From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755665AbXJ2XRf (ORCPT ); Mon, 29 Oct 2007 19:17:35 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1754563AbXJ2XRV (ORCPT ); Mon, 29 Oct 2007 19:17:21 -0400 Received: from mx3.mail.elte.hu ([157.181.1.138]:59533 "EHLO mx3.mail.elte.hu" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753644AbXJ2XRU (ORCPT ); Mon, 29 Oct 2007 19:17:20 -0400 Date: Tue, 30 Oct 2007 00:17:02 +0100 From: Ingo Molnar To: Zachary Amsden Cc: Glauber de Oliveira Costa , linux-kernel@vger.kernel.org, tglx@linutronix.de, rusty@rustcorp.com.au, jeremy@goop.org, --cc@redhat.com, avi@quramnet.com, kvm-devel@lists.sourceforge.net, Glauber de Oliveira Costa , Dan Hecht , Garrett Smith Subject: Re: [PATCH] raise tsc clocksource rating Message-ID: <20071029231702.GA8506@elte.hu> References: <11936994092607-git-send-email-gcosta@redhat.com> <1193697734.9793.86.camel@bodhitayantram.eng.vmware.com> <20071029224852.GA27547@elte.hu> <1193698505.9793.90.camel@bodhitayantram.eng.vmware.com> <20071029230213.GA1982@elte.hu> <1193699638.9793.97.camel@bodhitayantram.eng.vmware.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1193699638.9793.97.camel@bodhitayantram.eng.vmware.com> User-Agent: Mutt/1.5.16 (2007-06-09) X-ELTE-VirusStatus: clean X-ELTE-SpamScore: -1.5 X-ELTE-SpamLevel: X-ELTE-SpamCheck: no X-ELTE-SpamVersion: ELTE 2.0 X-ELTE-SpamCheck-Details: score=-1.5 required=5.9 tests=BAYES_00 autolearn=no SpamAssassin version=3.1.7-deb -1.5 BAYES_00 BODY: Bayesian spam probability is 0 to 1% [score: 0.0000] Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org * Zachary Amsden wrote: > On Tue, 2007-10-30 at 00:02 +0100, Ingo Molnar wrote: > > * Zachary Amsden wrote: > > > > Not every guest support paravirt, but for correctness, all guests > > > require TSC, which must be exposed all the way up to userspace, no > > > matter what the efficiency or accuracy may be. > > > > but if there's a perfect TSC available (there is such hardware) then the > > TSC _is_ the best clocksource. Paravirt now turns it off unconditionally > > in essence. > > No, if no paravirt clocksource is detected, nothing can override the > perfect TSC hardware clocksource rating of 400. And if a paravirt > clocksource is detected, it is always better than TSC. > > > anyway, that's at most an optimization issue. No strong feelings here, > > and we can certainly delay this patch until this gets all sorted out. > > This patch should be nacked, since it is just wrong. This is not an > optimization issue. It is an accuracy issue for all virtualization > environments that affects long term kernel clock stability, which is > important to fix, and the best way to do that is to use a paravirt > clocksource. i know it's not an optimization issue. Your current pessimisation of even perfect TSCs _is_ an optimization issue. (and note that if the TSC is unstable the guest will/should notice it and will fall back to the hyper clocksource) Ingo From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ingo Molnar Subject: Re: [PATCH] raise tsc clocksource rating Date: Tue, 30 Oct 2007 00:17:02 +0100 Message-ID: <20071029231702.GA8506@elte.hu> References: <11936994092607-git-send-email-gcosta@redhat.com> <1193697734.9793.86.camel@bodhitayantram.eng.vmware.com> <20071029224852.GA27547@elte.hu> <1193698505.9793.90.camel@bodhitayantram.eng.vmware.com> <20071029230213.GA1982@elte.hu> <1193699638.9793.97.camel@bodhitayantram.eng.vmware.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Cc: jeremy-TSDbQ3PG+2Y@public.gmane.org, avi-i/T1Zxg06OZWk0Htik3J/w@public.gmane.org, kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org, Dan Hecht , linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, Glauber de Oliveira Costa , Glauber de Oliveira Costa , --cc-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org, Garrett Smith To: Zachary Amsden Return-path: Content-Disposition: inline In-Reply-To: <1193699638.9793.97.camel-cxY/u30q8FloTgUnLF1by8fTvwmfpRNyZeezCHUQhQ4@public.gmane.org> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: kvm-devel-bounces-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org Errors-To: kvm-devel-bounces-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org List-Id: kvm.vger.kernel.org * Zachary Amsden wrote: > On Tue, 2007-10-30 at 00:02 +0100, Ingo Molnar wrote: > > * Zachary Amsden wrote: > > > > Not every guest support paravirt, but for correctness, all guests > > > require TSC, which must be exposed all the way up to userspace, no > > > matter what the efficiency or accuracy may be. > > > > but if there's a perfect TSC available (there is such hardware) then the > > TSC _is_ the best clocksource. Paravirt now turns it off unconditionally > > in essence. > > No, if no paravirt clocksource is detected, nothing can override the > perfect TSC hardware clocksource rating of 400. And if a paravirt > clocksource is detected, it is always better than TSC. > > > anyway, that's at most an optimization issue. No strong feelings here, > > and we can certainly delay this patch until this gets all sorted out. > > This patch should be nacked, since it is just wrong. This is not an > optimization issue. It is an accuracy issue for all virtualization > environments that affects long term kernel clock stability, which is > important to fix, and the best way to do that is to use a paravirt > clocksource. i know it's not an optimization issue. Your current pessimisation of even perfect TSCs _is_ an optimization issue. (and note that if the TSC is unstable the guest will/should notice it and will fall back to the hyper clocksource) Ingo ------------------------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/