[all-commits] [llvm/llvm-project] 5e79ae: DAG: Fix vector_shuffle -> splat fold defining und...

[Matt Arsenault via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 5e79ae60a67726805fcc27081f67c41cbd8a1e4e
      https://github.com/llvm/llvm-project/commit/5e79ae60a67726805fcc27081f67c41cbd8a1e4e
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2025-01-21 (Tue, 21 Jan 2025)

  Changed paths:
    M llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
    M llvm/test/CodeGen/AMDGPU/shufflevector.v4i64.v3i64.ll
    M llvm/test/CodeGen/AMDGPU/shufflevector.v4p0.v3p0.ll
    M llvm/test/CodeGen/PowerPC/vector-reduce-fadd.ll
    M llvm/test/CodeGen/WebAssembly/simd.ll
    M llvm/test/CodeGen/X86/urem-seteq-illegal-types.ll
    M llvm/test/CodeGen/X86/vec_smulo.ll
    M llvm/test/CodeGen/X86/vec_umulo.ll
    M llvm/test/CodeGen/X86/widen_shuffle-1.ll

  Log Message:
  -----------
  DAG: Fix vector_shuffle -> splat fold defining undef lanes (#123596)

For shuffle vector splats with undef lanes in the mask,
this was introducing real values. Filter out build_vector
results based on the undef elements in the mask.

This avoids AMDGPU test regressions in a future change.

test/CodeGen/X86/urem-seteq-illegal-types.ll looks worse
but I didn't investigate.



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