Merge branch 'sched-hrtimers-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull scheduler updates from Thomas Gleixner:
 "This series of scheduler updates depends on sched/core and timers/core
  branches, which are already in your tree:

   - Scheduler balancing overhaul to plug a hard to trigger race which
     causes an oops in the balancer (Peter Zijlstra)

   - Lockdep updates which are related to the balancing updates (Peter
     Zijlstra)"

* 'sched-hrtimers-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched,lockdep: Employ lock pinning
  lockdep: Implement lock pinning
  lockdep: Simplify lock_release()
  sched: Streamline the task migration locking a little
  sched: Move code around
  sched,dl: Fix sched class hopping CBS hole
  sched, dl: Convert switched_{from, to}_dl() / prio_changed_dl() to balance callbacks
  sched,dl: Remove return value from pull_dl_task()
  sched, rt: Convert switched_{from, to}_rt() / prio_changed_rt() to balance callbacks
  sched,rt: Remove return value from pull_rt_task()
  sched: Allow balance callbacks for check_class_changed()
  sched: Use replace normalize_task() with __sched_setscheduler()
  sched: Replace post_schedule with a balance callback list

6 files changed, 587 insertions, 492 deletions

diff --git a/kernel/locking/lockdep.c b/kernel/locking/lockdep.c
index 456614136f1a..8acfbf773e06 100644
--- a/kernel/locking/lockdep.c
+++ b/kernel/locking/lockdep.c
@@ -3157,6 +3157,7 @@ static int __lock_acquire(struct lockdep_map *lock, unsigned int subclass,
 	hlock->waittime_stamp = 0;
 	hlock->holdtime_stamp = lockstat_clock();
 #endif
+	hlock->pin_count = 0;
 
 	if (check && !mark_irqflags(curr, hlock))
 		return 0;
@@ -3260,26 +3261,6 @@ print_unlock_imbalance_bug(struct task_struct *curr, struct lockdep_map *lock,
 	return 0;
 }
 
-/*
- * Common debugging checks for both nested and non-nested unlock:
- */
-static int check_unlock(struct task_struct *curr, struct lockdep_map *lock,
-			unsigned long ip)
-{
-	if (unlikely(!debug_locks))
-		return 0;
-	/*
-	 * Lockdep should run with IRQs disabled, recursion, head-ache, etc..
-	 */
-	if (DEBUG_LOCKS_WARN_ON(!irqs_disabled()))
-		return 0;
-
-	if (curr->lockdep_depth <= 0)
-		return print_unlock_imbalance_bug(curr, lock, ip);
-
-	return 1;
-}
-
 static int match_held_lock(struct held_lock *hlock, struct lockdep_map *lock)
 {
 	if (hlock->instance == lock)
@@ -3376,31 +3357,35 @@ found_it:
 }
 
 /*
- * Remove the lock to the list of currently held locks in a
- * potentially non-nested (out of order) manner. This is a
- * relatively rare operation, as all the unlock APIs default
- * to nested mode (which uses lock_release()):
+ * Remove the lock to the list of currently held locks - this gets
+ * called on mutex_unlock()/spin_unlock*() (or on a failed
+ * mutex_lock_interruptible()).
+ *
+ * @nested is an hysterical artifact, needs a tree wide cleanup.
  */
 static int
-lock_release_non_nested(struct task_struct *curr,
-			struct lockdep_map *lock, unsigned long ip)
+__lock_release(struct lockdep_map *lock, int nested, unsigned long ip)
 {
+	struct task_struct *curr = current;
 	struct held_lock *hlock, *prev_hlock;
 	unsigned int depth;
 	int i;
 
-	/*
-	 * Check whether the lock exists in the current stack
-	 * of held locks:
-	 */
+	if (unlikely(!debug_locks))
+		return 0;
+
 	depth = curr->lockdep_depth;
 	/*
 	 * So we're all set to release this lock.. wait what lock? We don't
 	 * own any locks, you've been drinking again?
 	 */
-	if (DEBUG_LOCKS_WARN_ON(!depth))
-		return 0;
+	if (DEBUG_LOCKS_WARN_ON(depth <= 0))
+		 return print_unlock_imbalance_bug(curr, lock, ip);
 
