Re: [PATCH v2] rcu-tasks: Stop RCU Tasks scanning tasks which record on dyntick-idle entry

[Paul E. McKenney]

On Fri, May 20, 2022 at 04:41:55PM +0800, Zqiang wrote:
> When the RCU Tasks scanning on-rq tasks, there is no need to scanning
> tasks which record on dyntick-idle entry, at this time, these tasks are
> not remain within an RCU Tasks read-side critical section.
> 
> This commit skip scanning tasks which record on dyntick-idle entry in
> rcu_tasks_pertask().
> 
> Reported-by: kernel test robot <lkp@xxxxxxxxx>
> Signed-off-by: Zqiang <qiang1.zhang@xxxxxxxxx>

You lost me with this one.  Doesn't the rcu_tasks_pertask() function
already exclude the idle tasks that might possibly be running in
dyntick-idle state in non-CONFIG_NO_HZ_FULL systems?

What am I missing here?

							Thanx, Paul

> ---
>  v1->v2:
>  fix build test ERROR due to CONFIG_TASKS_RCU=n
> 
>  kernel/rcu/tasks.h | 12 +++++++++---
>  1 file changed, 9 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/rcu/tasks.h b/kernel/rcu/tasks.h
> index f6459343e4b6..7cb0f922478e 100644
> --- a/kernel/rcu/tasks.h
> +++ b/kernel/rcu/tasks.h
> @@ -809,10 +809,17 @@ static void rcu_tasks_pregp_step(struct list_head *hop)
>  	synchronize_rcu();
>  }
>  
> +static bool task_is_on_dyntick_idle(struct task_struct *t)
> +{
> +	return IS_ENABLED(CONFIG_NO_HZ_FULL) && !is_idle_task(t) &&
> +				t->rcu_tasks_idle_cpu >= 0;
> +}
> +
>  /* Per-task initial processing. */
>  static void rcu_tasks_pertask(struct task_struct *t, struct list_head *hop)
>  {
> -	if (t != current && READ_ONCE(t->on_rq) && !is_idle_task(t)) {
> +	if (t != current && READ_ONCE(t->on_rq) && !is_idle_task(t) &&
> +				!task_is_on_dyntick_idle(t)) {
>  		get_task_struct(t);
>  		t->rcu_tasks_nvcsw = READ_ONCE(t->nvcsw);
>  		WRITE_ONCE(t->rcu_tasks_holdout, true);
> @@ -842,8 +849,7 @@ static void check_holdout_task(struct task_struct *t,
>  	if (!READ_ONCE(t->rcu_tasks_holdout) ||
>  	    t->rcu_tasks_nvcsw != READ_ONCE(t->nvcsw) ||
>  	    !READ_ONCE(t->on_rq) ||
> -	    (IS_ENABLED(CONFIG_NO_HZ_FULL) &&
> -	     !is_idle_task(t) && t->rcu_tasks_idle_cpu >= 0)) {
> +	    task_is_on_dyntick_idle(t)) {
>  		WRITE_ONCE(t->rcu_tasks_holdout, false);
>  		list_del_init(&t->rcu_tasks_holdout_list);
>  		put_task_struct(t);
> -- 
> 2.25.1
>