From mboxrd@z Thu Jan 1 00:00:00 1970 From: Rik van Riel Subject: Re: [RFC -v2 PATCH 2/3] sched: add yield_to function Date: Tue, 28 Dec 2010 17:34:48 -0500 Message-ID: <4D1A6608.6040503@redhat.com> References: <20101213224434.7495edb2@annuminas.surriel.com> <20101213224657.7e141746@annuminas.surriel.com> <1292306896.7448.157.camel@marge.simson.net> <4D0A6D34.6070806@redhat.com> <1292569018.7772.75.camel@marge.simson.net> <1292570143.7772.84.camel@marge.simson.net> <4D0F78DF.9010908@redhat.com> <1292861059.15207.7.camel@marge.simson.net> <1293515683.6930.16.camel@marge.simson.net> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Cc: kvm@vger.kernel.org, linux-kernel@vger.kernel.org, Avi Kiviti , Srivatsa Vaddagiri , Peter Zijlstra , Chris Wright To: Mike Galbraith Return-path: Received: from mx1.redhat.com ([209.132.183.28]:3588 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752046Ab0L1WfD (ORCPT ); Tue, 28 Dec 2010 17:35:03 -0500 In-Reply-To: <1293515683.6930.16.camel@marge.simson.net> Sender: kvm-owner@vger.kernel.org List-ID: On 12/28/2010 12:54 AM, Mike Galbraith wrote: > On Mon, 2010-12-20 at 17:04 +0100, Mike Galbraith wrote: >> On Mon, 2010-12-20 at 10:40 -0500, Rik van Riel wrote: >>> On 12/17/2010 02:15 AM, Mike Galbraith wrote: >>> >>>> BTW, with this vruntime donation thingy, what prevents a task from >>>> forking off accomplices who do nothing but wait for a wakeup and >>>> yield_to(exploit)? >>>> >>>> Even swapping vruntimes in the same cfs_rq is dangerous as hell, because >>>> one party is going backward. >>> >>> I just realized the answer to this question. >>> >>> We only give cpu time to tasks that are runnable, but not >>> currently running. That ensures one task cannot block others >>> from running by having time yielded to it constantly. >> >> Hm. Don't think that will 100% sure prevent clock stoppage, because the >> running task doesn't necessarily advance min_vruntime. >> >> What about something like the below instead? It compiles, but may eat >> your first born child. > > (either it's christmas, or nobody cares to try it. ho ho hum:) It's christmas. I've replaced my original patch with your patch in my series, just haven't gotten around to running Avi's latest test on it yet :) -- All rights reversed