+	/*
+	 * Check whether the lock exists in the current stack
+	 * of held locks:
+	 */
 	prev_hlock = NULL;
 	for (i = depth-1; i >= 0; i--) {
 		hlock = curr->held_locks + i;
@@ -3419,6 +3404,8 @@ found_it:
 	if (hlock->instance == lock)
 		lock_release_holdtime(hlock);
 
+	WARN(hlock->pin_count, "releasing a pinned lock\n");
+
 	if (hlock->references) {
 		hlock->references--;
 		if (hlock->references) {
@@ -3456,91 +3443,66 @@ found_it:
 	 */
 	if (DEBUG_LOCKS_WARN_ON(curr->lockdep_depth != depth - 1))
 		return 0;
+
 	return 1;
 }
 
-/*
- * Remove the lock to the list of currently held locks - this gets
- * called on mutex_unlock()/spin_unlock*() (or on a failed
- * mutex_lock_interruptible()). This is done for unlocks that nest
- * perfectly. (i.e. the current top of the lock-stack is unlocked)
- */
-static int lock_release_nested(struct task_struct *curr,
-			       struct lockdep_map *lock, unsigned long ip)
+static int __lock_is_held(struct lockdep_map *lock)
 {
-	struct held_lock *hlock;
-	unsigned int depth;
-
-	/*
-	 * Pop off the top of the lock stack:
-	 */
-	depth = curr->lockdep_depth - 1;
-	hlock = curr->held_locks + depth;
-
-	/*
-	 * Is the unlock non-nested:
-	 */
-	if (hlock->instance != lock || hlock->references)
-		return lock_release_non_nested(curr, lock, ip);
-	curr->lockdep_depth--;
-
-	/*
-	 * No more locks, but somehow we've got hash left over, who left it?
-	 */
-	if (DEBUG_LOCKS_WARN_ON(!depth && (hlock->prev_chain_key != 0)))
-		return 0;
+	struct task_struct *curr = current;
+	int i;
 
-	curr->curr_chain_key = hlock->prev_chain_key;
+	for (i = 0; i < curr->lockdep_depth; i++) {
+		struct held_lock *hlock = curr->held_locks + i;
 
-	lock_release_holdtime(hlock);
+		if (match_held_lock(hlock, lock))
+			return 1;
+	}
 
-#ifdef CONFIG_DEBUG_LOCKDEP
-	hlock->prev_chain_key = 0;
-	hlock->class_idx = 0;
-	hlock->acquire_ip = 0;
-	hlock->irq_context = 0;
-#endif
-	return 1;
+	return 0;
 }
 
-/*
- * Remove the lock to the list of currently held locks - this gets
- * called on mutex_unlock()/spin_unlock*() (or on a failed
- * mutex_lock_interruptible()). This is done for unlocks that nest
- * perfectly. (i.e. the current top of the lock-stack is unlocked)
- */
-static void
-__lock_release(struct lockdep_map *lock, int nested, unsigned long ip)
+static void __lock_pin_lock(struct lockdep_map *lock)
 {
 	struct task_struct *curr = current;
+	int i;
 
-	if (!check_unlock(curr, lock, ip))
+	if (unlikely(!debug_locks))
 		return;
 
-	if (nested) {
-		if (!lock_release_nested(curr, lock, ip))
-			return;
-	} else {
-		if (!lock_release_non_nested(curr, lock, ip))
+	for (i = 0; i < curr->lockdep_depth; i++) {
+		struct held_lock *hlock = curr->held_locks + i;
+
+		if (match_held_lock(hlock, lock)) {
+			hlock->pin_count++;
 			return;
+		}
 	}
 
-	check_chain_key(curr);
+	WARN(1, "pinning an unheld lock\n");
 }
 
-static int __lock_is_held(struct lockdep_map *lock)
+static void __lock_unpin_lock(struct lockdep_map *lock)
 {
 	struct task_struct *curr = current;
 	int i;
 
+	if (unlikely(!debug_locks))
+		return;
+
 	for (i = 0; i < curr->lockdep_depth; i++) {
 		struct held_lock *hlock = curr->held_locks + i;
 
-		if (match_held_lock(hlock, lock))
-			return 1;
+		if (match_held_lock(hlock, lock)) {
+			if (WARN(!hlock->pin_count, "unpinning an unpinned lock\n"))
+				return;
+
+			hlock->pin_count--;
+			return;
+		}
 	}
 
