[PATCH] D74945: AMDGPU/GlobalISel: Fix xnor matching

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added a comment.

In D74945#1885979 <https://reviews.llvm.org/D74945#1885979>, @foad wrote:

> What was the bug? Could you pre-commit xnor.ll?


The patterns successfully imported, but we tried manually selecting the root operation before the generated matchers. We still need the manual selector because the patterns only catch i32/i64, and missing the legal v2i16/v2i32 etc. cases. I want to eventually add a vt32/vt64 type ignoring feature to tablegen so a single pattern matches any type of the given size


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D74945/new/

https://reviews.llvm.org/D74945