diff --git a/llvm/lib/Target/NVPTX/NVVMReflect.cpp b/llvm/lib/Target/NVPTX/NVVMReflect.cpp --- a/llvm/lib/Target/NVPTX/NVVMReflect.cpp +++ b/llvm/lib/Target/NVPTX/NVVMReflect.cpp @@ -40,6 +40,7 @@ #include #include #define NVVM_REFLECT_FUNCTION "__nvvm_reflect" +#define NVVM_REFLECT_OCL_FUNCTION "__nvvm_reflect_ocl" using namespace llvm; @@ -78,7 +79,8 @@ if (!NVVMReflectEnabled) return false; - if (F.getName() == NVVM_REFLECT_FUNCTION) { + if (F.getName() == NVVM_REFLECT_FUNCTION || + F.getName() == NVVM_REFLECT_OCL_FUNCTION) { assert(F.isDeclaration() && "_reflect function should not have a body"); assert(F.getReturnType()->isIntegerTy() && "_reflect's return type should be integer"); @@ -119,6 +121,7 @@ continue; Function *Callee = Call->getCalledFunction(); if (!Callee || (Callee->getName() != NVVM_REFLECT_FUNCTION && + Callee->getName() != NVVM_REFLECT_OCL_FUNCTION && Callee->getIntrinsicID() != Intrinsic::nvvm_reflect)) continue; diff --git a/llvm/test/CodeGen/NVPTX/nvvm-reflect-ocl.ll b/llvm/test/CodeGen/NVPTX/nvvm-reflect-ocl.ll new file mode 100644 --- /dev/null +++ b/llvm/test/CodeGen/NVPTX/nvvm-reflect-ocl.ll @@ -0,0 +1,20 @@ +; Verify that __nvvm_reflect_ocl() is replaced with an appropriate value +; +; RUN: opt %s -S -passes='default' -mtriple=nvptx64 \ +; RUN: | FileCheck %s --check-prefixes=COMMON,SM20 +; RUN: opt %s -S -passes='default' -mtriple=nvptx64 -mcpu=sm_35 \ +; RUN: | FileCheck %s --check-prefixes=COMMON,SM35 + +@"$str" = private addrspace(4) constant [12 x i8] c"__CUDA_ARCH\00" + +declare i32 @__nvvm_reflect_ocl(ptr addrspace(4) noundef) + +; COMMON-LABEL: @foo +define i32 @foo(float %a, float %b) { +; COMMON-NOT: call i32 @__nvvm_reflect_ocl + %reflect = tail call i32 @__nvvm_reflect_ocl(ptr addrspace(4) noundef getelementptr inbounds ([12 x i8], [12 x i8] addrspace(4)* @"$str", i64 0, i64 0)) +; SM20: ret i32 200 +; SM35: ret i32 350 + ret i32 %reflect +} +