maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[CUDA][HIP][NFC] Do not add `-fgpu-rdc` flag twice 

Summary:
We currently add the `-fgpu-rdc` flag twice. Once unconditionally for
both the host and device phases of compilation, and a second time only
for the host. When we moved to an unconditional addition of this flag it
the old one was most likely not removed. This patch simply removes the
redundant flag and changes no functionality.
This commit is contained in:
Joseph Huber 2022-07-13 15:56:15 -04:00
parent e32864b605
commit 6abaa8e210
2 changed files with 0 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
clang/lib/Driver/ToolChains/Clang.cpp
View File

@ -7011,8 +7011,6 @@ void Clang::ConstructJob(Compilation &C, const JobAction &JA,
if ((IsCuda || IsHIP) && CudaDeviceInput) {
CmdArgs.push_back("-fcuda-include-gpubinary");
CmdArgs.push_back(CudaDeviceInput->getFilename());
if (Args.hasFlag(options::OPT_fgpu_rdc, options::OPT_fno_gpu_rdc, false))
CmdArgs.push_back("-fgpu-rdc");
}
if (IsCuda) {

2
clang/test/Driver/cuda-external-tools.cu
View File

@ -160,7 +160,5 @@
// ARCH64-SAME: "-triple" "x86_64-
// ARCH32-SAME: "-triple" "i386-
// CHECK-SAME: "-fcuda-include-gpubinary" "[[FATBINARY]]"
// RDC-SAME: "-fgpu-rdc"
// CHECK-NOT: "-fgpu-rdc"
// CHK-PTXAS-VERBOSE: ptxas{{.*}}" "-v"