-	return 0;
+	WARN(1, "unpinning an unheld lock\n");
 }
 
 /*
@@ -3639,7 +3601,8 @@ void lock_release(struct lockdep_map *lock, int nested,
 	check_flags(flags);
 	current->lockdep_recursion = 1;
 	trace_lock_release(lock, ip);
-	__lock_release(lock, nested, ip);
+	if (__lock_release(lock, nested, ip))
+		check_chain_key(current);
 	current->lockdep_recursion = 0;
 	raw_local_irq_restore(flags);
 }
@@ -3665,6 +3628,40 @@ int lock_is_held(struct lockdep_map *lock)
 }
 EXPORT_SYMBOL_GPL(lock_is_held);
 
+void lock_pin_lock(struct lockdep_map *lock)
+{
+	unsigned long flags;
+
+	if (unlikely(current->lockdep_recursion))
+		return;
+
+	raw_local_irq_save(flags);
+	check_flags(flags);
+
+	current->lockdep_recursion = 1;
+	__lock_pin_lock(lock);
+	current->lockdep_recursion = 0;
+	raw_local_irq_restore(flags);
+}
+EXPORT_SYMBOL_GPL(lock_pin_lock);
+
+void lock_unpin_lock(struct lockdep_map *lock)
+{
+	unsigned long flags;
+
+	if (unlikely(current->lockdep_recursion))
+		return;
+
+	raw_local_irq_save(flags);
+	check_flags(flags);
+
+	current->lockdep_recursion = 1;
+	__lock_unpin_lock(lock);
+	current->lockdep_recursion = 0;
+	raw_local_irq_restore(flags);
+}
+EXPORT_SYMBOL_GPL(lock_unpin_lock);
+
 void lockdep_set_current_reclaim_state(gfp_t gfp_mask)
 {
 	current->lockdep_reclaim_gfp = gfp_mask;
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index c86935a7f1f8..b803e1b8ab0c 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -1000,7 +1000,11 @@ inline int task_curr(const struct task_struct *p)
 }
 
 /*
- * Can drop rq->lock because from sched_class::switched_from() methods drop it.
+ * switched_from, switched_to and prio_changed must _NOT_ drop rq->lock,
+ * use the balance_callback list if you want balancing.
+ *
+ * this means any call to check_class_changed() must be followed by a call to
+ * balance_callback().
  */
 static inline void check_class_changed(struct rq *rq, struct task_struct *p,
 				       const struct sched_class *prev_class,
@@ -1009,7 +1013,7 @@ static inline void check_class_changed(struct rq *rq, struct task_struct *p,
 	if (prev_class != p->sched_class) {
 		if (prev_class->switched_from)
 			prev_class->switched_from(rq, p);
-		/* Possble rq->lock 'hole'.  */
+
 		p->sched_class->switched_to(rq, p);
 	} else if (oldprio != p->prio || dl_task(p))
 		p->sched_class->prio_changed(rq, p, oldprio);
@@ -1041,6 +1045,177 @@ void check_preempt_curr(struct rq *rq, struct task_struct *p, int flags)
 }
 
