[all-commits] [llvm/llvm-project] 9e574a: DAG: Fix expansion of bf16 sourced extloads

[Matt Arsenault via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 9e574a3936881d4f4845f0e867380b290571d888
      https://github.com/llvm/llvm-project/commit/9e574a3936881d4f4845f0e867380b290571d888
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2023-12-20 (Wed, 20 Dec 2023)

  Changed paths:
    M llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
    M llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp
    M llvm/test/CodeGen/AMDGPU/bf16.ll

  Log Message:
  -----------
  DAG: Fix expansion of bf16 sourced extloads

Also fix assorted vector extload failures for AMDGPU.


  Commit: b01adc6bed7e5b924dd8a097be0aa893f4823905
      https://github.com/llvm/llvm-project/commit/b01adc6bed7e5b924dd8a097be0aa893f4823905
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2023-12-20 (Wed, 20 Dec 2023)

  Changed paths:
    M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.ll
    M llvm/test/CodeGen/AMDGPU/bf16.ll
    M llvm/test/CodeGen/AMDGPU/function-args-inreg.ll
    M llvm/test/CodeGen/AMDGPU/function-args.ll
    M llvm/test/CodeGen/AMDGPU/function-returns.ll
    M llvm/test/CodeGen/AMDGPU/gfx-callable-argument-types.ll
    M llvm/test/CodeGen/AMDGPU/global-atomics-fp.ll
    A llvm/test/CodeGen/AMDGPU/llvm.is.fpclass.bf16.ll
    M llvm/test/CodeGen/AMDGPU/local-atomics-fp.ll
    M llvm/test/CodeGen/AMDGPU/private-memory-atomics.ll
    M llvm/test/CodeGen/AMDGPU/select-undef.ll

  Log Message:
  -----------
  AMDGPU: Strengthen some bfloat tests

Fix bitcast test, which was splitting apart phis intended to force
bitcasts that survive all the way to selection.

Disable the amdgpu-codegenprepare phi splitting, which defeats the technique
of using a phi to ensure a bitcast reaches all the way to selection. Also
add a variety of bfloat tests. These probably need revisiting to avoid the
cast folding into argument loads. Also round out set of bfloat bitcast and
ABI tests.

Add codegen tests for more bf16 operations The promotion of these works
contrary to the comment.


Compare: https://github.com/llvm/llvm-project/compare/701f64790520...b01adc6bed7e