diff options
Diffstat (limited to 'clang/lib/CodeGen/Targets/AMDGPU.cpp')
| -rw-r--r-- | clang/lib/CodeGen/Targets/AMDGPU.cpp | 12 |
1 files changed, 9 insertions, 3 deletions
diff --git a/clang/lib/CodeGen/Targets/AMDGPU.cpp b/clang/lib/CodeGen/Targets/AMDGPU.cpp index 16d5919..e4ad078 100644 --- a/clang/lib/CodeGen/Targets/AMDGPU.cpp +++ b/clang/lib/CodeGen/Targets/AMDGPU.cpp @@ -439,9 +439,11 @@ void AMDGPUTargetCodeGenInfo::setTargetAttributes( return; const FunctionDecl *FD = dyn_cast_or_null<FunctionDecl>(D); - if (FD) + if (FD) { setFunctionDeclAttributes(FD, F, M); - + if (FD->hasAttr<DeviceKernelAttr>() && !M.getLangOpts().OpenCL) + F->setCallingConv(getDeviceKernelCallingConv()); + } if (!getABIInfo().getCodeGenOpts().EmitIEEENaNCompliantInsts) F->addFnAttr("amdgpu-ieee", "false"); } @@ -508,6 +510,10 @@ AMDGPUTargetCodeGenInfo::getLLVMSyncScopeID(const LangOptions &LangOpts, case SyncScope::WavefrontScope: Name = "wavefront"; break; + case SyncScope::HIPCluster: + case SyncScope::ClusterScope: + Name = "cluster"; + break; case SyncScope::HIPWorkgroup: case SyncScope::OpenCLWorkGroup: case SyncScope::WorkgroupScope: @@ -654,7 +660,7 @@ llvm::Value *AMDGPUTargetCodeGenInfo::createEnqueuedBlockKernel( // kernel address (only the kernel descriptor). auto *F = llvm::Function::Create(FT, llvm::GlobalValue::InternalLinkage, Name, &Mod); - F->setCallingConv(llvm::CallingConv::AMDGPU_KERNEL); + F->setCallingConv(getDeviceKernelCallingConv()); llvm::AttrBuilder KernelAttrs(C); // FIXME: The invoke isn't applying the right attributes either |