf::DW_CC_LLVM_PreserveAll; case CC_X86RegCall: return llvm::dwarf::DW_CC_LLVM_X86RegCall; + case CC_CUDAKernel: + // ToDo: Add llvm::dwarf::DW_CC_LLVM_CUDAKernel; + return 0; } return 0; } Modified: cfe/trunk/lib/CodeGen/TargetInfo.cpp URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/TargetInfo.cpp?rev=328795&r1=328794&r2=328795&view=diff ============================================================================== --- cfe/trunk/lib/CodeGen/TargetInfo.cpp (original) +++ cfe/trunk/lib/CodeGen/TargetInfo.cpp Thu Mar 29 08:02:08 2018 @@ -431,6 +431,10 @@ unsigned TargetCodeGenInfo::getOpenCLKer return llvm::CallingConv::SPIR_KERNEL; } +unsigned TargetCodeGenInfo::getCUDAKernelCallingConv() const { + return llvm::CallingConv::C; +} + llvm::Constant *TargetCodeGenInfo::getNullPointer(const CodeGen::CodeGenModule &CGM, llvm::PointerType *T, QualType QT) const { return llvm::ConstantPointerNull::get(T); @@ -7635,6 +7639,7 @@ public: void setTargetAttributes(const Decl *D, llvm::GlobalValue *GV, CodeGen::CodeGenModule &M) const override; unsigned getOpenCLKernelCallingConv() const override; + unsigned getCUDAKernelCallingConv() const override; llvm::Constant *getNullPointer(const CodeGen::CodeGenModule &CGM, llvm::PointerType *T, QualType QT) const override; @@ -7722,6 +7727,10 @@ unsigned AMDGPUTargetCodeGenInfo::getOpe return llvm::CallingConv::AMDGPU_KERNEL; } +unsigned AMDGPUTargetCodeGenInfo::getCUDAKernelCallingConv() const { + return llvm::CallingConv::AMDGPU_KERNEL; +} + // Currently LLVM assumes null pointers always have value 0, // which results in incorrectly transformed IR. Therefore, instead of // emitting null pointers in private and local address spaces, a null Modified: cfe/trunk/lib/CodeGen/TargetInfo.h URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/TargetInfo.h?rev=328795&r1=328794&r2=328795&view=diff ============================================================================== --- cfe/trunk/lib/CodeGen/TargetInfo.h (original) +++ cfe/trunk/lib/CodeGen/TargetInfo.h Thu Mar 29 08:02:08 2018 @@ -223,6 +223,9 @@ public: /// Get LLVM calling convention for OpenCL kernel. virtual unsigned getOpenCLKernelCallingConv() const; + /// Get LLVM calling convention for CUDA kernel. + virtual unsigned getCUDAKernelCallingConv() const; + /// Get target specific null pointer. /// \param T is the LLVM type of the null pointer. /// \param QT is the clang QualType of the null pointer. Modified: cfe/trunk/lib/Sema/SemaExpr.cpp URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Sema/SemaExpr.cpp?rev=328795&r1=328794&r2=328795&view=diff ============================================================================== --- cfe/trunk/lib/Sema/SemaExpr.cpp (original) +++ cfe/trunk/lib/Sema/SemaExpr.cpp Thu Mar 29 08:02:08 2018 @@ -25,6 +25,7 @@ #include "clang/AST/ExprObjC.h" #include "clang/AST/ExprOpenMP.h" #include "clang/AST/RecursiveASTVisitor.h" +#include "clang/AST/Type.h" #include "clang/AST/TypeLoc.h" #include "clang/Basic/PartialDiagnostic.h" #include "clang/Basic/SourceManager.h" @@ -1657,6 +1658,16 @@ Sema::BuildDeclRefExpr(ValueDecl *D, Qua isa<VarDecl>(D) && NeedToCaptureVariable(cast<VarDecl>(D), NameInfo.getLoc()); + // Drop CUDA kernel calling convention since it is invisible to the user + // in DRE. + if (const auto *FT = Ty->getAs<FunctionType>()) { + if (FT->getCallConv() == CC_CUDAKernel) { + FT = Context.adjustFunctionType(FT, + FT->getExtInfo().withCallingConv(CC_C)); + Ty = QualType(FT, Ty.getQualifiers().getAsOpaqueValue()); + } + } + DeclRefExpr *E; if (isa<VarTemplateSpecializationDecl>(D)) { VarTemplateSpecializationDecl *VarSpec = Modified: cfe/trunk/lib/Sema/SemaOverload.cpp URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Sema/SemaOverload.cpp?rev=328795&r1=328794&r2=328795&view=diff ============================================================================== --- cfe/trunk/lib/Sema/SemaOverload.cpp (original) +++ cfe/trunk/lib/Sema/SemaOverload.cpp Thu Mar 29 08:02:08 2018 @@ -1481,7 +1481,6 @@ bool Sema::IsFunctionConversion(QualType .getTypePtr()); Changed = true; } - // Convert FromFPT's ExtParameterInfo if necessary. The conversion is valid // only if the ExtParameterInfo lists of the two function prototypes can be // merged and the merged list is identical to ToFPT's ExtParameterInfo list. Modified: cfe/trunk/lib/Sema/SemaType.cpp URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Sema/SemaType.cpp?rev=328795&r1=328794&r2=328795&view=diff ============================================================================== --- cfe/trunk/lib/Sema/SemaType.cpp (original) +++ cfe/trunk/lib/Sema/SemaType.cpp Thu Mar 29 08:02:08 2018 @@ -3316,6 +3316,18 @@ getCCForDeclaratorChunk(Sema &S, Declara CallingConv CC = S.Context.getDefaultCallingConvention(FTI.isVariadic, IsCXXInstanceMethod); + // Attribute AT_CUDAGlobal affects the calling convention for AMDGPU targets. + // This is the simplest place to infer calling convention for CUDA kernels. + if (S.getLangOpts().CUDA && S.getLangOpts().CUDAIsDevice) { + for (const AttributeList *Attr = D.getDeclSpec().getAttributes().getList(); + Attr; Attr = Attr->getNext()) { + if (Attr->getKind() == AttributeList::AT_CUDAGlobal) { + CC = CC_CUDAKernel; + break; + } + } + } + // Attribute AT_OpenCLKernel affects the calling convention for SPIR // and AMDGPU targets, hence it cannot be treated as a calling // convention attribute. This is the simplest place to infer Added: cfe/trunk/test/CodeGenCUDA/kernel-amdgcn.cu URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGenCUDA/kernel-amdgcn.cu?rev=328795&view=auto ============================================================================== --- cfe/trunk/test/CodeGenCUDA/kernel-amdgcn.cu (added) +++ cfe/trunk/test/CodeGenCUDA/kernel-amdgcn.cu Thu Mar 29 08:02:08 2018 @@ -0,0 +1,29 @@ +// RUN: %clang_cc1 -triple amdgcn -fcuda-is-device -emit-llvm %s -o - | FileCheck %s +#include "Inputs/cuda.h" + +// CHECK: define amdgpu_kernel void @_ZN1A6kernelEv +class A { +public: + static __global__ void kernel(){} +}; + +// CHECK: define void @_Z10non_kernelv +__device__ void non_kernel(){} + +// CHECK: define amdgpu_kernel void @_Z6kerneli +__global__ void kernel(int x) { + non_kernel(); +} + +// CHECK: define amdgpu_kernel void @_Z15template_kernelI1AEvT_ +template<class T> +__global__ void template_kernel(T x) {} + +void launch(void *f); + +int main() { + launch((void*)A::kernel); + launch((void*)kernel); + launch((void*)template_kernel<A>); + return 0; +} Modified: cfe/trunk/tools/libclang/CXType.cpp URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/tools/libclang/CXType.cpp?rev=328795&r1=328794&r2=328795&view=diff ============================================================================== --- cfe/trunk/tools/libclang/CXType.cpp (original) +++ cfe/trunk/tools/libclang/CXType.cpp Thu Mar 29 08:02:08 2018 @@ -626,6 +626,7 @@ CXCallingConv clang_getFunctionTypeCalli TCALLINGCONV(PreserveAll); case CC_SpirFunction: return CXCallingConv_Unexposed; case CC_OpenCLKernel: return CXCallingConv_Unexposed; + case CC_CUDAKernel: return CXCallingConv_Unexposed; break; } #undef TCALLINGCONV _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
