[all-commits] [llvm/llvm-project] 00aa59: AMDGPU: Fix illegal commute with frame index

[Matt Arsenault via All-commits]

  Branch: refs/heads/users/arsenm/amdgpu-fix-illegal-commute-frame-index
  Home:   https://github.com/llvm/llvm-project
  Commit: 00aa59ff350963bf30c2d3716a6d7855576b10ae
      https://github.com/llvm/llvm-project/commit/00aa59ff350963bf30c2d3716a6d7855576b10ae
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2024-10-31 (Thu, 31 Oct 2024)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
    A llvm/test/CodeGen/AMDGPU/commute-frame-index-operand.mir
    M llvm/test/CodeGen/AMDGPU/eliminate-frame-index-v-add-u32.mir

  Log Message:
  -----------
  AMDGPU: Fix illegal commute with frame index

In ca409892c5396fa3fbb8ea4dbf53d0e952f36d09, frame indexes started
being treated more like registers, rather than immediates. Update
the commute logic to avoid failing the verifier by moving illegal
SGPR operands in place of a frame index.



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications