Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756693Ab0FOJ71 (ORCPT ); Tue, 15 Jun 2010 05:59:27 -0400 Received: from mail-iw0-f174.google.com ([209.85.214.174]:48917 "EHLO mail-iw0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751026Ab0FOJ70 convert rfc822-to-8bit (ORCPT ); Tue, 15 Jun 2010 05:59:26 -0400 DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type:content-transfer-encoding; b=J+SMdSYpiRijwMe+/nsVWEJM/CT2AyX1SMRooYNHFTHNRhlWu4zTVn5paef74dnnJ5 i3ERn2DbdVXywHwyC5TGSzV2HutXkkC2SF9GBfJxN8ptzUG6vUiYCtbqlk5lEQR5nrbc t/EBKdebWh86sYUuxx9sz2jNOCUzGyRs16DbA= MIME-Version: 1.0 In-Reply-To: <20100615152450.f82c1f8c.kamezawa.hiroyu@jp.fujitsu.com> References: <20100615152450.f82c1f8c.kamezawa.hiroyu@jp.fujitsu.com> Date: Tue, 15 Jun 2010 18:59:25 +0900 Message-ID: Subject: Re: [PATCH] use find_lock_task_mm in memory cgroups oom From: Minchan Kim To: KAMEZAWA Hiroyuki Cc: "linux-mm@kvack.org" , "linux-kernel@vger.kernel.org" , "nishimura@mxp.nes.nec.co.jp" , "balbir@linux.vnet.ibm.com" , Oleg Nesterov , "akpm@linux-foundation.org" Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8BIT Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 4241 Lines: 126 Hi, Kame. On Tue, Jun 15, 2010 at 3:24 PM, KAMEZAWA Hiroyuki wrote: > > based on  oom-introduce-find_lock_task_mm-to-fix-mm-false-positives.patch > tested on mm-of-the-moment snapshot 2010-06-11-16-40. > > == > From: KAMEZAWA Hiroyuki > > When the OOM killer scans task, it check a task is under memcg or > not when it's called via memcg's context. > > But, as Oleg pointed out, a thread group leader may have NULL ->mm > and task_in_mem_cgroup() may do wrong decision. We have to use > find_lock_task_mm() in memcg as generic OOM-Killer does. > > Cc: Oleg Nesterov > Cc: Daisuke Nishimura > Cc: Balbir Singh > Signed-off-by: KAMEZAWA Hiroyuki Reviewed-by: Minchan Kim I have a trivial comment below. > --- >  include/linux/oom.h |    2 ++ >  mm/memcontrol.c     |   10 +++++++--- >  mm/oom_kill.c       |    8 ++++++-- >  3 files changed, 15 insertions(+), 5 deletions(-) > > Index: mmotm-2.6.35-0611/include/linux/oom.h > =================================================================== > --- mmotm-2.6.35-0611.orig/include/linux/oom.h > +++ mmotm-2.6.35-0611/include/linux/oom.h > @@ -45,6 +45,8 @@ static inline void oom_killer_enable(voi >        oom_killer_disabled = false; >  } > > +extern struct task_struct *find_lock_task_mm(struct task_struct *p); > + >  /* sysctls */ >  extern int sysctl_oom_dump_tasks; >  extern int sysctl_oom_kill_allocating_task; > Index: mmotm-2.6.35-0611/mm/memcontrol.c > =================================================================== > --- mmotm-2.6.35-0611.orig/mm/memcontrol.c > +++ mmotm-2.6.35-0611/mm/memcontrol.c > @@ -47,6 +47,7 @@ >  #include >  #include >  #include > +#include >  #include "internal.h" > >  #include > @@ -838,10 +839,13 @@ int task_in_mem_cgroup(struct task_struc >  { >        int ret; >        struct mem_cgroup *curr = NULL; > +       struct task_struct *p; > > -       task_lock(task); > -       curr = try_get_mem_cgroup_from_mm(task->mm); > -       task_unlock(task); > +       p = find_lock_task_mm(task); > +       if (!p) > +               return 0; > +       curr = try_get_mem_cgroup_from_mm(p->mm); > +       task_unlock(p); >        if (!curr) >                return 0; >        /* > Index: mmotm-2.6.35-0611/mm/oom_kill.c > =================================================================== > --- mmotm-2.6.35-0611.orig/mm/oom_kill.c > +++ mmotm-2.6.35-0611/mm/oom_kill.c > @@ -81,13 +81,17 @@ static bool has_intersects_mems_allowed( >  } >  #endif /* CONFIG_NUMA */ > > -/* > +/** > + * find_lock_task_mm - Checking a process which a task belongs to has valid mm > + * and return a locked task which has a valid pointer to mm. > + * This comment should have been another patch. BTW, below comment uses "subthread" word. Personally it's easy to understand function's goal to me. :) How about following as? Checking a process which has any subthread with vaild mm .... > + * @p: the task of a process to be checked. >  * The process p may have detached its own ->mm while exiting or through >  * use_mm(), but one or more of its subthreads may still have a valid >  * pointer.  Return p, or any of its subthreads with a valid ->mm, with >  * task_lock() held. >  */ > -static struct task_struct *find_lock_task_mm(struct task_struct *p) > +struct task_struct *find_lock_task_mm(struct task_struct *p) >  { >        struct task_struct *t = p; > > > -- > To unsubscribe, send a message with 'unsubscribe linux-mm' in > the body to majordomo@kvack.org.  For more info on Linux MM, > see: http://www.linux-mm.org/ . > Don't email: email@kvack.org > -- Kind regards, Minchan Kim -- 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/