From mboxrd@z Thu Jan 1 00:00:00 1970 From: Peter Zijlstra Subject: Re: [RFC v2 4/7] change kernel accounting to include steal time Date: Fri, 03 Sep 2010 09:18:59 +0200 Message-ID: <1283498339.1783.24.camel@laptop> References: <1283184391-7785-8-git-send-email-glommer@redhat.com> <4C7BEA9C.1060605@goop.org> <4C7BFACD.4030409@redhat.com> <4C7C0187.7040401@goop.org> <4C7C03CB.1060700@redhat.com> <1283196005.1820.1340.camel@laptop> <4C7C0A57.2010906@redhat.com> <4C7C3709.3040706@goop.org> <4C7C38BC.1090907@redhat.com> <1283242309.1820.1471.camel@laptop> <20100902181956.GE5933@mothafucka.localdomain> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 8BIT Cc: Rik van Riel , Jeremy Fitzhardinge , kvm@vger.kernel.org, avi@redhat.com, zamsden@redhat.com, mtosatti@redhat.com, mingo@elte.hu To: Glauber Costa Return-path: Received: from bombadil.infradead.org ([18.85.46.34]:46551 "EHLO bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752755Ab0ICHTN convert rfc822-to-8bit (ORCPT ); Fri, 3 Sep 2010 03:19:13 -0400 In-Reply-To: <20100902181956.GE5933@mothafucka.localdomain> Sender: kvm-owner@vger.kernel.org List-ID: On Thu, 2010-09-02 at 15:19 -0300, Glauber Costa wrote: > So what's the big drawback of just making sched_clock return sched_clock_unstolen? > When there is no steal time involved, they will just be equal anyway. > And this way, everybody that relies on sched_clock for whatever reason, > will probably work. Because there's a number of things wanting sched_clock() and the likes to be synchronized across cpus, its impossible to make _unstolen() synchronized because different vcpus can get different service levels. Also, I think you want to keep measuring scheduling latency and other such things in actual wall-time.