Re: [RESEND][PATCH v18 6/8] sched: Add an initial sketch of the find_proxy_task() function

[Peter Zijlstra <peterz@infradead.org>]

On Mon, Jul 07, 2025 at 08:43:53PM +0000, John Stultz wrote:

> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 853157b27f384..dc82d9b8bee2c 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -6614,7 +6614,8 @@ pick_next_task(struct rq *rq, struct task_struct *prev, struct rq_flags *rf)
>   * Otherwise marks the task's __state as RUNNING
>   */
>  static bool try_to_block_task(struct rq *rq, struct task_struct *p,
> -			      unsigned long *task_state_p)
> +			      unsigned long *task_state_p,
> +			      bool deactivate_cond)
>  {
>  	unsigned long task_state = *task_state_p;
>  	int flags = DEQUEUE_NOCLOCK;
> @@ -6625,6 +6626,9 @@ static bool try_to_block_task(struct rq *rq, struct task_struct *p,
>  		return false;
>  	}
>  
> +	if (!deactivate_cond)
> +		return false;
> +
>  	p->sched_contributes_to_load =
>  		(task_state & TASK_UNINTERRUPTIBLE) &&
>  		!(task_state & TASK_NOLOAD) &&

I'm struggling with this; @deactivate_cond doesn't seem to adequately
cover what it actually does.

So far what it seems to do is when true, don't block. It still does the
signal thing -- but I can't tell if that is actually required or not.

Would 'should_block' be a better name? And maybe stick a little
something in the comment above try_to_block_task() or near the:

	if (!should_block)
		return false;

lines about why the signal bits are important to have done.

> @@ -6648,6 +6652,89 @@ static bool try_to_block_task(struct rq *rq, struct task_struct *p,
>  	return true;
>  }
>  
> +#ifdef CONFIG_SCHED_PROXY_EXEC
> +static inline struct task_struct *proxy_resched_idle(struct rq *rq)
> +{
> +	put_prev_set_next_task(rq, rq->donor, rq->idle);
> +	rq_set_donor(rq, rq->idle);
> +	set_tsk_need_resched(rq->idle);
> +	return rq->idle;
> +}

Nothing cares about the return value.