[PATCH] D30270: [InstCombine] don't try SimplifyDemandedInstructionBits from add/sub because it's slow and unlikely to succeed

[Eli Friedman via Phabricator via llvm-commits]

efriedma accepted this revision.
efriedma added a comment.
This revision is now accepted and ready to land.

It looks like in the current version of SimplifyDemandedUseBits, simplifying the demanded bits on an Add with an all-ones mask is equivalent to calling computeKnownBits.  That explains why you didn't find any regressions. :)

Like you've said, I can't see any reason why we'd want to do this; LGTM.


https://reviews.llvm.org/D30270