From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1759551Ab1CDMUh (ORCPT ); Fri, 4 Mar 2011 07:20:37 -0500 Received: from shutemov.name ([188.40.19.243]:42934 "EHLO shutemov.name" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1759400Ab1CDMUg (ORCPT ); Fri, 4 Mar 2011 07:20:36 -0500 Date: Fri, 4 Mar 2011 14:20:35 +0200 From: "Kirill A. Shutemov" To: Arnd Bergmann Cc: Paul Menage , Li Zefan , containers@lists.linux-foundation.org, jacob.jun.pan@linux.intel.com, Arjan van de Ven , linux-kernel@vger.kernel.org, Matt Helsley , Andrew Morton , linux-api@vger.kernel.org, Thomas Gleixner Subject: Re: [PATCH, v8 1/3] hrtimer: introduce effective timer slack Message-ID: <20110304122035.GB14933@shutemov.name> References: <1299161947-9903-1-git-send-email-kirill@shutemov.name> <1299161947-9903-2-git-send-email-kirill@shutemov.name> <201103031732.28964.arnd@arndb.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <201103031732.28964.arnd@arndb.de> User-Agent: Mutt/1.5.20 (2010-08-04) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Mar 03, 2011 at 05:32:28PM +0100, Arnd Bergmann wrote: > On Thursday 03 March 2011, Kirill A. Shutsemov wrote: > > task_get_effective_timer_slack() returns timer slack value to be used > > to configure per-task timers. It can be equal or higher than task's > > timer slack value. > > > > For now task_get_effective_timer_slack() returns timer_slack_ns of the > > task. Timer slack cgroup controller will implement a bit more > > sophisticated logic. > > Some time ago, there was a discussion about a method for automatically > determining timer slack values, and I think nobody ever implemented it. > > The idea was to penalize tasks that have timers expiring a lot, typically > a sign of programs that were not written with power consumption in mind. > > I think that could be nicely combined with your patch. Instead of setting > the effective timer slack for the entire control group, you could > set a target value that is applied only to tasks that have their timers > expire frequently. When a timer expires for a task, you increase the > effective slack up to the maximum, and when you set up a timer, you > decrease it again. The amount by which the effective slack gets changed > can depend on how long ago the last timer expired. > > Does this make sense to you? Hm.. I think we can add this kind of functionality to timer slack cgroup as a policy/stratagy later. I want to integrate basic functionality to kernel first. -- Kirill A. Shutemov