From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752955AbaCFJct (ORCPT ); Thu, 6 Mar 2014 04:32:49 -0500 Received: from relay.parallels.com ([195.214.232.42]:54861 "EHLO relay.parallels.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752781AbaCFJcJ (ORCPT ); Thu, 6 Mar 2014 04:32:09 -0500 Message-ID: <1394098321.19290.11.camel@tkhai> Subject: [PATCH tip 2/2] sched/core: Fix endless loop in pick_next_task() From: Kirill Tkhai To: CC: Peter Zijlstra , Ingo Molnar Date: Thu, 6 Mar 2014 13:32:01 +0400 Organization: Parallels Content-Type: text/plain; charset="UTF-8" X-Mailer: Evolution 3.8.5-2+b1 MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Originating-IP: [10.30.26.172] Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org 1)Single cpu machine case. When rq has only RT tasks, but no one of them can be picked because of throttling, we enter in endless loop. pick_next_task_{dl,rt} return NULL. In pick_next_task_fair() we permanently go to retry if (rq->nr_running != rq->cfs.h_nr_running) return RETRY_TASK; (rq->nr_running is not being decremented when rt_rq becomes throttled). No chances to unthrottle any rt_rq or to wake fair here, because of rq is locked permanently and interrupts are disabled. 2)In case of SMP this can cause a hang too. Although we unlock rq in idle_balance(), interrupts are still disabled. The solution is to check for available tasks in DL and RT classes instead of checking for sum. Signed-off-by: Kirill Tkhai CC: Peter Zijlstra CC: Ingo Molnar --- kernel/sched/fair.c | 4 +++- kernel/sched/rt.c | 10 ---------- kernel/sched/sched.h | 12 ++++++++++++ 3 files changed, 15 insertions(+), 11 deletions(-) diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index b956e70..10db4a8 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -6728,7 +6728,9 @@ static int idle_balance(struct rq *this_rq) out: /* Is there a task of a high priority class? */ - if (this_rq->nr_running != this_rq->cfs.h_nr_running) + if (this_rq->nr_running != this_rq->cfs.h_nr_running && + (this_rq->dl.dl_nr_running || + (this_rq->rt.rt_nr_running && !rt_rq_throttled(&this_rq->rt)))) pulled_task = -1; if (pulled_task) { diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c index f3cee0a..d8cdf16 100644 --- a/kernel/sched/rt.c +++ b/kernel/sched/rt.c @@ -470,11 +470,6 @@ static void sched_rt_rq_dequeue(struct rt_rq *rt_rq) dequeue_rt_entity(rt_se); } -static inline int rt_rq_throttled(struct rt_rq *rt_rq) -{ - return rt_rq->rt_throttled && !rt_rq->rt_nr_boosted; -} - static int rt_se_boosted(struct sched_rt_entity *rt_se) { struct rt_rq *rt_rq = group_rt_rq(rt_se); @@ -545,11 +540,6 @@ static inline void sched_rt_rq_dequeue(struct rt_rq *rt_rq) { } -static inline int rt_rq_throttled(struct rt_rq *rt_rq) -{ - return rt_rq->rt_throttled; -} - static inline const struct cpumask *sched_rt_period_mask(void) { return cpu_online_mask; diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index 378bff7..f2de7a1 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -423,6 +423,18 @@ struct rt_rq { #endif }; +#ifdef CONFIG_RT_GROUP_SCHED +static inline int rt_rq_throttled(struct rt_rq *rt_rq) +{ + return rt_rq->rt_throttled && !rt_rq->rt_nr_boosted; +} +#else +static inline int rt_rq_throttled(struct rt_rq *rt_rq) +{ + return rt_rq->rt_throttled; +} +#endif + /* Deadline class' related fields in a runqueue */ struct dl_rq { /* runqueue is an rbtree, ordered by deadline */