Re: [PATCH v4 1/7] sched: Unify runtime accounting across classes

[Steven Rostedt]

On Thu, 31 Aug 2023 22:28:52 +0200
Daniel Bristot de Oliveira <bristot@xxxxxxxxxx> wrote:

As I have a vested interest in this work, I started a deep dive into the
code.

> --- a/kernel/sched/deadline.c
> +++ b/kernel/sched/deadline.c
> @@ -1299,9 +1299,8 @@ static void update_curr_dl(struct rq *rq)
>  {
>  	struct task_struct *curr = rq->curr;
>  	struct sched_dl_entity *dl_se = &curr->dl;
> -	u64 delta_exec, scaled_delta_exec;
> +	s64 delta_exec, scaled_delta_exec;
>  	int cpu = cpu_of(rq);
> -	u64 now;
>  
>  	if (!dl_task(curr) || !on_dl_rq(dl_se))
>  		return;
> @@ -1314,21 +1313,13 @@ static void update_curr_dl(struct rq *rq)
>  	 * natural solution, but the full ramifications of this
>  	 * approach need further study.
>  	 */
> -	now = rq_clock_task(rq);
> -	delta_exec = now - curr->se.exec_start;
> -	if (unlikely((s64)delta_exec <= 0)) {
> +	delta_exec = update_curr_common(rq);

I have to say, mapping the update_curr_common() to the removed code below,
wasn't as easy as I thought it would be, as the above function is broken up
slightly differently.

But the conclusion does appear to be pretty much the same.

Reviewed-by: Steven Rostedt (Google) <rostedt@xxxxxxxxxxx>

-- Steve


> +	if (unlikely(delta_exec <= 0)) {
>  		if (unlikely(dl_se->dl_yielded))
>  			goto throttle;
>  		return;
>  	}
>  
> -	schedstat_set(curr->stats.exec_max,
> -		      max(curr->stats.exec_max, delta_exec));
> -
> -	trace_sched_stat_runtime(curr, delta_exec, 0);
> -
> -	update_current_exec_runtime(curr, now, delta_exec);
> -
>  	if (dl_entity_is_special(dl_se))
>  		return;
>