https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117100
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Ever confirmed|0 |1
Last reconfirmed| |2024-10-12
Status|UNCONFIRMED |NEW
--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
auto_vec<int_range_max> edge_range;
...
edge_range.safe_grow_cleared (EDGE_COUNT (gimple_bb (stmt)->succs),
true);
Hmmm, is this even valid to do? Seems like that might be broken ...
Someone who knows the range API better can tell if not running the constructor
on int_range_max and just having it be all 0s will be fine.
That might be the reason why bisecting to r13-7841-g70639014a69cf5 makes
sense. It might also be a latent bug on the trunk too now.
Anyways confirmed.
