From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1757481Ab3KMR2M (ORCPT ); Wed, 13 Nov 2013 12:28:12 -0500 Received: from mx1.redhat.com ([209.132.183.28]:29596 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755507Ab3KMR2J (ORCPT ); Wed, 13 Nov 2013 12:28:09 -0500 Date: Wed, 13 Nov 2013 18:29:18 +0100 From: Oleg Nesterov To: Sameer Nanda Cc: akpm@linux-foundation.org, mhocko@suse.cz, rientjes@google.com, hannes@cmpxchg.org, rusty@rustcorp.com.au, semenzato@google.com, murzin.v@gmail.com, dserrg@gmail.com, msb@chromium.org, linux-mm@kvack.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH v6] mm, oom: Fix race when selecting process to kill Message-ID: <20131113172918.GA20145@redhat.com> References: <1384363093-8025-1-git-send-email-snanda@chromium.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1384363093-8025-1-git-send-email-snanda@chromium.org> User-Agent: Mutt/1.5.18 (2008-05-17) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 11/13, Sameer Nanda wrote: > > The selection of the process to be killed happens in two spots: > first in select_bad_process and then a further refinement by > looking for child processes in oom_kill_process. Since this is > a two step process, it is possible that the process selected by > select_bad_process may get a SIGKILL just before oom_kill_process > executes. If this were to happen, __unhash_process deletes this > process from the thread_group list. This results in oom_kill_process > getting stuck in an infinite loop when traversing the thread_group > list of the selected process. > > Fix this race by adding a pid_alive check for the selected process > with tasklist_lock held in oom_kill_process. I am fine with this patch as well, but honestly I'd prefer the previous v5. I won't argue though. > +/* > + * Careful: while_each_thread is not RCU safe. Callers should hold > + * read_lock(tasklist_lock) across while_each_thread loops. > + */ (tasklist_lock or siglock, in fact but this doesn't matter). This is not that simple, even tasklist_lock can't help if the task is already dead. Oh. Yes, sorry. I promised to send the patches "soon" many times, but still didn't find the time. Perhaps I should try to start with the "make this all less buggy" changes, the "complete" fix needs to change the callers as well. Oleg.