[PATCH] D106053: [CodeGen] Remove pending AssertZext AssertSext in promoting FP_TO_INT

[Craig Topper via Phabricator via llvm-commits]

craig.topper added a comment.

Here's what I think is the issue. After type legalization we have a bitcast from <4 x i32> to <16 x i8>. When computeKnownBits sees a bitcast from a wider element type to a smaller element type, it converts DemandedElts into DemandedBits to try to get an answer from a portion of the wider input element elements.

How does poison interact with a bitcast from a wide element type to a narrower element type? Is what computeKnownBits does for this poison safe?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D106053/new/

https://reviews.llvm.org/D106053