[all-commits] [llvm/llvm-project] bf4dc4: [AMDGPU] Don't transform illegal intrinsics to V_I...

[Jay Foad via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: bf4dc4381e3023c110fd3efbc9caa5e280373fbb
      https://github.com/llvm/llvm-project/commit/bf4dc4381e3023c110fd3efbc9caa5e280373fbb
  Author: Jay Foad <jay.foad at amd.com>
  Date:   2023-04-19 (Wed, 19 Apr 2023)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIISelLowering.cpp
    M llvm/lib/Target/AMDGPU/SIISelLowering.h
    M llvm/lib/Target/AMDGPU/SIInstructions.td
    A llvm/test/CodeGen/AMDGPU/unsupported-atomics.ll
    A llvm/test/CodeGen/AMDGPU/unsupported-image-sample.ll
    R llvm/test/CodeGen/AMDGPU/v_illegal-atomics.ll
    R llvm/test/CodeGen/AMDGPU/v_illegal-image_sample.ll
    M llvm/test/MC/AMDGPU/v_illegal-atomics.s

  Log Message:
  -----------
  [AMDGPU] Don't transform illegal intrinsics to V_ILLEGAL

This reverts parts of D123693. The functionality of allowing unsupported
intrinsics to select has been superseded by D139000 "Remove function
with incompatible features".

Retain assembler/disassembler support for v_illegal on GFX10+ only,
where it is documented.

Differential Revision: https://reviews.llvm.org/D148127