https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118232
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |jakub at gcc dot gnu.org
--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #4)
> It's going to be difficult to prevent. At some point I considered making
> FRE dumb down addresses it CSEs to &base + offset. The late diagnostics are
> too late here.
Could be also just a flag set on the ADDR_EXPR or the references in its
argument which could tell the late warning passes and get_range_strlen_tree
(and whatever other code misuses that) to punt, don't assume anything about
those.
Note, it would be useful to do this dumbing down only if the CSE origin and dup
aren't the same type wise etc.
