[all-commits] [llvm/llvm-project] 6d4555: Mips/GlobalISel: Add stub post-legalizer combiner

[Matt Arsenault via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 6d45558c1a05d4ae925bdf628b8cefa947a589f3
      https://github.com/llvm/llvm-project/commit/6d45558c1a05d4ae925bdf628b8cefa947a589f3
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2022-04-14 (Thu, 14 Apr 2022)

  Changed paths:
    M llvm/lib/Target/Mips/CMakeLists.txt
    M llvm/lib/Target/Mips/Mips.h
    M llvm/lib/Target/Mips/Mips.td
    A llvm/lib/Target/Mips/MipsCombine.td
    A llvm/lib/Target/Mips/MipsPostLegalizerCombiner.cpp
    M llvm/lib/Target/Mips/MipsTargetMachine.cpp

  Log Message:
  -----------
  Mips/GlobalISel: Add stub post-legalizer combiner

This enables no combines, just adds the boilerplate for the new pass.


  Commit: a1303b23c9de6ef6d667aa923ec266ca4a0334e7
      https://github.com/llvm/llvm-project/commit/a1303b23c9de6ef6d667aa923ec266ca4a0334e7
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2022-04-14 (Thu, 14 Apr 2022)

  Changed paths:
    M clang/lib/Basic/Targets/AMDGPU.cpp
    M clang/test/Driver/amdgpu-macros.cl

  Log Message:
  -----------
  clang/AMDGPU: Define macro for -munsafe-fp-atomics

The HIP headers want to use this to swap the implementation of the
function, rather than relying on backend expansion of the generic
atomic instruction.

Fixes: SWDEV-332998


Compare: https://github.com/llvm/llvm-project/compare/2d9b7fdaa379...a1303b23c9de