Re: [RESEND PATCH] sched/fair: consider RT/IRQ pressure in select_idle_sibling

[Rohit Jain <rohit.k.jain@oracle.com>]

On 02/05/2018 10:42 PM, Joel Fernandes wrote:
> On Tue, Jan 30, 2018 at 11:47 AM, Rohit Jain <rohit.k.jain@oracle.com> wrote:
> [...]
>>>> @@ -6102,7 +6107,8 @@ static int select_idle_core(struct task_struct *p,
>>>> struct sched_domain *sd, int
>>>>     */
>>>>    static int select_idle_smt(struct task_struct *p, struct sched_domain
>>>> *sd, int target)
>>>>    {
>>>> -       int cpu;
>>>> +       int cpu, rcpu = -1;
>>>> +       unsigned long max_cap = 0;
>>>>
>>>>           if (!static_branch_likely(&sched_smt_present))
>>>>                   return -1;
>>>> @@ -6110,11 +6116,13 @@ static int select_idle_smt(struct task_struct *p,
>>>> struct sched_domain *sd, int t
>>>>           for_each_cpu(cpu, cpu_smt_mask(target)) {
>>>>                   if (!cpumask_test_cpu(cpu, &p->cpus_allowed))
>>>>                           continue;
>>>> -               if (idle_cpu(cpu))
>>>> -                       return cpu;
>>>> +               if (idle_cpu(cpu) && (capacity_of(cpu) > max_cap)) {
>>>> +                       max_cap = capacity_of(cpu);
>>>> +                       rcpu = cpu;
>>> At the SMT level, do you need to bother with choosing best capacity
>>> among threads? If RT is eating into one of the SMT thread's underlying
>>> capacity, it would eat into the other's. Wondering what's the benefit
>>> of doing this here.
>>
>> Yes, you are right because of SD_SHARE_CPUCAPACITY, however the benefit
>> is that if don't do this check, we might end up picking a SMT thread
>> which has "high" RT/IRQ activity and be on the run queue for a while,
>> till the pull side can bail us out.
> Do your tests show a difference in results though with such change
> (for select_idle_smt)?

I don't have the numbers readily available, but I did see a measurable
difference with the select_idle_smt changes.

Thanks,
Rohit

>
> thanks,
>
> - Joel