[Devel] Re: Re: Hang with fair cgroup scheduler (reproducer is attached.)

Dmitry Adamushko dmitry.adamushko at gmail.com
Fri Dec 14 08:25:39 PST 2007


On 14/12/2007, Dhaval Giani <dhaval at linux.vnet.ibm.com> > >
> Actually no, its another bug. Thanks for the program!
>

Humm... this crash is very likely to be caused by the same bug. It
just reveals itself in a different place, but effectivelly the pattern
looks similar. Anyway, the rb-tree gets corrupted... and for both
cases, at the very least the 'current' must be within the tree.
I think, if you repeat your test a number of times, you'll likely get
the very same crash as was reported by Kame.

ia64 does define __ARCH_WANT_UNLOCKED_CTXSW (I checked against
2.6.23.1 that I have at hand)

x86 -- not (it's not reproducible there, right?)

so for ia64 task_running() makes use of 'p->oncpu' to determine
whether a given task is currently running
(as opposed to 'rq->curr == p' otherwise)...

But at first glance, it looks like there shouldn't be situations
leading to some sort of de-synchronization in determining the real
'current'.

Will look at it closer.


>
> --
> regards,
> Dhaval
>

-- 
Best regards,
Dmitry Adamushko
_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers




More information about the Devel mailing list