[PATCH] D89076: AMDGPU: Update AMDHSA code object version handling

Konstantin Zhuravlyov via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Thu Oct 8 17:38:20 PDT 2020


kzhuravl marked 3 inline comments as done.
kzhuravl added inline comments.


================
Comment at: clang/docs/ClangCommandLineReference.rst:179
 
+.. option:: -dsym-dir<dir>
+
----------------
t-tye wrote:
> scott.linder wrote:
> > There seem to be a lot of unrelated changes here; does the patch need to be rebased? Or are you just cleaning up the whole file?
> > 
> > If the latter could these be in a separate diff?
> Many changes in this file are the result of earlier changes in the options.td file not updating this file with the build result. So suggest submitting a review for updating this file without the code version changes first. Then the changes to this file in this review will just be code object version related.
> 
> Otherwise I think all the changes in this review are related to cleaning up the handling of the existing V2 and V3 code object formats in preparation for submitting the V4 code object changes.
Re-generating the whole file, as I think it might have been stale.


================
Comment at: llvm/docs/AMDGPUUsage.rst:901
 
-Code Object V2 Note Records (-mattr=-code-object-v3)
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+Code Object V2 Note Records (--amdhsa-code-object-version=2)
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
----------------
t-tye wrote:
> Were there also cleanup in the ELF section that describes the e_flags and defines the usage of the new --amdhsa-code-object-version option?
I think doc updates will come in a separate patch as we discussed earlier today.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D89076/new/

https://reviews.llvm.org/D89076



More information about the llvm-commits mailing list