Re: [patch v8 4/9] sched: fix slept time double counting in enqueue entity

From: Lei Wen
Date: Wed Jun 19 2013 - 21:43:51 EST


Hi Alex,

On Mon, Jun 17, 2013 at 11:41 PM, Alex Shi <alex.shi@xxxxxxxxx> wrote:
> On 06/17/2013 07:51 PM, Paul Turner wrote:
>> Can you add something like:
>>
>> + /*
>> + * Task re-woke on same cpu (or else
>> migrate_task_rq_fair()
>> + * would have made count negative); we must be careful
>> to avoid
>> + * double-accounting blocked time after synchronizing
>> decays.
>> + */
>>
>>
>> Reviewed-by: Paul Turner <pjt@xxxxxxxxxx>
>
> thanks for review!
> ---
>
> From 24d9b43e7a269e6ffee5b874d39812b83812a809 Mon Sep 17 00:00:00 2001
> From: Alex Shi <alex.shi@xxxxxxxxx>
> Date: Thu, 9 May 2013 10:54:13 +0800
> Subject: [PATCH] sched: fix slept time double counting in enqueue entity
>
> The wakeuped migrated task will __synchronize_entity_decay(se); in
> migrate_task_fair, then it needs to set

Should be migrate_task_rq_fair, right? :)

> `se->avg.last_runnable_update -= (-se->avg.decay_count) << 20'
> before update_entity_load_avg, in order to avoid slept time is updated
> twice for se.avg.load_avg_contrib in both __syncchronize and
> update_entity_load_avg.
>
> but if the slept task is waked up from self cpu, it miss the
> last_runnable_update before update_entity_load_avg(se, 0, 1), then the
> slept time was used twice in both functions.
> So we need to remove the double slept time counting.
>
> Signed-off-by: Alex Shi <alex.shi@xxxxxxxxx>
> Reviewed-by: Paul Turner <pjt@xxxxxxxxxx>
> ---
> kernel/sched/fair.c | 8 +++++++-
> 1 files changed, 7 insertions(+), 1 deletions(-)
>
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index df5b8a9..1e5a5e6 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -1571,7 +1571,13 @@ static inline void enqueue_entity_load_avg(struct cfs_rq *cfs_rq,
> }
> wakeup = 0;
> } else {
> - __synchronize_entity_decay(se);
> + /*
> + * Task re-woke on same cpu (or else migrate_task_rq_fair()
> + * would have made count negative); we must be careful to avoid
> + * double-accounting blocked time after synchronizing decays.
> + */
> + se->avg.last_runnable_update += __synchronize_entity_decay(se)
> + << 20;

I'm kind of getting confused at here, since migrate_task_rq_fair use
below equation to
avoid sleep time be accounted for twice.
`se->avg.last_runnable_update -= (-se->avg.decay_count) << 20'

Why here we need to use "+=", which the reversed version comparing
with previous?

Thanks,
Lei

> }
>
> /* migrated tasks did not contribute to our blocked load */
> --
> 1.7.5.4
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/