[all-commits] [llvm/llvm-project] 36ef29: [AMDGPU] Fix hang caused by VS_CNT handling at cal...

[Jay Foad via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 36ef291d632e3c0ac109e8e1b19ae09d39ac5496
      https://github.com/llvm/llvm-project/commit/36ef291d632e3c0ac109e8e1b19ae09d39ac5496
  Author: Jay Foad <jay.foad at amd.com>
  Date:   2024-01-17 (Wed, 17 Jan 2024)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
    A llvm/test/CodeGen/AMDGPU/insert-waitcnts-hang.mir

  Log Message:
  -----------
  [AMDGPU] Fix hang caused by VS_CNT handling at calls (#78318)

Fix a potential hang introduced by #77439 and #77935. This line:

  setScoreUB(VS_CNT, getScoreLB(VS_CNT) + getWaitCountMax(VS_CNT));

could potentialy set UB lower than it was before, which confused
SIInsertWaitcnts's fixed point algorithm.

This was only triggered a STORE instruction with an implicit-def, which
seems odd but apparently happens for some spills.