* Re: [PATCH] oom: do not kill tasks with oom_score_adj OOM_SCORE_ADJ_MIN
2011-11-07 21:54 ` [PATCH] oom: do not kill tasks with oom_score_adj OOM_SCORE_ADJ_MIN David Rientjes
@ 2011-11-07 22:18 ` Michal Hocko
2011-11-07 22:22 ` KOSAKI Motohiro
1 sibling, 0 replies; 3+ messages in thread
From: Michal Hocko @ 2011-11-07 22:18 UTC (permalink / raw)
To: David Rientjes
Cc: linux-mm, linux-kernel, Oleg Nesterov, Ying Han, KOSAKI Motohiro,
Andrew Morton
On Mon 07-11-11 13:54:38, David Rientjes wrote:
> On Fri, 4 Nov 2011, Michal Hocko wrote:
>
> > diff --git a/mm/oom_kill.c b/mm/oom_kill.c
> > index e916168..4883514 100644
> > --- a/mm/oom_kill.c
> > +++ b/mm/oom_kill.c
> > @@ -185,6 +185,9 @@ unsigned int oom_badness(struct task_struct *p, struct mem_cgroup *mem,
> > if (!p)
> > return 0;
> >
> > + if (p->signal->oom_score_adj == OOM_SCORE_ADJ_MIN)
> > + return 0;
> > +
> > /*
> > * The memory controller may have a limit of 0 bytes, so avoid a divide
> > * by zero, if necessary.
>
> This leaves p locked, you need to do task_unlock(p) first.
Yes, right you are. Thanks for spotting this out.
>
> Once that's fixed, please add my
>
> Acked-by: David Rientjes <rientjes@google.com>
Thanks.
> and resubmit to Andrew for the 3.2 rc series. Thanks!
Andrew, could you push this for 3.2 (bugfix for post 3.1 kernel).
---
>From 10e06e7c0682af58236055ed38bfab58f7935253 Mon Sep 17 00:00:00 2001
From: Michal Hocko <mhocko@suse.cz>
Date: Fri, 4 Nov 2011 12:59:44 +0100
Subject: [PATCH] oom: do not kill tasks with oom_score_adj OOM_SCORE_ADJ_MIN
To: David Rientjes <rientjes@google.com>
Cc: linux-mm@kvack.org,
linux-kernel@vger.kernel.org,
Oleg Nesterov <oleg@redhat.com>,
Ying Han <yinghan@google.com>,
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>,
Andrew Morton <akpm@linux-foundation.org>
c9f01245 (oom: remove oom_disable_count) has removed oom_disable_count
counter which has been used for early break out from oom_badness so we
could never select a task with oom_score_adj set to OOM_SCORE_ADJ_MIN
(oom disabled).
Now that the counter is gone we are always going through heuristics
calculation and we always return a non zero positive value. This
means that we can end up killing a task with OOM disabled because it is
indistinguishable from regular tasks with 1% resp. CAP_SYS_ADMIN tasks
with 3% usage of memory or tasks with oom_score_adj set but OOM enabled.
Let's break out early if the task should have OOM disabled.
Signed-off-by: Michal Hocko <mhocko@suse.cz>
Acked-by: David Rientjes <rientjes@google.com>
---
mm/oom_kill.c | 5 +++++
1 files changed, 5 insertions(+), 0 deletions(-)
diff --git a/mm/oom_kill.c b/mm/oom_kill.c
index e916168..4465fb8 100644
--- a/mm/oom_kill.c
+++ b/mm/oom_kill.c
@@ -185,6 +185,11 @@ unsigned int oom_badness(struct task_struct *p, struct mem_cgroup *mem,
if (!p)
return 0;
+ if (p->signal->oom_score_adj == OOM_SCORE_ADJ_MIN) {
+ task_unlock(p);
+ return 0;
+ }
+
/*
* The memory controller may have a limit of 0 bytes, so avoid a divide
* by zero, if necessary.
--
1.7.7.1
--
Michal Hocko
SUSE Labs
SUSE LINUX s.r.o.
Lihovarska 1060/12
190 00 Praha 9
Czech Republic
^ permalink raw reply [flat|nested] 3+ messages in thread* Re: [PATCH] oom: do not kill tasks with oom_score_adj OOM_SCORE_ADJ_MIN
2011-11-07 21:54 ` [PATCH] oom: do not kill tasks with oom_score_adj OOM_SCORE_ADJ_MIN David Rientjes
2011-11-07 22:18 ` Michal Hocko
@ 2011-11-07 22:22 ` KOSAKI Motohiro
1 sibling, 0 replies; 3+ messages in thread
From: KOSAKI Motohiro @ 2011-11-07 22:22 UTC (permalink / raw)
To: rientjes; +Cc: mhocko, linux-mm, linux-kernel, oleg, yinghan, akpm
(11/7/2011 1:54 PM), David Rientjes wrote:
> On Fri, 4 Nov 2011, Michal Hocko wrote:
>
>> diff --git a/mm/oom_kill.c b/mm/oom_kill.c
>> index e916168..4883514 100644
>> --- a/mm/oom_kill.c
>> +++ b/mm/oom_kill.c
>> @@ -185,6 +185,9 @@ unsigned int oom_badness(struct task_struct *p, struct mem_cgroup *mem,
>> if (!p)
>> return 0;
>>
>> + if (p->signal->oom_score_adj == OOM_SCORE_ADJ_MIN)
>> + return 0;
>> +
>> /*
>> * The memory controller may have a limit of 0 bytes, so avoid a divide
>> * by zero, if necessary.
>
> This leaves p locked, you need to do task_unlock(p) first.
>
> Once that's fixed, please add my
>
> Acked-by: David Rientjes <rientjes@google.com>
Agreed.
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
^ permalink raw reply [flat|nested] 3+ messages in thread