[PATCH] D54713: [SCEV] Guard movement of insertion point for loop-invariants
Warren Ristow via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Wed Nov 28 22:46:28 PST 2018
wristow added a comment.
> I'm not sure whether you use a custom pipeline or not. But my point was that code came to vectorization pass seems strange because trivial loop unswitch optimization (provided by LoopUnswitch pass) could hoist the invariant check. By some reason it did not happen.
To be clear, I'm not using a custom pipeline. I just happen to have a customer who reported a runtime crash because of an integer-div-by-0 (compiled simply as `clang -O2`), and that led me to PR30806. Which led me to looking into the vectorizer,.
All that said, your comment about it should have been caught by the LoopUnswitch pass, now makes me more optimistic about tackling that. Still not committing that I'll get to it soon, but it seems less daunting now. :)
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D54713/new/
https://reviews.llvm.org/D54713
More information about the llvm-commits
mailing list