[PATCH] D148843: [X86 isel] Fix permute mask calculation in lowerShuffleAsUNPCKAndPermute

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon added inline comments.


================
Comment at: llvm/lib/Target/X86/X86ISelLowering.cpp:13343
+      PermuteMask[Elt] = BaseMaskElt + 1;
   }
 
----------------
Is the OpsSwapped logic necessary?
```
  for (int Elt = 0; Elt != NumElts; ++Elt) {
    int M = Mask[Elt];
    if (M < 0)
      continue;
    int NormM = M;
    if (NumElts <= M)
      NormM -= NumElts;
    bool IsFirstOp = M < NumElts;
    int BaseMaskElt =
        NumLaneElts * (NormM / NumLaneElts) + (2 * (NormM % NumHalfLaneElts));
    if ((IsFirstOp && V1 == Ops[0]) || (!IsFirstOp && V2 == Ops[0]))
      PermuteMask[Elt] = BaseMaskElt;
    else if if ((IsFirstOp && V1 == Ops[1]) || (!IsFirstOp && V2 == Ops[1]))
      PermuteMask[Elt] = BaseMaskElt + 1;
  }
```
I think there probably needs to be an assertion there as well someplace.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D148843/new/

https://reviews.llvm.org/D148843