From mboxrd@z Thu Jan 1 00:00:00 1970 From: morten.rasmussen@arm.com (Morten Rasmussen) Date: Wed, 8 Oct 2014 12:13:26 +0100 Subject: [PATCH 4/7] sched: Track group sched_entity usage contributions In-Reply-To: References: <1412684017-16595-1-git-send-email-vincent.guittot@linaro.org> <1412684017-16595-5-git-send-email-vincent.guittot@linaro.org> Message-ID: <20141008111326.GB1788@e105550-lin.cambridge.arm.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Tue, Oct 07, 2014 at 09:15:39PM +0100, bsegall at google.com wrote: > Vincent Guittot writes: > > > From: Morten Rasmussen > > > > Adds usage contribution tracking for group entities. Unlike > > se->avg.load_avg_contrib, se->avg.utilization_avg_contrib for group > > entities is the sum of se->avg.utilization_avg_contrib for all entities on the > > group runqueue. It is _not_ influenced in any way by the task group > > h_load. Hence it is representing the actual cpu usage of the group, not > > its intended load contribution which may differ significantly from the > > utilization on lightly utilized systems. > > > Just noting that this version also has usage disappear immediately when > a task blocks, although it does what you probably want on migration. Yes, as it was discussed at Ksummit, we should include blocked usage. It gives a much more stable metric. > Also, group-ses don't ever use their running_avg_sum so it's kinda a > waste, but I'm not sure it's worth doing anything about. Yeah, but since we still have to maintain runnable_avg_sum for group-ses it isn't much that can be saved I think. Maybe avoid calling update_entity_load_avg() from set_next_entity() for group-ses, but I'm not sure how much it helps.