This is an archive of the discontinued LLVM Phabricator instance.

[AMDGPU][MC] Fix for Bug 28211 + LIT tests
ClosedPublic

Authored by dp on Apr 5 2017, 6:36 AM.

Details

Summary
  • corrected DS_GWS_* opcodes (see VI_Shader_Programming#16.pdf for detailed description)
    • address operand is not used
    • several opcodes have data operand
    • all opcodes have offset modifier
  • DS_AND_SRC2_B32: corrected typo in mnemo
  • DS_WRAP_RTN_F32 replaced with DS_WRAP_RTN_B32
  • added CI/VI opcodes:
    • DS_CONDXCHG32_RTN_B64
    • DS_GWS_SEMA_RELEASE_ALL
  • added VI opcodes:
    • DS_CONSUME
    • DS_APPEND
    • DS_ORDERED_COUNT

https://bugs.llvm.org//show_bug.cgi?id=28211

Diff Detail

Repository
rL LLVM

Event Timeline

dp created this revision.Apr 5 2017, 6:36 AM
vpykhtin accepted this revision.Apr 5 2017, 7:02 AM

LGTM.

This revision is now accepted and ready to land.Apr 5 2017, 7:02 AM
This revision was automatically updated to reflect the committed changes.
jvesely added a subscriber: jvesely.Apr 7 2017, 7:50 AM

Hi,

can you please use more descriptive short messages? "Fix for bug XYZ" makes viewing commit history unusable.
Moreover, VI_Shader_programming#16.pdf does not appear to be publicly available, so the reference is useless.

Jan

dp added a comment.Apr 10 2017, 11:20 AM

I'll provide more detailed description in the future.

Bug description is available here:
https://bugs.llvm.org//show_bug.cgi?id=28211

I'd have referred a publicly available document, but I found no relevant information elsewhere.

In D31707#722640, @dp wrote:

I'll provide more detailed description in the future.

Bug description is available here:
https://bugs.llvm.org//show_bug.cgi?id=28211

thank you. I can dig through review and bugzilla when looking for details.
it's the initial overview to determine interesting commits (potential fixes/regressions) that suffers

AMDGPU: Actually write nops for writeNopData
[AMDGPU] Unroll more to eliminate phis and conditions
[AMDGPU][MC] Fix for Bug 28211 + LIT tests
[AMDGPU] Move SiShrinkInstruction and SDWAPeephole to SSAOptimization passes
AMDGPU/GFX9: Fix shared and private aperture queries

thanks again,
Jan

dp added a comment.Apr 11 2017, 12:30 AM

Thanks, I got the idea. Sorry for confusion.