From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754156AbdLFJAW (ORCPT ); Wed, 6 Dec 2017 04:00:22 -0500 Received: from mx2.suse.de ([195.135.220.15]:51147 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753562AbdLFJAV (ORCPT ); Wed, 6 Dec 2017 04:00:21 -0500 Date: Wed, 6 Dec 2017 10:00:19 +0100 From: Michal Hocko To: David Rientjes Cc: Tetsuo Handa , Andrew Morton , Andrea Arcangeli , linux-kernel@vger.kernel.org, linux-mm@kvack.org Subject: Re: Multiple oom_reaper BUGs: unmap_page_range racing with exit_mmap Message-ID: <20171206090019.GE16386@dhcp22.suse.cz> References: <201712060328.vB63SrDK069830@www262.sakura.ne.jp> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.9.1 (2017-09-22) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue 05-12-17 23:48:21, David Rientjes wrote: [...] > I think this argues to do MMF_REAPING-style behavior at the beginning of > exit_mmap() and avoid reaping all together once we have reached that > point. There are no more users of the mm and we are in the process of > tearing it down, I'm not sure that the oom reaper should be in the > business with trying to interfere with that. Or are there actual bug > reports where an oom victim gets wedged while in exit_mmap() prior to > releasing its memory? Something like that seem to work indeed. But we should better understand what is going on here before adding new oom reaper specific kludges. So let's focus on getting more information from your crashes first. -- Michal Hocko SUSE Labs