From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756070Ab1HaNmb (ORCPT ); Wed, 31 Aug 2011 09:42:31 -0400 Received: from mail-ww0-f44.google.com ([74.125.82.44]:42790 "EHLO mail-ww0-f44.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755464Ab1HaNma (ORCPT ); Wed, 31 Aug 2011 09:42:30 -0400 Date: Wed, 31 Aug 2011 15:42:24 +0200 From: Frederic Weisbecker To: Tejun Heo Cc: rjw@sisk.pl, paul@paulmenage.org, lizf@cn.fujitsu.com, linux-pm@lists.linux-foundation.org, linux-kernel@vger.kernel.org, containers@lists.linux-foundation.org, matthltc@us.ibm.com, kamezawa.hiroyu@jp.fujitsu.com Subject: Re: [PATCH 6/6] cgroup: kill subsys->can_attach_task(), pre_attach() and attach_task() Message-ID: <20110831134220.GB20598@somewhere> References: <1314312192-26885-1-git-send-email-tj@kernel.org> <1314312192-26885-7-git-send-email-tj@kernel.org> <20110830201030.GC15953@somewhere.redhat.com> <20110831070313.GA29179@mtj.dyndns.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20110831070313.GA29179@mtj.dyndns.org> 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, Aug 31, 2011 at 09:03:13AM +0200, Tejun Heo wrote: > Hello, Frederic. > > On Tue, Aug 30, 2011 at 10:10:32PM +0200, Frederic Weisbecker wrote: > > In order to keep the fix queued in -mm (https://lkml.org/lkml/2011/8/26/262) > > the tasks that have failed to migrate should be removed from the iterator > > so that they are not included in the batch in ->attach(). > > I don't think that's a good approach. It breaks the symmetry when > calling different callbacks. What if ->can_attach() allocates > per-task resources and the task exits in the middle? I think we > better lock down fork/exit/exec. I'll send patches but I'm currently > moving / traveling w/ limited access to my toys so it might take some > time. My task counter subsystem patchset brings a cancel_attach_task() callback that cancels can_attach_task() effects. I thought that rebased on top of your patch it's going to be merged inside cancel_attach() but OTOH we can't cancel the effect of failed migration on a thread that way. May be we need to keep a cancel_attach_task() just for that purpose?