[PATCH] D148182: ValueTracking: Handle extractelement and extractvalue in computeKnownFPClass

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added a comment.

In D148182#4264236 <https://reviews.llvm.org/D148182#4264236>, @foad wrote:

> Seems reasonable. It would be nice to have some tests for vector sources where only some elements are known to be non-nan, but I guess they are hard to write.

I think they're unwritable as-is. This is testable after insertelement is handled in D148186 <https://reviews.llvm.org/D148186>


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D148182/new/

https://reviews.llvm.org/D148182