[Mesa-dev] [Bug 98172] Concurrent call to glClientWaitSync results in segfault in one of the waiters.

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=98172

--- Comment #35 from Marek Olšák <maraeo at gmail.com> ---
(In reply to Michel Dänzer from comment #34)
> (In reply to Suzuki, Shinji from comment #31)
> > Why is this patch rejected?
> 
> Because it'll have to be rebased on top of the updated patch 1.
> 
> 
> > Now I'm revisiting your patch. Do we need to have mutual exclusion on
> >    screen->fence_reference(screen, &fence, so->fence);
> > and
> >       screen->fence_reference(screen, &so->fence, NULL);
> > ?
> 
> I'm not sure r600_fence_reference() is thread safe.

OK, again:

fence_reference is thread-safe with regard to the reference counter and fence
destruction. It doesn't, however, protect the "dst" pointer itself. Therefore:

Not thread safe (race condition on so->fence):
  screen->fence_reference(screen, &so->fence, NULL);

Always thread safe (if fence is a local variable):
  screen->fence_reference(screen, &fence, NULL);

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20161019/e4198e77/attachment.html>