Re: [PATCH] tty: Only wakeup the line discipline idle queue when queue is active

[Jiri Slaby <jslaby@suse.cz>]

On 12/18/2012 03:48 PM, Ivo Sieben wrote:
> Before waking up the tty line discipline idle queue first check if the queue is
> active (non empty). This prevents unnecessary entering the critical section in
> the wake_up() function and therefore avoid needless scheduling overhead on a
> PREEMPT_RT system caused by two processes being in the same critical section.
> 
> Signed-off-by: Ivo Sieben <meltedpianoman@gmail.com>
> ---
> 
>  Remark:
>  This patch has kind of a long history... I first tried to prevent the critical
>  section in the wakeup() function itself by a change in the scheduler. But after
>  review remarks from Oleg Nesterov it turned out that using the
>  waitqueue_active() was a much nicer way to prevent it. See also
>  https://lkml.org/lkml/2012/10/25/159
> 
>  drivers/tty/tty_ldisc.c |    4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/tty/tty_ldisc.c b/drivers/tty/tty_ldisc.c
> index c578229..e96d187 100644
> --- a/drivers/tty/tty_ldisc.c
> +++ b/drivers/tty/tty_ldisc.c
> @@ -64,7 +64,9 @@ static void put_ldisc(struct tty_ldisc *ld)
>  		return;
>  	}
>  	raw_spin_unlock_irqrestore(&tty_ldisc_lock, flags);
> -	wake_up(&ld->wq_idle);
> +
> +	if (waitqueue_active(&ld->wq_idle))
> +		wake_up(&ld->wq_idle);

Looks good, but I would prefer the layer to provide us with
wake_up_if_active...

-- 
js
suse labs