replica
/
hanchenye-llvm-project
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

AMDGPU/NFC: Use already available explicit kernarg 

size instead of calculating it again when filling
out the metadata.

llvm-svn: 336825
This commit is contained in:
Konstantin Zhuravlyov 2018-07-11 17:27:17 +00:00
parent c094d23f6f
commit bde59e9989
1 changed files with 2 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
llvm/lib/Target/AMDGPU/AMDGPUHSAMetadataStreamer.cpp
View File

@ -212,7 +212,8 @@ Kernel::CodeProps::Metadata MetadataStreamer::getHSACodeProps(
if (F.getCallingConv() != CallingConv::AMDGPU_KERNEL)
return HSACodeProps;
HSACodeProps.mKernargSegmentSize = STM.getKernArgSegmentSize(F);
HSACodeProps.mKernargSegmentSize =
STM.getKernArgSegmentSize(F, MFI.getExplicitKernArgSize());
HSACodeProps.mGroupSegmentFixedSize = ProgramInfo.LDSSize;
HSACodeProps.mPrivateSegmentFixedSize = ProgramInfo.ScratchSize;
HSACodeProps.mKernargSegmentAlign =