https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86056
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Component|target |middle-end
--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
I don't see why you think atomic_thread_fence should prevent this kind of
optimization here as no other thread can "see" the stores that would be doing.
The only thing I see missing is the merging of the __atomic_thread_fence
