On Wed, Mar 15, 2023 at 01:07:56PM +0000, Richard Biener wrote:
> The following patch was bootstrapped and tested on
> x86_64-unknown-linux-gnu. It now doesn't regress anything in the
> testsuite but on its own it has the chance to (by luck maybe
> avoided previosuly due to the PHI compare bug) introduce extra
> false-positives. Fixing only the comparison causes false negatives
> in the testsuite - as said we rely on processing PHIs before the
> invalidation stmt.
>
> Together with the second patch I'm confident the overall experience
> will be better than what we have now.
>
> OK?
LGTM.
Jakub
