[PATCH] D151122: [GlobalISel] Check if ShiftAmt is greater then size of operand

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added inline comments.


================
Comment at: llvm/lib/CodeGen/GlobalISel/CombinerHelper.cpp:1726
   unsigned MinLeadingZeros = KB->getKnownZeroes(ExtSrc).countl_one();
-  return MinLeadingZeros >= ShiftAmt;
+  unsigned SrcTySize = MRI.getType(ExtSrc).getSizeInBits();
+  return MinLeadingZeros >= ShiftAmt && ShiftAmt < SrcTySize;
----------------
Broken for vectors, getScalarSizeInBits


================
Comment at: llvm/test/CodeGen/AMDGPU/GlobalISel/combine-shl-from-extend-narrow.postlegal.mir:332
+---
+name: do_not_shl_s32_by_16_from_zext_s16
+tracksRegLiveness: true
----------------
Add a vector version. Also would be good to have one that couldn't be trivially constant folded 


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D151122/new/

https://reviews.llvm.org/D151122