Re: [PATCH v4] signal: Let tasks cache one sigqueue struct.

[Hillf Danton <hdanton@sina.com>]

On Thu, 6 Apr 2023 22:47:21 +0200 Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> The sigqueue caching originated in the PREEMPT_RT tree. A few of the
> applications, that were ported to Linux, were ported from OS-9. Sending
> notifications from one task to another via a signal was a common
> communication model there and so the applications are heavy signal
> users. Removing the allocation reduces the critical path, avoids locks
> and so lowers the maximal latency of the task while sending a signal.

This is needed only if slab does not work for you.
Why not make it work better instead of peeling slab one more off just to
make signal/rt richer?
> 
> --- a/kernel/signal.c
> +++ b/kernel/signal.c
> @@ -432,7 +432,18 @@ static struct sigqueue *
>  		return NULL;
> =20
>  	if (override_rlimit || likely(sigpending <=3D task_rlimit(t, RLIMIT_SIGPE=
> NDING))) {
> -		q =3D kmem_cache_alloc(sigqueue_cachep, gfp_flags);
> +
> +		if (!sigqueue_flags) {
> +			struct sighand_struct *sighand =3D t->sighand;
> +
> +			lockdep_assert_held(&sighand->siglock);
> +			if (sighand->sigqueue_cache) {
> +				q =3D sighand->sigqueue_cache;
> +				sighand->sigqueue_cache =3D NULL;
> +			}
> +		}
> +		if (!q)
> +			q =3D kmem_cache_alloc(sigqueue_cachep, gfp_flags);
>  	} else {
>  		print_dropped_signal(sig);
>  	}