[PATCH] Possible threads race condition in src/alarm.c

Andy Dougherty doughera at lafayette.edu
Fri Aug 31 14:15:39 UTC 2012


␟On Wed, 29 Aug 2012, Andrew Whitworth wrote:

> On Wed, Aug 29, 2012 at 4:50 PM, Reini Urban <rurban at x-ray.at> wrote:
> > I verified that it fixes the sleep deadlock, but with
> > the deadlock fix there come 3 more races detected by tsan.
> > (See http://blogs.perl.org/users/rurban/2012/08/threadsanitizer.html)
> >
> > Before:
> > $ tsan --announce-threads ./parrot t/pmc/task.t
> > 1..6
> > ok 1 - initialized
> > ok 2 task1 ran
> > ok 3 task2 ran
> > ok 4 sub1 ran
> > .. Deadlock in sleep cond_wait.
> >
> > After:
> > ...
> 
> Do those deadlocks cause any test failures? Are they serious concerns?
> I think we're getting pretty darn close to 100% here, I don't want to
> rush a merge until we're ready.

The tests pass for me, but I don't know how thorough they are.  There are
also examples, but most of them simply die.  I don't know what they are
supposed to do, so I don't know if that is expected or not.  I haven't 
tested the branch with rakudo.

Of course, it may still be worlds better than what's currently in there.

-- 
    Andy Dougherty		doughera at lafayette.edu


More information about the parrot-dev mailing list