lkml.org 
[lkml]   [2011]   [Sep]   [27]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 1/2] oom: do not live lock on frozen tasks
On Mon 26-09-11 18:03:26, David Rientjes wrote:
> On Mon, 26 Sep 2011, Rafael J. Wysocki wrote:
>
> > > Konstantin Khlebnikov has reported (https://lkml.org/lkml/2011/8/23/45)
> > > that OOM can end up in a live lock if select_bad_process picks up a frozen
> > > task.
> > > Unfortunately we cannot mark such processes as unkillable to ignore them
> > > because we could panic the system even though there is a chance that
> > > somebody could thaw the process so we can make a forward process (e.g. a
> > > process from another cpuset or with a different nodemask).
> > >
> > > Let's thaw an OOM selected frozen process right after we've sent fatal
> > > signal from oom_kill_task.
> > > Thawing is safe if the frozen task doesn't access any suspended device
> > > (e.g. by ioctl) on the way out to the userspace where we handle the
> > > signal and die. Note, we are not interested in the kernel threads because
> > > they are not oom killable.
> > >
> > > Accessing suspended devices by a userspace processes shouldn't be an
> > > issue because devices are suspended only after userspace is already
> > > frozen and oom is disabled at that time.
> > >
> > > run_guest (drivers/lguest/core.c) calls try_to_freeze with an user
> > > context but it seems it is able to cope with signals because it
> > > explicitly checks for pending signals so we should be safe.
> > >
> > > Other than that userspace accesses the fridge only from the
> > > signal handling routines so we are able to handle SIGKILL without any
> > > negative side effects.
> > >
> > > Signed-off-by: Michal Hocko <mhocko@suse.cz>
> > > Reported-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
> >
> > Acked-by: Rafael J. Wysocki <rjw@sisk.pl>
> >
>
> Acked-by: David Rientjes <rientjes@google.com>

Thanks!

>
> Although this still seems to be problematic if the chosen thread gets
> frozen before the SIGKILL can be handled. We don't have any checks for
> fatal_signal_pending() when freezing threads and waiting for them to exit?

I guess you mean a situation when select_bad_process picks up a process
which is not marked as frozen yet but we send SIGKILL right before
schedule is called in refrigerator.
In that case either schedule should catch it by signal_pending_state
check or we will pick it up next OOM round when we pick up the same
process (if nothing else is eligible). Or am I missing something?

> Michal, could you send Andrew your revised patch with all the acked-bys?

Yes I will. I would just like to hear back from Konstantin who
originally reported the issue. Maybe he has a test case.

>
> Thanks!
> --
> 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/

--
Michal Hocko
SUSE Labs
SUSE LINUX s.r.o.
Lihovarska 1060/12
190 00 Praha 9
Czech Republic


\
 
 \ /
  Last update: 2011-09-27 09:55    [W:0.068 / U:0.272 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site