[PATCH] Fix a race condition in indirect call value profiling

Jack Howarth howarth@bromo.med.uc.edu
Tue Dec 30 17:48:00 GMT 2008


On Tue, Dec 30, 2008 at 11:55:37AM +0100, Richard Guenther wrote:
> On Tue, Dec 30, 2008 at 5:35 AM, Andrew Pinski <pinskia@gmail.com> wrote:
> > 2008/12/29 Seongbae Park 박성배 朴成培 <seongbae.park@gmail.com>:
> >> I was afraid something like this could happen on different targets.
> >> What's the appropriate flag or #ifdef guard ?
> >
> > It is only darwin where the problem shows up really.  There has been
> > some talk about how to fix darwin's libgcc.  But nothing much.  Can
> > this patch wait until 4.5?
> 
> Huh, weird Darwin again.  Yes, please back this out for now.
> 
> Thanks,
> Richard.
> 
> > Thanks,
> > Andrew Pinski
> >
  Shouldn't we open a PR for the general issue of new libgcc
symbols on darwin? It would seem that we will be forced to do
something like always link against both the versioned libgcc
and then the libgcc_s.1 to pick up these symbols. 
                Jack



More information about the Gcc-patches mailing list