From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751372Ab3LLIoZ (ORCPT ); Thu, 12 Dec 2013 03:44:25 -0500 Received: from cantor2.suse.de ([195.135.220.15]:51480 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750873Ab3LLIoX (ORCPT ); Thu, 12 Dec 2013 03:44:23 -0500 Date: Thu, 12 Dec 2013 09:44:20 +0100 From: Michal Hocko To: David Rientjes Cc: Andrew Morton , Johannes Weiner , KOSAKI Motohiro , KAMEZAWA Hiroyuki , linux-mm@kvack.org, LKML Subject: Re: [PATCH] memcg, oom: lock mem_cgroup_print_oom_info Message-ID: <20131212084420.GA2630@dhcp22.suse.cz> References: <1386776545-24916-1-git-send-email-mhocko@suse.cz> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed 11-12-13 14:23:18, David Rientjes wrote: > On Wed, 11 Dec 2013, Michal Hocko wrote: > > > mem_cgroup_print_oom_info uses a static buffer (memcg_name) to store the > > name of the cgroup. This is not safe as pointed out by David Rientjes > > because memcg oom is locked only for its hierarchy and nothing prevents > > another parallel hierarchy to trigger oom as well and overwrite the > > already in-use buffer. > > > > This patch introduces oom_info_lock hidden inside mem_cgroup_print_oom_info > > which is held throughout the function. It make access to memcg_name safe > > and as a bonus it also prevents parallel memcg ooms to interleave their > > statistics which would make the printed data hard to analyze otherwise. > > > > Signed-off-by: Michal Hocko > > Acked-by: David Rientjes Thanks > > > --- > > mm/memcontrol.c | 12 +++++++----- > > 1 file changed, 7 insertions(+), 5 deletions(-) > > > > diff --git a/mm/memcontrol.c b/mm/memcontrol.c > > index 28c9221b74ea..c72b03bf9679 100644 > > --- a/mm/memcontrol.c > > +++ b/mm/memcontrol.c > > @@ -1647,13 +1647,13 @@ static void move_unlock_mem_cgroup(struct mem_cgroup *memcg, > > */ > > void mem_cgroup_print_oom_info(struct mem_cgroup *memcg, struct task_struct *p) > > { > > - struct cgroup *task_cgrp; > > - struct cgroup *mem_cgrp; > > /* > > - * Need a buffer in BSS, can't rely on allocations. The code relies > > - * on the assumption that OOM is serialized for memory controller. > > - * If this assumption is broken, revisit this code. > > + * protects memcg_name and makes sure that parallel ooms do not > > + * interleave > > Parallel memcg oom kills can happen in disjoint memcg hierarchies, this > just prevents the printing of the statistics from interleaving. I'm not > sure if that's clear from this comment. What about this instead: * Protects memcg_name and makes sure that ooms from parallel * hierarchies do not interleave. ? -- Michal Hocko SUSE Labs