https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121112
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords|documentation |wrong-code
Ever confirmed|0 |1
Last reconfirmed| |2025-07-16
Summary|Issue about alias function |alias and ifunc attribute
|call under high level |together incorrectly causes
|optimization |inlining of the resolver
Status|UNCONFIRMED |NEW
--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Confirmed.
The resolver does NOT have any side effects.
In this case we have
bar_alias is an alias of bar
and bar is an ifunc whos resolver is bar_ifunc.
cgraph decided that bar_alias is an alias to bar_ifunc due to the way ifunc are
represented into the cgraph but that messes up things when it comes to
inlining.
