linux-kernel - Re: [PATCH v2] mm,oom: exclude oom_task_origin processes if they are OOM-unkillable.

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20160218080909.GA18149@dhcp22.suse.cz>
Date:	Thu, 18 Feb 2016 09:09:09 +0100
From:	Michal Hocko <mhocko@...nel.org>
To:	David Rientjes <rientjes@...gle.com>
Cc:	Tetsuo Handa <penguin-kernel@...ove.sakura.ne.jp>,
	akpm@...ux-foundation.org, mgorman@...e.de, oleg@...hat.com,
	torvalds@...ux-foundation.org, hughd@...gle.com, andrea@...nel.org,
	riel@...hat.com, linux-mm@...ck.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] mm,oom: exclude oom_task_origin processes if they are
 OOM-unkillable.

On Wed 17-02-16 14:31:54, David Rientjes wrote:
> On Wed, 17 Feb 2016, Tetsuo Handa wrote:
> 
> > oom_scan_process_thread() returns OOM_SCAN_SELECT when there is a
> > thread which returns oom_task_origin() == true. But it is possible
> > that such thread is marked as OOM-unkillable. In that case, the OOM
> > killer must not select such process.
> > 
> > Since it is meaningless to return OOM_SCAN_OK for OOM-unkillable
> > process because subsequent oom_badness() call will return 0, this
> > patch changes oom_scan_process_thread to return OOM_SCAN_CONTINUE
> > if that process is marked as OOM-unkillable (regardless of
> > oom_task_origin()).
> > 
> > Signed-off-by: Tetsuo Handa <penguin-kernel@...ove.SAKURA.ne.jp>
> > Suggested-by: Michal Hocko <mhocko@...nel.org>
> > ---
> >  mm/oom_kill.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/mm/oom_kill.c b/mm/oom_kill.c
> > index 7653055..cf87153 100644
> > --- a/mm/oom_kill.c
> > +++ b/mm/oom_kill.c
> > @@ -282,7 +282,7 @@ enum oom_scan_t oom_scan_process_thread(struct oom_control *oc,
> >  		if (!is_sysrq_oom(oc))
> >  			return OOM_SCAN_ABORT;
> >  	}
> > -	if (!task->mm)
> > +	if (!task->mm || task->signal->oom_score_adj == OOM_SCORE_ADJ_MIN)
> >  		return OOM_SCAN_CONTINUE;
> >  
> >  	/*
> 
> I'm getting multiple emails from you with the identical patch, something 
> is definitely wacky in your toolchain.
> 
> Anyway, this is NACK'd since task->signal->oom_score_adj is checked under 
> task_lock() for threads with memory attached, that's the purpose of 
> finding the correct thread in oom_badness() and taking task_lock().  We 
> aren't going to duplicate logic in several functions that all do the same 
> thing.

Is the task_lock really necessary, though? E.g. oom_task_origin()
doesn't seem to depend on it for task->signal safety. If you are
referring to races with changing oom_score_adj does such a race matter
at all?

To me this looks like a reasonable cleanup because we _know_ that
OOM_SCORE_ADJ_MIN means OOM_SCAN_CONTINUE and do not really have to go
down to oom_badness to find that out. Or what am I missing?

-- 
Michal Hocko
SUSE Labs