From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1161498AbcFGQLK (ORCPT ); Tue, 7 Jun 2016 12:11:10 -0400 Received: from foss.arm.com ([217.140.101.70]:46008 "EHLO foss.arm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753967AbcFGQLJ (ORCPT ); Tue, 7 Jun 2016 12:11:09 -0400 Date: Tue, 7 Jun 2016 17:11:34 +0100 From: Juri Lelli To: Daniel Bristot de Oliveira Cc: linux-kernel@vger.kernel.org, Rik van Riel , "Luis Claudio R. Goncalves" , Tejun Heo , Li Zefan , Johannes Weiner , cgroups@vger.kernel.org Subject: Re: [PATCH] cgroup: disable irqs while holding css_set_lock Message-ID: <20160607161134.GU9340@e106622-lin> References: <20160607101402.GP9340@e106622-lin> <20160607133009.GS9340@e106622-lin> <494c1f85-bbab-99f1-4095-914aaebdd708@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <494c1f85-bbab-99f1-4095-914aaebdd708@redhat.com> 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 07/06/16 12:26, Daniel Bristot de Oliveira wrote: > Ciao Juri, > > On 06/07/2016 10:30 AM, Juri Lelli wrote: > > So, this and the partitioned one could actually overlap, since we don't > > set cpu_exclusive. Is that right? > > > > I guess affinity mask of both m processes gets set correclty, but I'm > > not sure if we are missing one check in the admission control. Can you > > actually create two overlapping sets and get DEADLINE tasks running in > > them? For example, what happens if partitioned is [4] and clustered is > > [4-7]? Does setattr() fail? > > That is what I was trying to understand/break. Fortunately, I still > can't break it! bravo! :-) > > In the test you mentioned the task in the "clustered" fails to > sched_setattr(). > > Test output example: > + '[' '!' -d partitioned ']' > + mkdir partitioned > + echo 4 > + echo 0 > + echo 0 > + echo 1155 > + cat /proc/self/cpuset > /partitioned > + /root/m > main thread [1162] > deadline thread started [1164] Mmm. The other one is good, but this one looks still suspect to me. We shouldn't have created a new root domain at this point (since cpu_exclusive is not set), so we shouldn't be able to admit DEADLINE tasks with a smaller affinity mask than the full root domain mask (which should still be 0-7 at this point), if I'm interpreting your script correctly and I'm not wrong about root domains stuff. Unfortunately I'm not able to test this myself right now. Hopefully next week. Thanks a lot for testing, though! Best, - Juri > + '[' '!' -d clustered ']' > + mkdir clustered > + echo 4-7 > + echo 0 > + echo 0 > + echo 1 > + echo 1155 > + cat /proc/self/cpuset > + /root/m > /clustered > + /root/m > main thread [1166] > sched_setattr: Operation not permitted > > I will let you know if I find something odd. > > -- Daniel >