From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754970Ab1DSMRi (ORCPT ); Tue, 19 Apr 2011 08:17:38 -0400 Received: from relay.parallels.com ([195.214.232.42]:49102 "EHLO relay.parallels.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754570Ab1DSMRh (ORCPT ); Tue, 19 Apr 2011 08:17:37 -0400 Message-ID: <4DAD7D5F.7040407@parallels.com> Date: Tue, 19 Apr 2011 16:17:35 +0400 From: Konstantin Khlebnikov User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.1.18) Gecko/20110416 SeaMonkey/2.0.13 MIME-Version: 1.0 To: Paul Menage CC: "linux-kernel@vger.kernel.org" Subject: Re: [PATCH] cpuset: allow empty cpu/node masks References: <20110419100739.8464.80475.stgit@localhost6> <4DAD6D22.7000002@parallels.com> In-Reply-To: Content-Type: text/plain; charset="ISO-8859-1"; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Paul Menage wrote: > On Tue, Apr 19, 2011 at 1:08 PM, Konstantin Khlebnikov > wrote: >> Is cgroup.clone_children=1 planned to be default? > > I don't think so. It would be a user-visible API change, for no > significant benefit. > >> >> So, I just want to make cgroup worked out of the box: >> >> mount -t cgroup cgroup /cgroup >> mkdir /cgroup/foo >> echo $$> /cgroup/foo/tasks > > Just because cgroups has a very simple filesystem-based ASCII API, > that doesn't stop it from being a complex and > barely-human-comprehendable system. :-) > > I'm not sure what we gain by making that approach work - to do useful > stuff with cgroups (rather than simply playing around with moving > tasks into cgroups) you do need to have a better understanding of > what's going on. It's not too hard to make the initial instructions > explain how to only mount the subsystems that you want, or pass the > clone_children option. > Ok, then the problem is that the cgroupfs can be mounted without the specifying set of subsystems. Thus this operation is unsafe for future compatibily. I think it is best to oblige all subsystems to initialize their css by default with reasonable unlimited values or to inherit they from parent. Currently cpuset is only one cgroup that "does not work by default", and I see no reason why it should be special.