From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754876Ab1IWPrc (ORCPT ); Fri, 23 Sep 2011 11:47:32 -0400 Received: from mx2.parallels.com ([64.131.90.16]:54649 "EHLO mx2.parallels.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754861Ab1IWPra (ORCPT ); Fri, 23 Sep 2011 11:47:30 -0400 Message-ID: <4E7CA9B1.9070903@parallels.com> Date: Fri, 23 Sep 2011 12:45:53 -0300 From: Glauber Costa User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:6.0.2) Gecko/20110906 Thunderbird/6.0.2 MIME-Version: 1.0 To: Balbir Singh CC: Peter Zijlstra , Srivatsa Vaddagiri , , , , , , , Subject: Re: [PATCH 1/9] Remove parent field in cpuacct cgroup References: <1316030695-19826-1-git-send-email-glommer@parallels.com> <1316030695-19826-2-git-send-email-glommer@parallels.com> <1316448186.1511.19.camel@twins> <4E776937.1070108@parallels.com> <1316449160.6091.5.camel@twins> <4E776E36.6040906@parallels.com> <1316450385.6091.9.camel@twins> <4E7770BD.6050902@parallels.com> <1316457653.6091.16.camel@twins> <20110920172930.GA5318@linux.vnet.ibm.com> <1316704637.31429.25.camel@twins> In-Reply-To: Content-Type: text/plain; charset="ISO-8859-1"; format=flowed Content-Transfer-Encoding: 7bit X-Originating-IP: [201.82.134.226] Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 09/23/2011 05:09 AM, Balbir Singh wrote: >>> Android makes heavy use of cpuacct as well. We invented it to do cpu >>> accounting independent of control >> >> But its a massive waste of time to have to iterate a double hierarchy, >> cache-miss heaven. >> > > By double hierarchy you mean both cpu and cpuacct? What if they are > not mounted together? > >> Or so people really have independent cgroups as well? I thought people >> just co-mount all this crap. >> > > Please see > http://code.google.com/p/cgroupmgr/wiki/cgroupsForAndroid > http://stackoverflow.com/questions/6814207/android-cpuacct-usage-file > > AFAIK, some versions mount cpuacct at /acct and independent of cpu cgroups > > Balbir But then, are they using both cpuacct and cpu cgroup for an independent group of tasks? The problem here is that we're gathering the same statistics 3 times just because we need/want to present then in a different way. This is really bad design.