Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756542AbcCaJOi (ORCPT ); Thu, 31 Mar 2016 05:14:38 -0400 Received: from casper.infradead.org ([85.118.1.10]:43007 "EHLO casper.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756212AbcCaJOd (ORCPT ); Thu, 31 Mar 2016 05:14:33 -0400 Date: Thu, 31 Mar 2016 11:14:26 +0200 From: Peter Zijlstra To: Steve Muckle Cc: Ingo Molnar , "Rafael J. Wysocki" , linux-kernel@vger.kernel.org, linux-pm@vger.kernel.org, Vincent Guittot , Morten Rasmussen , Dietmar Eggemann , Juri Lelli , Patrick Bellasi , Michael Turquette , Byungchul Park Subject: Re: [RFC PATCH] sched/fair: call cpufreq hook in additional paths Message-ID: <20160331091426.GE11035@twins.programming.kicks-ass.net> References: <1458858367-2831-1-git-send-email-smuckle@linaro.org> <20160331075951.GG3408@twins.programming.kicks-ass.net> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20160331075951.GG3408@twins.programming.kicks-ass.net> User-Agent: Mutt/1.5.21 (2012-12-30) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2783 Lines: 75 On Thu, Mar 31, 2016 at 09:59:51AM +0200, Peter Zijlstra wrote: > > - passing an argument into attach_entity_load_avg() to indicate > > whether calling the cpufreq hook is necessary > > > > Both of these are ugly in their own way but would avoid a runtime > > cost. Opinions welcome. > > Lemme see what this would look like while I throw the below into the bit > bucket. OK, so the below looks a lot more sane; and has the surprising benefit of actually shrinking the text size.. 43675 1226 24 44925 af7d defconfig-build/kernel/sched/fair.o.base 43723 1226 24 44973 afad defconfig-build/kernel/sched/fair.o.patch 43595 1226 24 44845 af2d defconfig-build/kernel/sched/fair.o.patch+ --- --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -2851,7 +2851,8 @@ static inline void cfs_rq_util_change(st } /* Group cfs_rq's load_avg is used for task_h_load and update_cfs_share */ -static inline int update_cfs_rq_load_avg(u64 now, struct cfs_rq *cfs_rq) +static inline int +update_cfs_rq_load_avg(u64 now, struct cfs_rq *cfs_rq, bool update_freq) { struct sched_avg *sa = &cfs_rq->avg; int decayed, removed_load = 0, removed_util = 0; @@ -2878,7 +2879,7 @@ static inline int update_cfs_rq_load_avg cfs_rq->load_last_update_time_copy = sa->last_update_time; #endif - if (decayed || removed_util) + if (update_freq && (decayed || removed_util)) cfs_rq_util_change(cfs_rq); return decayed || removed_load; @@ -2900,7 +2901,7 @@ static inline void update_load_avg(struc se->on_rq * scale_load_down(se->load.weight), cfs_rq->curr == se, NULL); - if (update_cfs_rq_load_avg(now, cfs_rq) && update_tg) + if (update_cfs_rq_load_avg(now, cfs_rq, true) && update_tg) update_tg_load_avg(cfs_rq, 0); } @@ -2962,7 +2963,7 @@ enqueue_entity_load_avg(struct cfs_rq *c cfs_rq->curr == se, NULL); } - decayed = update_cfs_rq_load_avg(now, cfs_rq); + decayed = update_cfs_rq_load_avg(now, cfs_rq, !migrated); cfs_rq->runnable_load_avg += sa->load_avg; cfs_rq->runnable_load_sum += sa->load_sum; @@ -6170,7 +6171,7 @@ static void update_blocked_averages(int if (throttled_hierarchy(cfs_rq)) continue; - if (update_cfs_rq_load_avg(cfs_rq_clock_task(cfs_rq), cfs_rq)) + if (update_cfs_rq_load_avg(cfs_rq_clock_task(cfs_rq), cfs_rq, true)) update_tg_load_avg(cfs_rq, 0); } raw_spin_unlock_irqrestore(&rq->lock, flags); @@ -6231,7 +6232,7 @@ static inline void update_blocked_averag raw_spin_lock_irqsave(&rq->lock, flags); update_rq_clock(rq); - update_cfs_rq_load_avg(cfs_rq_clock_task(cfs_rq), cfs_rq); + update_cfs_rq_load_avg(cfs_rq_clock_task(cfs_rq), cfs_rq, true); raw_spin_unlock_irqrestore(&rq->lock, flags); }