 #ifdef CONFIG_SMP
+/*
+ * This is how migration works:
+ *
+ * 1) we invoke migration_cpu_stop() on the target CPU using
+ *    stop_one_cpu().
+ * 2) stopper starts to run (implicitly forcing the migrated thread
+ *    off the CPU)
+ * 3) it checks whether the migrated task is still in the wrong runqueue.
+ * 4) if it's in the wrong runqueue then the migration thread removes
+ *    it and puts it into the right queue.
+ * 5) stopper completes and stop_one_cpu() returns and the migration
+ *    is done.
+ */
+
+/*
+ * move_queued_task - move a queued task to new rq.
+ *
+ * Returns (locked) new rq. Old rq's lock is released.
+ */
+static struct rq *move_queued_task(struct rq *rq, struct task_struct *p, int new_cpu)
+{
+	lockdep_assert_held(&rq->lock);
+
+	dequeue_task(rq, p, 0);
+	p->on_rq = TASK_ON_RQ_MIGRATING;
+	set_task_cpu(p, new_cpu);
+	raw_spin_unlock(&rq->lock);
+
+	rq = cpu_rq(new_cpu);
+
+	raw_spin_lock(&rq->lock);
+	BUG_ON(task_cpu(p) != new_cpu);
+	p->on_rq = TASK_ON_RQ_QUEUED;
+	enqueue_task(rq, p, 0);
+	check_preempt_curr(rq, p, 0);
+
+	return rq;
+}
+
+struct migration_arg {
+	struct task_struct *task;
+	int dest_cpu;
+};
+
+/*
+ * Move (not current) task off this cpu, onto dest cpu. We're doing
+ * this because either it can't run here any more (set_cpus_allowed()
+ * away from this CPU, or CPU going down), or because we're
+ * attempting to rebalance this task on exec (sched_exec).
+ *
+ * So we race with normal scheduler movements, but that's OK, as long
+ * as the task is no longer on this CPU.
+ */
+static struct rq *__migrate_task(struct rq *rq, struct task_struct *p, int dest_cpu)
+{
+	if (unlikely(!cpu_active(dest_cpu)))
+		return rq;
+
+	/* Affinity changed (again). */
+	if (!cpumask_test_cpu(dest_cpu, tsk_cpus_allowed(p)))
+		return rq;
+
+	rq = move_queued_task(rq, p, dest_cpu);
+
+	return rq;
+}
+
+/*
+ * migration_cpu_stop - this will be executed by a highprio stopper thread
+ * and performs thread migration by bumping thread off CPU then
+ * 'pushing' onto another runqueue.
+ */
+static int migration_cpu_stop(void *data)
+{
+	struct migration_arg *arg = data;
+	struct task_struct *p = arg->task;
+	struct rq *rq = this_rq();
+
+	/*
+	 * The original target cpu might have gone down and we might
+	 * be on another cpu but it doesn't matter.
+	 */
+	local_irq_disable();
+	/*
+	 * We need to explicitly wake pending tasks before running
+	 * __migrate_task() such that we will not miss enforcing cpus_allowed
+	 * during wakeups, see set_cpus_allowed_ptr()'s TASK_WAKING test.
+	 */
+	sched_ttwu_pending();
+
+	raw_spin_lock(&p->pi_lock);
+	raw_spin_lock(&rq->lock);
+	/*
+	 * If task_rq(p) != rq, it cannot be migrated here, because we're
+	 * holding rq->lock, if p->on_rq == 0 it cannot get enqueued because
+	 * we're holding p->pi_lock.
+	 */
+	if (task_rq(p) == rq && task_on_rq_queued(p))
+		rq = __migrate_task(rq, p, arg->dest_cpu);
+	raw_spin_unlock(&rq->lock);
+	raw_spin_unlock(&p->pi_lock);
+
+	local_irq_enable();
+	return 0;
+}
+
+void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask)
+{
+	if (p->sched_class->set_cpus_allowed)
+		p->sched_class->set_cpus_allowed(p, new_mask);
+
+	cpumask_copy(&p->cpus_allowed, new_mask);
+	p->nr_cpus_allowed = cpumask_weight(new_mask);
+}
+
+/*
+ * Change a given task's CPU affinity. Migrate the thread to a
+ * proper CPU and schedule it away if the CPU it's executing on
+ * is removed from the allowed bitmask.
+ *
+ * NOTE: the caller must have a valid reference to the task, the
+ * task must not exit() & deallocate itself prematurely. The
+ * call is not atomic; no spinlocks may be held.
+ */
+int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask)
+{
+	unsigned long flags;
+	struct rq *rq;
+	unsigned int dest_cpu;
+	int ret = 0;
+
+	rq = task_rq_lock(p, &flags);
+
+	if (cpumask_equal(&p->cpus_allowed, new_mask))
+		goto out;
+
+	if (!cpumask_intersects(new_mask, cpu_active_mask)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	do_set_cpus_allowed(p, new_mask);
+
+	/* Can the task run on the task's current CPU? If so, we're done */
+	if (cpumask_test_cpu(task_cpu(p), new_mask))
+		goto out;
+
+	dest_cpu = cpumask_any_and(cpu_active_mask, new_mask);
+	if (task_running(rq, p) || p->state == TASK_WAKING) {
+		struct migration_arg arg = { p, dest_cpu };
+		/* Need help from migration thread: drop lock and wait. */
+		task_rq_unlock(rq, p, &flags);
+		stop_one_cpu(cpu_of(rq), migration_cpu_stop, &arg);
+		tlb_migrate_finish(p->mm);
+		return 0;
+	} else if (task_on_rq_queued(p)) {
+		/*
+		 * OK, since we're going to drop the lock immediately
+		 * afterwards anyway.
+		 */
+		lockdep_unpin_lock(&rq->lock);
+		rq = move_queued_task(rq, p, dest_cpu);
+		lockdep_pin_lock(&rq->lock);
+	}
+out:
+	task_rq_unlock(rq, p, &flags);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(set_cpus_allowed_ptr);
+
 void set_task_cpu(struct task_struct *p, unsigned int new_cpu)
 {
 #ifdef CONFIG_SCHED_DEBUG
@@ -1181,13 +1356,6 @@ out:
 	return ret;
 }
 
