From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754111AbYJCRsH (ORCPT ); Fri, 3 Oct 2008 13:48:07 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752935AbYJCRrz (ORCPT ); Fri, 3 Oct 2008 13:47:55 -0400 Received: from ms01.sssup.it ([193.205.80.99]:45395 "EHLO sssup.it" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1752332AbYJCRrz (ORCPT ); Fri, 3 Oct 2008 13:47:55 -0400 X-Greylist: delayed 3599 seconds by postgrey-1.27 at vger.kernel.org; Fri, 03 Oct 2008 13:47:54 EDT Subject: [PATCH] sched_rt.c: resch needed in rt_rq_enqueue() for the root rt_rq From: Dario Faggioli To: Linux Kernel Mailing List Cc: Michael Trimarchi , Peter Zijlstra , Ingo Molnar , Thomas Gleixner Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="=-5/ttfLF2kyKWjOHrVugj" Date: Fri, 03 Oct 2008 17:40:46 +0200 Message-Id: <1223048447.6714.46.camel@Palanthas> Mime-Version: 1.0 X-Mailer: Evolution 2.22.3.1 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org --=-5/ttfLF2kyKWjOHrVugj Content-Type: text/plain Content-Transfer-Encoding: quoted-printable Hi all, While working on the new version of the code for SCHED_SPORADIC I noticed something strange in the present throttling mechanism. More specifically in the throttling timer handler in sched_rt.c (do_sched_rt_period_timer()) and in rt_rq_enqueue(). The problem is that, when unthrottling a runqueue, rt_rq_enqueue() only asks for rescheduling if the runqueue has a sched_entity associated to it (i.e., rt_rq->rt_se !=3D NULL). Now, if the runqueue is the root rq (which has a rt_se =3D NULL) rescheduling does not take place, and it is delayed to some undefined instant in the future. This imply some random bandwidth usage by the RT tasks under throttling. For instance, setting rt_runtime_us/rt_period_us =3D 950ms/1000ms an RT task will get less than 95%. In our tests we got something varying between 70% to 95%. Using smaller time values, e.g., 95ms/100ms, things are even worse, and I can see values also going down to 20-25%!! The tests we performed are simply running 'yes' as a SCHED_FIFO task, and checking the CPU usage with top, but we can investigate thoroughly if you think it is needed. Things go much better, for us, with the attached patch... Don't know if it is the best approach, but it solved the issue for us. Best Regards, Dario Faggioli --- Signed-off-by: Dario Faggioli Signed-off-by: Michael Trimarchi --- diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index 1113157..37f0721 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -102,12 +102,12 @@ static void dequeue_rt_entity(struct sched_rt_entity = *rt_se); =20 static void sched_rt_rq_enqueue(struct rt_rq *rt_rq) { + struct task_struct *curr =3D rq_of_rt_rq(rt_rq)->curr; struct sched_rt_entity *rt_se =3D rt_rq->rt_se; =20 - if (rt_se && !on_rt_rq(rt_se) && rt_rq->rt_nr_running) { - struct task_struct *curr =3D rq_of_rt_rq(rt_rq)->curr; - - enqueue_rt_entity(rt_se); + if (rt_rq->rt_nr_running) { + if (rt_se && !on_rt_rq(rt_se)) + enqueue_rt_entity(rt_se); if (rt_rq->highest_prio < curr->prio) resched_task(curr); } --=20 <> (Raistlin Majere, DragonLance Chronicles -Dragons of Spring Drawning-) ---------------------------------------------------------------------- Dario Faggioli GNU/Linux Registered User: #340657 Web: http://www.linux.it/~raistlin Blog: http://blog.linux.it/raistlin SIP Account: dario.faggioli@sipproxy.wengo.fr or raistlin@ekiga.net Jabber Account: dario.faggioli@jabber.org/WengoPhone GnuPG Key ID: 4DC83AC4 GnuPG Key Fingerprint: 2A78 AD5D B9CF A082 0836 08AD 9385 DA04 4DC8 3AC4 --=-5/ttfLF2kyKWjOHrVugj Content-Type: application/pgp-signature; name=signature.asc Content-Description: This is a digitally signed message part -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) iD8DBQBI5jz+k4XaBE3IOsQRAq9gAJ9EadnhW3Q7pmBgfhOF3mQNvIahnwCgihQY 3kVH6t+dMYHgruVdzUKUq6Q= =WMst -----END PGP SIGNATURE----- --=-5/ttfLF2kyKWjOHrVugj--