Re: [PATCH V2] sched: fair: Use the earliest break even

[Valentin Schneider]

Hi Daniel,

One more comment on the break even itself, ignoring the rest:

On Wed, Mar 11 2020, Daniel Lezcano wrote:
> diff --git a/kernel/sched/idle.c b/kernel/sched/idle.c
> index b743bf38f08f..3342e7bae072 100644
> --- a/kernel/sched/idle.c
> +++ b/kernel/sched/idle.c
> @@ -19,7 +19,13 @@ extern char __cpuidle_text_start[], __cpuidle_text_end[];
>   */
>  void sched_idle_set_state(struct cpuidle_state *idle_state)
>  {
> -	idle_set_state(this_rq(), idle_state);
> +	struct rq *rq = this_rq();
> +
> +	idle_set_state(rq, idle_state);
> +
> +	if (idle_state)
> +		idle_set_break_even(rq, ktime_get_ns() +
> +				    idle_state->exit_latency_ns);

I'm not sure I follow why we go for entry time + exit latency. If this
is based on the minimum residency, shouldn't this be something depending
on the entry latency? i.e. something like

  break_even = now + entry_latency + idling_time
                     \_________________________/
                            min-residency

or am I missing something?

>  }
>
>  static int __read_mostly cpu_idle_force_poll;