---

yaml
---
r: 288754
b: refs/heads/master
c: 1c4dd99
h: refs/heads/master
v: v3

[refs]

@@ -1,2 +1,2 @@
 ---
-refs/heads/master: 63b2001169e75cd71e917ec953fdab572e3f944a
+refs/heads/master: 1c4dd99bed5f6f70932bf8dacdd54d04a2619778

trunk/kernel/sched/core.c

@@ -3779,6 +3779,24 @@ void rt_mutex_setprio(struct task_struct *p, int prio)
 
 	rq = __task_rq_lock(p);
 
+	/*
+	 * Idle task boosting is a nono in general. There is one
+	 * exception, when PREEMPT_RT and NOHZ is active:
+	 *
+	 * The idle task calls get_next_timer_interrupt() and holds
+	 * the timer wheel base->lock on the CPU and another CPU wants
+	 * to access the timer (probably to cancel it). We can safely
+	 * ignore the boosting request, as the idle CPU runs this code
+	 * with interrupts disabled and will complete the lock
+	 * protected section without being interrupted. So there is no
+	 * real need to boost.
+	 */
+	if (unlikely(p == rq->idle)) {
+		WARN_ON(p != rq->curr);
+		WARN_ON(p->pi_blocked_on);
+		goto out_unlock;
+	}
+
 	trace_sched_pi_setprio(p, prio);
 	oldprio = p->prio;
 	prev_class = p->sched_class;
@@ -3802,11 +3820,10 @@ void rt_mutex_setprio(struct task_struct *p, int prio)
 		enqueue_task(rq, p, oldprio < prio ? ENQUEUE_HEAD : 0);
 
 	check_class_changed(rq, p, prev_class, oldprio);
+out_unlock:
 	__task_rq_unlock(rq);
 }
-
 #endif
-
 void set_user_nice(struct task_struct *p, long nice)
 {
 	int old_prio, delta, on_rq;