[Bug nptl/13165] pthread_cond_wait() can consume a signal that was sent before it started waiting

["mihaylov.mihail at gmail dot com" <sourceware-bugzilla at sourceware dot org>]

http://sourceware.org/bugzilla/show_bug.cgi?id=13165

--- Comment #33 from Mihail Mihaylov <mihaylov.mihail at gmail dot com> 2012-10-18 06:26:22 UTC ---
(In reply to comment #32)
> (In reply to comment #28)
> ....
> ....
> > Anyway, this whole dispute has been reduced to the question of which threads
> > are eligible for wakeup, so I've taken the liberty to post a clarification
> > request to the Austin Group, asking them to add explicit text explaining which
> > threads should be considered blocked with respect to a pthread_cond_signal()
> > call. The clarification request is at
> > http://austingroupbugs.net/view.php?id=609. Torvald, please correct me if have
> > inadvertently misrepresented your position.
> 
> Thanks.  I have added a note there that summarizes the interpretation that the
> current implementation is based on.  Let's see what they think about this
> issue.

The Austin Group have reached an official position. They have decided to make
changes to some of the texts related to condition variables. I believe that the
changes as they announced them yesterday invalidate glibc's interpretation of
the spec.

Let me point out that these changes do not add new requirements to the spec.
They just make explicit the requirements that were already suggested by the
spec.

In my opinion, at this point it is already clear that NPTL's implementation of
condition variables does not conform to the POSIX spec, therefore this bug is
actually really a bug. I hope that the NPTL team will acknowledge it as such.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.