https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96564

--- Comment #24 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Jeffrey A. Law from comment #23)
> What I was suggesting was to use the flow sensitive capabilities of DOM to
> track the fact that the pointers can't be NULL at the comparison point. 
> Then use an updated API into the alias oracle that would essentially ask
> "can pointers A and B be the same, ignoring NULL".  Or something along those
> lines.

I would assume ranger could be hooked up that way really. Since it tracks
nullness now too. I thought Andrew MacLeod was working on something for GCC 17
to prange to do more inline with this. But maybe I am wrong.

Reply via email to