[PATCH] D98422: [Alias] Add a ah-hoc pattern with two PHI for isKnownNonEqual

JinGu Kang via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Thu Mar 18 13:56:08 PDT 2021


jaykang10 added a comment.

In D98422#2635678 <https://reviews.llvm.org/D98422#2635678>, @nikic wrote:

> I think the main problem here is that the review tries doing too much. Yes, the BasicAA test case can be handled as a combination of simple phi handling plus some additional mul handling. But from the code comments and discussion the intention is to also handle some kind of additional case involving add recurrences.
>
> It would be great if this patch can be cut down to just basic phi-phi handling, with dedicated ValueTracking tests -- even if that does not cover the motivational case in isolation.
>
> As @reames has mentioned before, the code should also only recurse into one pair of operands, and require trivially non-equal constants for the rest. At least we have been following this rule in the rest of the isKnownNonEqual code.

Ok, I agree with your opinion. Let me cut down this patch to the basic case. Thanks for suggestion @nikic @lebedev.ri


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D98422/new/

https://reviews.llvm.org/D98422



More information about the llvm-commits mailing list