From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S964972AbWBANWd (ORCPT ); Wed, 1 Feb 2006 08:22:33 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S964976AbWBANWd (ORCPT ); Wed, 1 Feb 2006 08:22:33 -0500 Received: from mx2.mail.elte.hu ([157.181.151.9]:27578 "EHLO mx2.mail.elte.hu") by vger.kernel.org with ESMTP id S964972AbWBANWc (ORCPT ); Wed, 1 Feb 2006 08:22:32 -0500 Date: Wed, 1 Feb 2006 14:20:54 +0100 From: Ingo Molnar To: Nick Piggin Cc: Steven Rostedt , Peter Williams , Thomas Gleixner , Andrew Morton , LKML Subject: Re: [PATCH] Avoid moving tasks when a schedule can be made. Message-ID: <20060201132054.GA31156@elte.hu> References: <1138736609.7088.35.camel@localhost.localdomain> <43E02CC2.3080805@bigpond.net.au> <1138797874.7088.44.camel@localhost.localdomain> <43E0B24E.8080508@yahoo.com.au> <43E0B342.6090700@yahoo.com.au> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <43E0B342.6090700@yahoo.com.au> User-Agent: Mutt/1.4.2.1i X-ELTE-SpamScore: -2.2 X-ELTE-SpamLevel: X-ELTE-SpamCheck: no X-ELTE-SpamVersion: ELTE 2.0 X-ELTE-SpamCheck-Details: score=-2.2 required=5.9 tests=ALL_TRUSTED,AWL autolearn=no SpamAssassin version=3.0.3 -2.8 ALL_TRUSTED Did not pass through any untrusted hosts 0.6 AWL AWL: From: address is in the auto white-list X-ELTE-VirusStatus: clean Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org * Nick Piggin wrote: > >Once we've gone to the trouble of deciding which tasks to move and how > >many (and the estimate should be very conservative), and locked the source > >and destination runqueues, it is a very good idea to follow up with our > >threat of actually moving the tasks rather than bail out early. > > Oh, I forgot: Ingo once introduced some code to bail early (though for > different reasons and under different conditions), and this actually > was found to cause significant regressions in some database workloads. well, we both did changes with that effect - pretty much any change in this area can cause a regression on _some_ workload ;) So there wont be any silver bullet. > So it is not a nice thing to tinker with unless there is good reason. unbound latencies with hardirqs off are obviously a good reason - but i agree that the solution is not good enough, yet. Ingo