[all-commits] [llvm/llvm-project] 370aa2: InlineSpiller: Don't fold spills into undef reads

[Matt Arsenault via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 370aa2f88ffabae5831bbc350c03d7dcc757580b
      https://github.com/llvm/llvm-project/commit/370aa2f88ffabae5831bbc350c03d7dcc757580b
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2022-06-22 (Wed, 22 Jun 2022)

  Changed paths:
    M llvm/lib/CodeGen/InlineSpiller.cpp
    A llvm/test/CodeGen/AMDGPU/fold-restore-undef-use.mir

  Log Message:
  -----------
  InlineSpiller: Don't fold spills into undef reads

This was producing a load into a dead register which was a verifier
error.


  Commit: b03d902b61099cc5bfe7d17bbca59fccc1fd064a
      https://github.com/llvm/llvm-project/commit/b03d902b61099cc5bfe7d17bbca59fccc1fd064a
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2022-06-22 (Wed, 22 Jun 2022)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIOptimizeExecMaskingPreRA.cpp
    A llvm/test/CodeGen/AMDGPU/opt-exec-masking-pre-ra-update-liveness-wave32.mir
    A llvm/test/CodeGen/AMDGPU/opt-exec-masking-pre-ra-update-liveness.mir

  Log Message:
  -----------
  AMDGPU: Fix invalid liveness after si-optimize-exec-masking-pre-ra

This was leaving behind a use at the deleted instruction which the
verifier would fail during allocation.


Compare: https://github.com/llvm/llvm-project/compare/a35141d39501...b03d902b6109