[all-commits] [llvm/llvm-project] 9bd108: AMDGPU: Undo atomicrmw add/sub/xor 0 -> atomicrmw ...

[Matt Arsenault via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 9bd10853e505b61f5fe2e3b3592c76787a06aa31
      https://github.com/llvm/llvm-project/commit/9bd10853e505b61f5fe2e3b3592c76787a06aa31
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2024-04-13 (Sat, 13 Apr 2024)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIISelLowering.cpp
    M llvm/test/CodeGen/AMDGPU/global_atomics_i32_system.ll
    A llvm/test/Transforms/AtomicExpand/AMDGPU/expand-atomicrmw-integer-ops-0-to-add-0.ll

  Log Message:
  -----------
  AMDGPU: Undo atomicrmw add/sub/xor 0 -> atomicrmw or canonicalization (#87533)

InstCombine transforms add of 0 to or of 0. For system atomics, this is
problematic because while PCIe supports add, it does not support the
other operations. Undo this for system scope atomics.



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