[PATCH] D24365: [PR29121] Don't fold if it would produce atomic vector loads or stores

[Steven Wu via llvm-commits]

steven_wu added a comment.

Thanks for cleaning up my proposed fix and creating the review. I think this is the correct fix in short term before we can legalize the vector atomic in instruction selection for all backend. LGTM.


https://reviews.llvm.org/D24365