From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: From: Kirill Tkhai To: Peter Zijlstra Cc: Ingo Molnar , "linux-kernel@vger.kernel.org" , Steven Rostedt , "stable@vger.kernel.org" In-Reply-To: <20131217124656.GI21999@twins.programming.kicks-ass.net> References: <49231385567953@web4m.yandex.ru> <20131213154211.GP21999@twins.programming.kicks-ass.net> <122071387281778@web18h.yandex.ru> <20131217124656.GI21999@twins.programming.kicks-ass.net> Subject: Re: [PATCH] sched/rt: Fix rq's cpupri leak while enqueue/dequeue child RT entities MIME-Version: 1.0 Message-Id: <96341387285682@web18g.yandex.ru> Date: Tue, 17 Dec 2013 17:08:02 +0400 Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=koi8-r Sender: linux-kernel-owner@vger.kernel.org List-ID: 17.12.2013, 16:47, "Peter Zijlstra" : > On Tue, Dec 17, 2013 at 04:02:58PM +0400, Kirill Tkhai wrote: > >> �13.12.2013, 19:42, "Peter Zijlstra" : >>> �On Wed, Nov 27, 2013 at 07:59:13PM +0400, Kirill Tkhai wrote: >>>> ��This patch touches RT group scheduling case. >>>> >>>> ��Functions inc_rt_prio_smp() and dec_rt_prio_smp() change (global) rq's priority, >>>> ��while rt_rq passed to them may be not the top-level rt_rq. This is wrong, because >>>> ��changing of priority on a child level does not guarantee that the priority is >>>> ��the highest all over the rq. So, this leak makes RT balancing unusable. >>>> >>>> ��The short example: the task having the highest priority among all rq's RT tasks >>>> ��(no one other task has the same priority) are waking on a throttle rt_rq. >>>> ��The rq's cpupri is set to the task's priority equivalent, but real >>>> ��rq->rt.highest_prio.curr is less. >>>> >>>> ��The patch below fixes the problem. >>>> >>>> ��It looks like all version have this bug, so I CC'ed stable mailing list. >>> �Yeah, I think this is right. >>> >>> �cpupri stuff should indeed only be changed for the top level group. >> �Ingo, are you going to apply this patch? Or will you give any comments? > > I queued it, Ingo should get it through me somewhere today if all things > go well. > > Thanks Thanks, Peter