From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753990AbbI2APM (ORCPT ); Mon, 28 Sep 2015 20:15:12 -0400 Received: from mail-pa0-f50.google.com ([209.85.220.50]:34631 "EHLO mail-pa0-f50.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752887AbbI2APJ (ORCPT ); Mon, 28 Sep 2015 20:15:09 -0400 From: Steve Muckle Subject: Re: [RFCv5 PATCH 32/46] sched: Energy-aware wake-up task placement To: Leo Yan References: <1436293469-25707-1-git-send-email-morten.rasmussen@arm.com> <1436293469-25707-33-git-send-email-morten.rasmussen@arm.com> <20150902171131.GA12510@leoy-linaro> <55FBE8AF.3050200@arm.com> <55FEFD54.3010206@linaro.org> <20150920220324.GA20859@leoy-linaro> Cc: Dietmar Eggemann , Morten Rasmussen , "peterz@infradead.org" , "mingo@redhat.com" , "vincent.guittot@linaro.org" , "daniel.lezcano@linaro.org" , "yuyang.du@intel.com" , "mturquette@baylibre.com" , "rjw@rjwysocki.net" , Juri Lelli , "sgurrappadi@nvidia.com" , "pang.xunlei@zte.com.cn" , "linux-kernel@vger.kernel.org" , "linux-pm@vger.kernel.org" X-Enigmail-Draft-Status: N1110 Message-ID: <5609D80B.5060507@linaro.org> Date: Mon, 28 Sep 2015 17:15:07 -0700 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.2.0 MIME-Version: 1.0 In-Reply-To: <20150920220324.GA20859@leoy-linaro> Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 09/20/2015 03:03 PM, Leo Yan wrote: > In this case of CPU is running at fmax, it's true that > task_fits_capacity() will return true. But here i think > cpu_overutilized() also will return true, so that means scheduler will > go back to use CFS's old way for loading balance. Finally tasks also > will be spread into two clusters. Agreed that once the first cluster is overutilized, the load will definitely spread to both clusters. My concern though is that for this to occur, the first cluster will likely be pushed to a high OPP. For power (and even perhaps performance) spreading the load earlier may be better. Or not, my observation is really just that we're encoding policy here which ideally would be the result of calculations in the energy model.