[Bug ipa/123459] [16 Regression] speculative devirtualization and always_inlining doesn't play nice together

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123459

--- Comment #9 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #8)
> (In reply to Orion Poplawski from comment #7)
> > Any idea when this might get addressed or what a good workaround would be? 
> > Thanks.
> 
> I think the best fix is to fix the source and remove always_inline when it
> comes to indirect function calls. That is openmpi should be fixed.

That is I don't see why __opal_attribute_always_inline__ is there in the first
place. The functions that have __opal_attribute_always_inline__  are big enough
that I am not sure always_inline would work anyways.