Re: [PATCH v3 1/1] kthread: do not preempt current task if it is going to call schedule()

[Steven Rostedt <rostedt@goodmis.org>]

On Fri,  6 Mar 2020 15:01:33 +0800
<cl@rock-chips.com> wrote:

> From: Liang Chen <cl@rock-chips.com>
> 
> when we create a kthread with ktrhead_create_on_cpu(),the child thread
> entry is ktread.c:ktrhead() which will be preempted by the parent after
> call complete(done) while schedule() is not called yet,then the parent
> will call wait_task_inactive(child) but the child is still on the runqueue,
> so the parent will schedule_hrtimeout() for 1 jiffy,it will waste a lot of
> time,especially on startup.
> 
>   parent                             child
> ktrhead_create_on_cpu()
>   wait_fo_completion(&done) -----> ktread.c:ktrhead()
>                              |----- complete(done);--wakeup and preempted by parent
>  kthread_bind() <------------|  |-> schedule();--dequeue here
>   wait_task_inactive(child)     |
>    schedule_hrtimeout(1 jiffy) -|
> 
> So we hope the child just wakeup parent but not preempted by parent, and the
> child is going to call schedule() soon,then the parent will not call
> schedule_hrtimeout(1 jiffy) as the child is already dequeue.
> 
> The same issue for ktrhead_park()&&kthread_parkme().
> This patch can save 120ms on rk312x startup with CONFIG_HZ=300.
> 
> Signed-off-by: Liang Chen <cl@rock-chips.com>
> ---
>  kernel/kthread.c | 17 +++++++++++++++--
>  1 file changed, 15 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/kthread.c b/kernel/kthread.c
> index b262f47046ca..bfbfa481be3a 100644
> --- a/kernel/kthread.c
> +++ b/kernel/kthread.c
> @@ -199,8 +199,15 @@ static void __kthread_parkme(struct kthread *self)
>  		if (!test_bit(KTHREAD_SHOULD_PARK, &self->flags))
>  			break;
>  
> +		/*
> +		 * Thread is going to call schedule(), do not preempt it,
> +		 * or the caller of kthread_park() may spend more time in
> +		 * wait_task_inactive().
> +		 */
> +		preempt_disable();
>  		complete(&self->parked);

I first was concerned that this could break PREEMPT_RT, as complete() calls
spin_locks() which are turned into sleeping locks when PREEMPT_RT is
enabled. But looking at the latest PREEMPT_RT patch, it appears that it
converts the locks in complete() into raw_spin_locks (and using swake).

I don't see any other issue with this patch.

Reviewed-by: Steven Rostedt (VMware) <rostedt@goodmis.org>

-- Steve

> -		schedule();
> +		schedule_preempt_disabled();
> +		preempt_enable();
>  	}
>  	__set_current_state(TASK_RUNNING);
>  }
> @@ -245,8 +252,14 @@ static int kthread(void *_create)
>  	/* OK, tell user we're spawned, wait for stop or wakeup */
>  	__set_current_state(TASK_UNINTERRUPTIBLE);
>  	create->result = current;
> +	/*
> +	 * Thread is going to call schedule(), do not preempt it,
> +	 * or the creator may spend more time in wait_task_inactive().
> +	 */
> +	preempt_disable();
>  	complete(done);
> -	schedule();
> +	schedule_preempt_disabled();
> +	preempt_enable();
>  
>  	ret = -EINTR;
>  	if (!test_bit(KTHREAD_SHOULD_STOP, &self->flags)) {