From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail202.messagelabs.com (mail202.messagelabs.com [216.82.254.227]) by kanga.kvack.org (Postfix) with ESMTP id CC31F6B007B for ; Tue, 24 Nov 2009 22:30:06 -0500 (EST) Received: from d28relay05.in.ibm.com (d28relay05.in.ibm.com [9.184.220.62]) by e28smtp02.in.ibm.com (8.14.3/8.13.1) with ESMTP id nAP3U0di003433 for ; Wed, 25 Nov 2009 09:00:00 +0530 Received: from d28av04.in.ibm.com (d28av04.in.ibm.com [9.184.220.66]) by d28relay05.in.ibm.com (8.13.8/8.13.8/NCO v10.0) with ESMTP id nAP3TxQ11192038 for ; Wed, 25 Nov 2009 09:00:00 +0530 Received: from d28av04.in.ibm.com (loopback [127.0.0.1]) by d28av04.in.ibm.com (8.14.3/8.13.1/NCO v10.0 AVout) with ESMTP id nAP3Tx00007057 for ; Wed, 25 Nov 2009 14:29:59 +1100 Date: Wed, 25 Nov 2009 08:59:55 +0530 From: Balbir Singh Subject: Re: [BUGFIX][PATCH -mmotm] memcg: avoid oom-killing innocent task in case of use_hierarchy Message-ID: <20091125032954.GC3365@balbir.in.ibm.com> Reply-To: balbir@linux.vnet.ibm.com References: <20091124145759.194cfc9f.nishimura@mxp.nes.nec.co.jp> <661de9470911240531p5e587c42w96995fde37dbd401@mail.gmail.com> <20091124230029.7245e1b8.d-nishimura@mtf.biglobe.ne.jp> <20091124170402.GB3365@balbir.in.ibm.com> <20091125084910.16d9095d.nishimura@mxp.nes.nec.co.jp> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <20091125084910.16d9095d.nishimura@mxp.nes.nec.co.jp> Sender: owner-linux-mm@kvack.org To: Daisuke Nishimura Cc: Andrew Morton , LKML , linux-mm , stable , David Rientjes , KAMEZAWA Hiroyuki , KOSAKI Motohiro List-ID: * nishimura@mxp.nes.nec.co.jp [2009-11-25 08:49:10]: > On Tue, 24 Nov 2009 22:34:02 +0530, Balbir Singh wrote: > > * Daisuke Nishimura [2009-11-24 23:00:29]: > > > > > On Tue, 24 Nov 2009 19:01:54 +0530 > > > Balbir Singh wrote: > > > > > > > On Tue, Nov 24, 2009 at 11:27 AM, Daisuke Nishimura > > > > wrote: > > > > > task_in_mem_cgroup(), which is called by select_bad_process() to check whether > > > > > a task can be a candidate for being oom-killed from memcg's limit, checks > > > > > "curr->use_hierarchy"("curr" is the mem_cgroup the task belongs to). > > > > > > > > > > But this check return true(it's false positive) when: > > > > > > > > > > /00 use_hierarchy == 0 <- hitting limit > > > > > /00/aa use_hierarchy == 1 <- "curr" > > > > > > > > > > This leads to killing an innocent task in 00/aa. This patch is a fix for this > > > > > bug. And this patch also fixes the arg for mem_cgroup_print_oom_info(). We > > > > > should print information of mem_cgroup which the task being killed, not current, > > > > > belongs to. > > > > > > > > > > > > > Quick Question: What happens if /00 has no tasks in it > > > > after your patches? > > > > > > > Nothing would happen because /00 never hit its limit. > > > > Why not? I am talking of a scenario where /00 is set to a > > limit (similar to your example) and hits its limit, but the groups > > under it have no limits, but tasks. Shouldn't we be scanning > > /00/aa as well? > > > > > > > > The bug that this patch fixes is: > > > > > > - create a dir /00 and set some limits. > > > - create a sub dir /00/aa w/o any limits, and enable hierarchy. > > > - run some programs in both in 00 and 00/aa. programs in 00 should be > > > big enough to cause oom by its limit. > > > - when oom happens by 00's limit, tasks in 00/aa can also be killed. > > > > > > > To be honest, the last part is fair, specifically if 00/aa has a task > > that is really the heaviest task as per the oom logic. no? Are you > > suggesting that only tasks in /00 should be selected by the > > oom logic? > > > All of your comments would be rational if hierarchy is enabled in 00(it's > also enabled in 00/aa automatically in this case). > I'm saying about the case where it's disabled in 00 but enabled in 00/aa. > OK, I misunderstood the example then, so even though hierarchy is disabled, we kill a task in 00/aa when 00 hits the limit. Thanks for clarifying. > In this scenario, charges by tasks in 00/aa is(and should not be) charged to 00. > And oom caused by 00's limit should not affect the task in 00/aa. > > > Regards, > Daisuke Nishimura. > > -- > 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 > -- Balbir -- 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