Re: [PATCH] sched/fair: Optimize select_idle_cpu

[Valentin Schneider <valentin.schneider@arm.com>]

On 12/12/2019 15:24, Peter Zijlstra wrote:
> On Thu, Dec 12, 2019 at 10:41:02PM +0800, Cheng Jian wrote:
> 
>> Fixes: 1ad3aaf3fcd2 ("sched/core: Implement new approach to scale select_idle_cpu()")
> 
> The 'funny' thing is that select_idle_core() actually does the right
> thing.
> 
> Copying that should work:
> 
> 
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index 08a233e97a01..416d574dcebf 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -5828,6 +5837,7 @@ static inline int select_idle_smt(struct task_struct *p, int target)
>   */
>  static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, int target)
>  {
> +	struct cpumask *cpus = this_cpu_cpumask_var_ptr(select_idle_mask);
>  	struct sched_domain *this_sd;
>  	u64 avg_cost, avg_idle;
>  	u64 time, cost;
> @@ -5859,11 +5869,11 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, int t
>  
>  	time = cpu_clock(this);
>  
> -	for_each_cpu_wrap(cpu, sched_domain_span(sd), target) {
> +	cpumask_and(cpus, sched_domain_span(sd), p->cpus_ptr);
> +
> +	for_each_cpu_wrap(cpu, cpus, target) {
>  		if (!--nr)
>  			return si_cpu;
> -		if (!cpumask_test_cpu(cpu, p->cpus_ptr))
> -			continue;
>  		if (available_idle_cpu(cpu))
>  			break;
>  		if (si_cpu == -1 && sched_idle_cpu(cpu))
> 

That looks sane enough. I'd only argue the changelog should directly point
out that the issue is we consume some CPUs out of 'nr' that are not allowed
for the task and thus waste our attempts.