[all-commits] [llvm/llvm-project] 666333: [X86][AVX] canonicalizeLaneShuffleWithRepeatedOps ...

Simon Pilgrim via All-commits all-commits at lists.llvm.org
Thu Jan 28 04:12:38 PST 2021


  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 6663330bc8c84a75ea092272297b557bfc310380
      https://github.com/llvm/llvm-project/commit/6663330bc8c84a75ea092272297b557bfc310380
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2021-01-28 (Thu, 28 Jan 2021)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp
    M llvm/test/CodeGen/X86/vector-shuffle-combining-avx.ll

  Log Message:
  -----------
  [X86][AVX] canonicalizeLaneShuffleWithRepeatedOps - don't merge VPERMILPD ops with different low/high masks.

Unlike VPERMILPS, VPERMILPD can have non-repeating masks in each 128-bit subvector, we weren't accounting for this when folding vperm2f128(vpermilpd(x,c),vpermilpd(y,c)) -> vpermilpd(vperm2f128(x,y),c).

I'm intending to add support for this but wanted to get a minimal fix in first for merging into 12.xx.

Fixes PR48908




More information about the All-commits mailing list