linux-kernel - Re: [RFC PATCH 1/3] sched/fair: Aggregate task utilization only on root cfs

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 2 Jun 2016 10:23:58 +0100
From:	Juri Lelli <juri.lelli@....com>
To:	Dietmar Eggemann <dietmar.eggemann@....com>
Cc:	Peter Zijlstra <peterz@...radead.org>,
	linux-kernel@...r.kernel.org,
	Vincent Guittot <vincent.guittot@...aro.org>,
	Ben Segall <bsegall@...gle.com>,
	Morten Rasmussen <morten.rasmussen@....com>,
	Yuyang Du <yuyang.du@...el.com>
Subject: Re: [RFC PATCH 1/3] sched/fair: Aggregate task utilization only on
 root cfs_rq

Hi,

minor comment below.

On 01/06/16 20:39, Dietmar Eggemann wrote:
> cpu utilization (cpu_util()) is defined as the cpu (original) capacity
> capped cfs_rq->avg->util_avg signal of the root cfs_rq.
> 
> With the current pelt version, the utilization of a task [en|de]queued
> on/from a cfs_rq, representing a task group other than the root task group
> on a cpu, is not immediately propagated down to the root cfs_rq.
> 
> This makes decisions based on cpu_util() for scheduling or cpu frequency
> settings less accurate in case tasks are running in task groups.
> 
> This patch aggregates the task utilization only on the root cfs_rq,
> essentially avoiding maintaining utilization for a se/cfs_rq representing
> task groups other than the root task group (!entity_is_task(se) and
> &rq_of(cfs_rq)->cfs != cfs_rq).
> 
> The additional if/else condition to set @update_util in
> __update_load_avg() is replaced in 'sched/fair: Change @running of
> __update_load_avg() to @update_util' by providing the information whether
> utilization has to be maintained via an argument to this function.
> 
> The additional requirements for the alignment of the last_update_time of a
> se and the root cfs_rq are handled by the patch 'sched/fair: Sync se with
> root cfs_rq'.
> 
> Signed-off-by: Dietmar Eggemann <dietmar.eggemann@....com>
> ---
>  kernel/sched/fair.c | 48 ++++++++++++++++++++++++++++++++++++------------
>  1 file changed, 36 insertions(+), 12 deletions(-)
> 
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index 218f8e83db73..212becd3708f 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -2705,6 +2705,7 @@ __update_load_avg(u64 now, int cpu, struct sched_avg *sa,
>  	u32 contrib;
>  	unsigned int delta_w, scaled_delta_w, decayed = 0;
>  	unsigned long scale_freq, scale_cpu;
> +	int update_util = 0;
>  
>  	delta = now - sa->last_update_time;
>  	/*
> @@ -2725,6 +2726,12 @@ __update_load_avg(u64 now, int cpu, struct sched_avg *sa,
>  		return 0;
>  	sa->last_update_time = now;
>  
> +	if (cfs_rq) {
> +		if (&rq_of(cfs_rq)->cfs == cfs_rq)

Maybe we can wrap this sort of checks in a static inline improving
readability?

Best,

- Juri