lkml.org 
[lkml]   [2021]   [Oct]   [11]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: [PATCH] sched/fair: Use this_sd->weight to calculate span_avg
On Sun, 10 Oct 2021 02:10:55 +0800
Tao Zhou <tao.zhou@linux.dev> wrote:

> avg_idle, avg_cost got from this_rq and this_sd. I think
> use this_sd->weight to calculate and estimate the number
> of loop cpus in the target domain.

If that's the case, then shouldn't the CPUs to be checked come from this_sd
as well? I mean, at the beginning of the function we have:

this_sd = rcu_dereference(*this_cpu_ptr(&sd_llc));
if (!this_sd)
return -1;

cpumask_and(cpus, sched_domain_span(sd), p->cpus_ptr);

Where "cpus" comes from sd, and not this_sd.


> ---
> kernel/sched/fair.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index f6a05d9b5443..7fab7b70814c 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -6300,7 +6300,7 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, bool
> avg_idle = this_rq->wake_avg_idle;
> avg_cost = this_sd->avg_scan_cost + 1;
>
> - span_avg = sd->span_weight * avg_idle;
> + span_avg = this_sd->span_weight * avg_idle;
> if (span_avg > 4*avg_cost)
> nr = div_u64(span_avg, avg_cost);
> else


And after this code, the nr that is determined from the above, is for
limiting the looping over those CPUs from sd, not this_sd:

for_each_cpu_wrap(cpu, cpus, target + 1) {
if (has_idle_core) {
i = select_idle_core(p, cpu, cpus, &idle_cpu);
if ((unsigned int)i < nr_cpumask_bits)
return i;

} else {
if (!--nr)
return -1;
idle_cpu = __select_idle_cpu(cpu, p);
if ((unsigned int)idle_cpu < nr_cpumask_bits)
break;
}
}

I'm guessing there's nothing wrong here. But, I don't fully understand it
myself.

-- Steve
\
 
 \ /
  Last update: 2021-10-11 16:59    [W:0.115 / U:0.208 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site