> James Morris writes:
> > BUG: seems like it, if they can point to the same thing. ERROR: 1 thread deadlock.
> > <struct in_device.lock (<local>:0)>-><struct ip_mc_list.lock (<local>:0)> occurred 5 times
> > <struct ip_mc_list.lock (<local>:0)>-><struct in_device.lock (<local>:0)> occurred 5 times
>
> See below.
>
> > BUG? very hard to follow, but interesting if a real bug. unfortunately,
> > could also be a false positive because of
> > 1. infeasible callchain path or
> >
> > 2. the various in_dev and im pointers never actually point to
> > the same object.
> >
> > requires three threads:
> > thread 1: acquires im->lock then tries to get inetdev_lock
> > thread 2: acquires inetdev_lock and tries to get in_dev->lock.
> > thread 3: acquires in_dev->lock and tries to get im->lock.
> >
> > ERROR: 2 thread deadlock.
> > <struct ip_mc_list.lock (<local>:0)>-><&inetdev_lock> occurred 5 times
> > <&inetdev_lock>-><struct ip_mc_list.lock (<local>:0)> occurred 4 times
>
> These are indeed potential deadlock cases, caused by holding im->lock for
> too long, now fixed by Alexey (in 2.5 bk at least).
great! Thanks very much for the feedback. I'd given up on anyone
looking at these.
If people are interested, I can release more deadlock bugs pretty easily.
Dawson
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Mar 15 2003 - 22:00:29 EST