[webkit-reviews] review denied: [Bug 184372] Executing known edge types may reveal a contradiction causing us to emit an exit at a node that is not allowed to exit : [Attachment 337401] patch

[bugzilla-daemon at webkit.org]

Filip Pizlo <fpizlo at apple.com> has denied Saam Barati <sbarati at apple.com>'s
request for review:
Bug 184372: Executing known edge types may reveal a contradiction causing us to
emit an exit at a node that is not allowed to exit
https://bugs.webkit.org/show_bug.cgi?id=184372

Attachment 337401: patch

https://bugs.webkit.org/attachment.cgi?id=337401&action=review




--- Comment #5 from Filip Pizlo <fpizlo at apple.com> ---
Comment on attachment 337401
  --> https://bugs.webkit.org/attachment.cgi?id=337401
patch

Isn’t the bug here that the SetLocal would try to emit a check for a known
edge?

It feels shady that you’re handling only the case of AI state becoming invalid.
There’s a sauntle version of this bug where AI state isn’t invalid: like a cell
check that wasn’t visible to AI. 

Therefore the correct fix is to make lowCell not emit a check for KnownCell.
Similarly other edge uses should not emit checks for known edges, regardless of
AI state. 

A good snort check here is that we want compilation to succeed even if AI
thinks things are top. We have a lot of quirks like this that prevent that
right now but we should fix them.