Re: [RFC PATCH] sched/fair: Choose the CPU where short task is running during wake up

[Chen Yu <yu.c.chen@intel.com>]

On 2022-09-15 at 10:10:25 -0700, Tim Chen wrote:
> On Fri, 2022-09-16 at 00:54 +0800, Chen Yu wrote:
> > 
> > +/*
> > + * If a task switches in and then voluntarily relinquishes the
> > + * CPU quickly, it is regarded as a short running task.
> > + * sysctl_sched_min_granularity is chosen as the threshold,
> > + * as this value is the minimal slice if there are too many
> > + * runnable tasks, see __sched_period().
> > + */
> > +static int is_short_task(struct task_struct *p)
> > +{
> > +	return (p->se.sum_exec_runtime <=
> > +		(p->nvcsw * sysctl_sched_min_granularity));
> > +}
> > +
> >  /*
> >   * The purpose of wake_affine() is to quickly determine on which CPU we can run
> >   * soonest. For the purpose of speed we only consider the waking and previous
> > @@ -6050,7 +6063,8 @@ wake_affine_idle(int this_cpu, int prev_cpu, int sync)
> >  	if (available_idle_cpu(this_cpu) && cpus_share_cache(this_cpu, prev_cpu))
> >  		return available_idle_cpu(prev_cpu) ? prev_cpu : this_cpu;
> >  
> > -	if (sync && cpu_rq(this_cpu)->nr_running == 1)
> > +	if ((sync && cpu_rq(this_cpu)->nr_running == 1) ||
> > +	    is_short_task(cpu_curr(this_cpu)))
> >  		return this_cpu;
> >  
> >  	if (available_idle_cpu(prev_cpu))
> > @@ -6434,6 +6448,21 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, bool
> >  			/* overloaded LLC is unlikely to have idle cpu/core */
> >  			if (nr == 1)
> >  				return -1;
> > +
> > +			/*
> > +			 * If nr is smaller than 60% of llc_weight, it
> > +			 * indicates that the util_avg% is higher than 50%.
> > +			 * This is calculated by SIS_UTIL in
> > +			 * update_idle_cpu_scan(). The 50% util_avg indicates
> > +			 * a half-busy LLC domain. System busier than this
> > +			 * level could lower its bar to choose a compromised
> > +			 * "idle" CPU. If the waker on target CPU is a short
> > +			 * task and the wakee is also a short task, pick
> > +			 * target directly.
> > +			 */
> > +			if (!has_idle_core && (5 * nr < 3 * sd->span_weight) &&
> > +			    is_short_task(p) && is_short_task(cpu_curr(target)))
> 
> Should we check if target's rq's nr_running is 1, and if there's pending waking
> task before picking it?
>
Yes we can consider the two factors, then the criteria to pick up a target CPU
would be more strict. After taking nr_running and the pending wakeup request into
consideration, I think it would be a variant of WF_SYNC and we can get rid of
'system should be busy' restriction. I'll do some test in this direction.

thanks,
Chenyu
> > +				return target;
> >  		}
> >  	}
> >  
> 
> Thanks.
> 
> Tim
>