1// REQUIRES: amdgpu-registered-target 2// RUN: %clang_cc1 -x ir -fcuda-is-device -triple amdgcn-amd-amdhsa -emit-llvm-bc -disable-llvm-passes -o %t.bc %S/Inputs/device-lib-code.ll 3// RUN: %clang_cc1 -x hip -fcuda-is-device -triple amdgcn-amd-amdhsa -mlink-builtin-bitcode %t.bc -emit-llvm -disable-llvm-passes -o - %s | FileCheck %s 4 5#include "Inputs/cuda.h" 6 7extern "C" __device__ float __ocml_fma_f32(float x, float y, float z); 8 9__device__ float foo(float x) { 10 return __ocml_fma_f32(x, x, x); 11} 12 13// CHECK: {{^}}define{{.*}} @__ocml_fma_f32{{.*}} [[ATTR1:#[0-9]+]] 14// CHECK: {{^}}declare{{.*}} @llvm.fma.f32{{.*}} [[ATTR2:#[0-9]+]] 15// CHECK: attributes [[ATTR1]] = { convergent 16// CHECK: attributes [[ATTR2]] = { 17// CHECK-NOT: convergent 18// CHECK: } 19