-struct migration_arg {
-	struct task_struct *task;
-	int dest_cpu;
-};
-
-static int migration_cpu_stop(void *data);
-
 /*
  * wait_task_inactive - wait for a thread to unschedule.
  *
@@ -1320,9 +1488,7 @@ void kick_process(struct task_struct *p)
 	preempt_enable();
 }
 EXPORT_SYMBOL_GPL(kick_process);
-#endif /* CONFIG_SMP */
 
-#ifdef CONFIG_SMP
 /*
  * ->cpus_allowed is protected by both rq->lock and p->pi_lock
  */
@@ -1402,6 +1568,8 @@ out:
 static inline
 int select_task_rq(struct task_struct *p, int cpu, int sd_flags, int wake_flags)
 {
+	lockdep_assert_held(&p->pi_lock);
+
 	if (p->nr_cpus_allowed > 1)
 		cpu = p->sched_class->select_task_rq(p, cpu, sd_flags, wake_flags);
 
@@ -1427,7 +1595,7 @@ static void update_avg(u64 *avg, u64 sample)
 	s64 diff = sample - *avg;
 	*avg += diff >> 3;
 }
-#endif
+#endif /* CONFIG_SMP */
 
 static void
 ttwu_stat(struct task_struct *p, int cpu, int wake_flags)
@@ -1490,8 +1658,15 @@ ttwu_do_wakeup(struct rq *rq, struct task_struct *p, int wake_flags)
 
 	p->state = TASK_RUNNING;
 #ifdef CONFIG_SMP
-	if (p->sched_class->task_woken)
+	if (p->sched_class->task_woken) {
+		/*
+		 * Our task @p is fully woken up and running; so its safe to
+		 * drop the rq->lock, hereafter rq is only used for statistics.
+		 */
+		lockdep_unpin_lock(&rq->lock);
 		p->sched_class->task_woken(rq, p);
+		lockdep_pin_lock(&rq->lock);
+	}
 
 	if (rq->idle_stamp) {
 		u64 delta = rq_clock(rq) - rq->idle_stamp;
@@ -1510,6 +1685,8 @@ ttwu_do_wakeup(struct rq *rq, struct task_struct *p, int wake_flags)
 static void
 ttwu_do_activate(struct rq *rq, struct task_struct *p, int wake_flags)
 {
+	lockdep_assert_held(&rq->lock);
+
 #ifdef CONFIG_SMP
 	if (p->sched_contributes_to_load)
 		rq->nr_uninterruptible--;
@@ -1554,6 +1731,7 @@ void sched_ttwu_pending(void)
 		return;
 
 	raw_spin_lock_irqsave(&rq->lock, flags);
+	lockdep_pin_lock(&rq->lock);
 
 	while (llist) {
 		p = llist_entry(llist, struct task_struct, wake_entry);
@@ -1561,6 +1739,7 @@ void sched_ttwu_pending(void)
 		ttwu_do_activate(rq, p, 0);
 	}
 
+	lockdep_unpin_lock(&rq->lock);
 	raw_spin_unlock_irqrestore(&rq->lock, flags);
 }
 
@@ -1657,7 +1836,9 @@ static void ttwu_queue(struct task_struct *p, int cpu)
 #endif
 
 	raw_spin_lock(&rq->lock);
+	lockdep_pin_lock(&rq->lock);
 	ttwu_do_activate(rq, p, 0);
+	lockdep_unpin_lock(&rq->lock);
 	raw_spin_unlock(&rq->lock);
 }
 
