Page MenuHomePhabricator

Pre-commit test for D131587
Needs ReviewPublic

Authored by piggynl on Aug 14 2022, 11:07 AM.

Details

Summary

Depends on D131863.

Diff Detail

Event Timeline

piggynl created this revision.Aug 14 2022, 11:07 AM
piggynl requested review of this revision.Aug 14 2022, 11:07 AM
piggynl added inline comments.Aug 14 2022, 11:16 AM
llvm/test/CodeGen/AMDGPU/branch-relax-spill.ll
2036

I have no knowledge of AMDGPU and don't know how to construct some branches that can share or can't share a same restore block, so please kindly let me know how to construct the test properly.

This is the first added function. I copied @spill() in this file and simply added another branch after the previous one. The two restore blocks .LBB2_5 and .LBB2_7 are identical.

3004

This is the second added function. I copied @spill_func() in this file and simply added another branch after the previous one. It seems the two branches don't share a same destination.

3413–3422

and I don't know why this sequence is forming. Is this intended?

arsenm added inline comments.Thu, Sep 15, 6:17 AM
llvm/test/CodeGen/AMDGPU/branch-relax-spill.ll
2

Why do you need to increase this? This just forces you to make the test functions bigger? If it's just for the new test, I'd rather split that into a separate file

3413–3422

Do you mean the copies to shift down by one? I'd guess this is somehow related to the asm sequence trying to use a bunch of reserved registers as if they were all allocatable, and with such high register usage you hit some poor allocation decision