[PATCH] D72330: [ARM][MVE] Enable masked gathers from base + vector of offsets

[Anna Welker via Phabricator via llvm-commits]

anwel marked 3 inline comments as done.
anwel added inline comments.


================
Comment at: llvm/lib/Target/ARM/MVEGatherScatterLowering.cpp:197
   Value *Load = nullptr;
-  // Look through bitcast instruction if #elements is the same
-  if (auto *BitCast = dyn_cast<BitCastInst>(Ptr)) {
-    Type *BCTy = BitCast->getType();
-    Type *BCSrcTy = BitCast->getOperand(0)->getType();
-    if (BCTy->getVectorNumElements() == BCSrcTy->getVectorNumElements()) {
-      LLVM_DEBUG(dbgs() << "masked gathers: looking through bitcast\n");
-      Ptr = BitCast->getOperand(0);
-    }
+  if (!createMaskedGatherOffset(Load, Builder))
+    if (!createMaskedGatherBase(Load, Builder))
----------------
dmgreen wrote:
> These could be `Value *MVEGatherScatterLowering::tryCreateMaskedGatherOffset` and return nullptr on failure. It could then look something like:
> ```
>   Value *Load = tryCreateMaskedGatherOffset(Builder, Ty, Ptr, Mask);
>   if (!Load)
>     Load = tryCreateMaskedGatherBase(Builder, Ty, Ptr, Mask);
>   if (!Load)
>     return false;
> ```
Yes, that makes one less argument to the function


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D72330/new/

https://reviews.llvm.org/D72330