@@ -1752,9 +1933,17 @@ static void try_to_wake_up_local(struct task_struct *p)
 	lockdep_assert_held(&rq->lock);
 
 	if (!raw_spin_trylock(&p->pi_lock)) {
+		/*
+		 * This is OK, because current is on_cpu, which avoids it being
+		 * picked for load-balance and preemption/IRQs are still
+		 * disabled avoiding further scheduler activity on it and we've
+		 * not yet picked a replacement task.
+		 */
+		lockdep_unpin_lock(&rq->lock);
 		raw_spin_unlock(&rq->lock);
 		raw_spin_lock(&p->pi_lock);
 		raw_spin_lock(&rq->lock);
+		lockdep_pin_lock(&rq->lock);
 	}
 
 	if (!(p->state & TASK_NORMAL))
@@ -2294,23 +2483,35 @@ static struct rq *finish_task_switch(struct task_struct *prev)
 #ifdef CONFIG_SMP
 
 /* rq->lock is NOT held, but preemption is disabled */
-static inline void post_schedule(struct rq *rq)
+static void __balance_callback(struct rq *rq)
 {
-	if (rq->post_schedule) {
-		unsigned long flags;
+	struct callback_head *head, *next;
+	void (*func)(struct rq *rq);
+	unsigned long flags;
 
-		raw_spin_lock_irqsave(&rq->lock, flags);
-		if (rq->curr->sched_class->post_schedule)
-			rq->curr->sched_class->post_schedule(rq);
-		raw_spin_unlock_irqrestore(&rq->lock, flags);
+	raw_spin_lock_irqsave(&rq->lock, flags);
+	head = rq->balance_callback;
+	rq->balance_callback = NULL;
+	while (head) {
+		func = (void (*)(struct rq *))head->func;
+		next = head->next;
+		head->next = NULL;
+		head = next;
 
-		rq->post_schedule = 0;
+		func(rq);
 	}
+	raw_spin_unlock_irqrestore(&rq->lock, flags);
+}
+
+static inline void balance_callback(struct rq *rq)
+{
+	if (unlikely(rq->balance_callback))
+		__balance_callback(rq);
 }
 
 #else
 
-static inline void post_schedule(struct rq *rq)
+static inline void balance_callback(struct rq *rq)
 {
 }
 
@@ -2328,7 +2529,7 @@ asmlinkage __visible void schedule_tail(struct task_struct *prev)
 	/* finish_task_switch() drops rq->lock and enables preemtion */
 	preempt_disable();
 	rq = finish_task_switch(prev);
-	post_schedule(rq);
+	balance_callback(rq);
 	preempt_enable();
 
 	if (current->set_child_tid)
@@ -2372,6 +2573,7 @@ context_switch(struct rq *rq, struct task_struct *prev,
 	 * of the scheduler it's an obvious special-case), so we
 	 * do an early lockdep release here:
 	 */
+	lockdep_unpin_lock(&rq->lock);
 	spin_release(&rq->lock.dep_map, 1, _THIS_IP_);
 
 	/* Here we just switch the register state and the stack. */
@@ -2794,6 +2996,7 @@ static void __sched __schedule(void)
 	 */
 	smp_mb__before_spinlock();
 	raw_spin_lock_irq(&rq->lock);
+	lockdep_pin_lock(&rq->lock);
 
 	rq->clock_skip_update <<= 1; /* promote REQ to ACT */
 
@@ -2836,10 +3039,12 @@ static void __sched __schedule(void)
 
 		rq = context_switch(rq, prev, next); /* unlocks the rq */
 		cpu = cpu_of(rq);
-	} else
+	} else {
+		lockdep_unpin_lock(&rq->lock);
 		raw_spin_unlock_irq(&rq->lock);
+	}
 
-	post_schedule(rq);
+	balance_callback(rq);
 }
 
 static inline void sched_submit_work(struct task_struct *tsk)
@@ -3103,7 +3308,11 @@ void rt_mutex_setprio(struct task_struct *p, int prio)
 
 	check_class_changed(rq, p, prev_class, oldprio);
 out_unlock:
+	preempt_disable(); /* avoid rq from going away on us */
 	__task_rq_unlock(rq);
+
+	balance_callback(rq);
+	preempt_enable();
 }
 #endif
 
@@ -3441,7 +3650,7 @@ static bool dl_param_changed(struct task_struct *p,
 
 static int __sched_setscheduler(struct task_struct *p,
 				const struct sched_attr *attr,
-				bool user)
+				bool user, bool pi)
 {
 	int newprio = dl_policy(attr->sched_policy) ? MAX_DL_PRIO - 1 :
 		      MAX_RT_PRIO - 1 - attr->sched_priority;
@@ -3627,18 +3836,20 @@ change:
 	p->sched_reset_on_fork = reset_on_fork;
 	oldprio = p->prio;
 
