Re: [patch] sched/fair: delayed dequeued entity bias wakeup preempt short

[Qais Yousef <qyousef@layalina.io>]

On 04/10/26 16:48, Vincent Guittot wrote:
> When checking if a shorter slice entity can preempt curr, we compare it with
> the next entity to be picked but delayed dequeue entities can screw the
> decision whereas they will be dequeued when picking next entity.
> Dequeue them while checking for preemption as they will be dequeued anyway
> when picking next entity.
> 
> Test on snapdragon rb5:
> hackbench -T -p -l 16000000 -g 2 1> /dev/null &
> cyclictest -t 1 -i 2777 -D 63 --policy=fair --mlock  -h 20000 -q
> 
>                      tip/sched/core  tip/sched/core    +this patch
> cyclictest slice  (ms) (default)2.8             8               8
> hackbench slice   (ms) (default)2.8            20              20
> Total Samples          |    22559           22595           22683
> Average           (us) |      157              64( 59%)        59(  8%)
> Median (P50)      (us) |       57              57(  0%)        58(- 2%)
> 90th Percentile   (us) |       64              60(  6%)        60(  0%)
> 99th Percentile   (us) |     2407              67( 97%)        67(  0%)
> 99.9th Percentile (us) |     3400            2288( 33%)       727( 68%)
> Maximum           (us) |     5037            9252(-84%)      7461( 19%)

Max is still high. The clear buddies patch reduces it though, I assume the
combined effect of the two deals with that?