[PATCH tip 2/2] sched/core: Fix endless loop in pick_next_task()

[Kirill Tkhai]

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 <ktkhai@xxxxxxxxxxxxx>
CC: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
CC: Ingo Molnar <mingo@xxxxxxxxxx>
---
 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 */



--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/