-	/*
-	 * Take priority boosted tasks into account. If the new
-	 * effective priority is unchanged, we just store the new
-	 * normal parameters and do not touch the scheduler class and
-	 * the runqueue. This will be done when the task deboost
-	 * itself.
-	 */
-	new_effective_prio = rt_mutex_get_effective_prio(p, newprio);
-	if (new_effective_prio == oldprio) {
-		__setscheduler_params(p, attr);
-		task_rq_unlock(rq, p, &flags);
-		return 0;
+	if (pi) {
+		/*
+		 * Take priority boosted tasks into account. If the new
+		 * effective priority is unchanged, we just store the new
+		 * normal parameters and do not touch the scheduler class and
+		 * the runqueue. This will be done when the task deboost
+		 * itself.
+		 */
+		new_effective_prio = rt_mutex_get_effective_prio(p, newprio);
+		if (new_effective_prio == oldprio) {
+			__setscheduler_params(p, attr);
+			task_rq_unlock(rq, p, &flags);
+			return 0;
+		}
 	}
 
 	queued = task_on_rq_queued(p);
@@ -3649,7 +3860,7 @@ change:
 		put_prev_task(rq, p);
 
 	prev_class = p->sched_class;
-	__setscheduler(rq, p, attr, true);
+	__setscheduler(rq, p, attr, pi);
 
 	if (running)
 		p->sched_class->set_curr_task(rq);
@@ -3662,9 +3873,17 @@ change:
 	}
 
 	check_class_changed(rq, p, prev_class, oldprio);
+	preempt_disable(); /* avoid rq from going away on us */
 	task_rq_unlock(rq, p, &flags);
 
-	rt_mutex_adjust_pi(p);
+	if (pi)
+		rt_mutex_adjust_pi(p);
+
+	/*
+	 * Run balance callbacks after we've adjusted the PI chain.
+	 */
+	balance_callback(rq);
+	preempt_enable();
 
 	return 0;
 }
@@ -3685,7 +3904,7 @@ static int _sched_setscheduler(struct task_struct *p, int policy,
 		attr.sched_policy = policy;
 	}
 
-	return __sched_setscheduler(p, &attr, check);
+	return __sched_setscheduler(p, &attr, check, true);
 }
 /**
  * sched_setscheduler - change the scheduling policy and/or RT priority of a thread.
@@ -3706,7 +3925,7 @@ EXPORT_SYMBOL_GPL(sched_setscheduler);
 
 int sched_setattr(struct task_struct *p, const struct sched_attr *attr)
 {
-	return __sched_setscheduler(p, attr, true);
+	return __sched_setscheduler(p, attr, true, true);
 }
 EXPORT_SYMBOL_GPL(sched_setattr);
 
@@ -4754,149 +4973,6 @@ out:
 }
 
 #ifdef CONFIG_SMP
-/*
- * move_queued_task - move a queued task to new rq.
- *
- * Returns (locked) new rq. Old rq's lock is released.
- */
-static struct rq *move_queued_task(struct task_struct *p, int new_cpu)
-{
-	struct rq *rq = task_rq(p);
-
-	lockdep_assert_held(&rq->lock);
-
-	dequeue_task(rq, p, 0);
-	p->on_rq = TASK_ON_RQ_MIGRATING;
-	set_task_cpu(p, new_cpu);
-	raw_spin_unlock(&rq->lock);
-
-	rq = cpu_rq(new_cpu);
-
-	raw_spin_lock(&rq->lock);
-	BUG_ON(task_cpu(p) != new_cpu);
-	p->on_rq = TASK_ON_RQ_QUEUED;
-	enqueue_task(rq, p, 0);
-	check_preempt_curr(rq, p, 0);
-
-	return rq;
-}
-
-void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask)
-{
-	if (p->sched_class->set_cpus_allowed)
-		p->sched_class->set_cpus_allowed(p, new_mask);
-
-	cpumask_copy(&p->cpus_allowed, new_mask);
-	p->nr_cpus_allowed = cpumask_weight(new_mask);
-}
-
-/*
- * This is how migration works:
- *
- * 1) we invoke migration_cpu_stop() on the target CPU using
- *    stop_one_cpu().
- * 2) stopper starts to run (implicitly forcing the migrated thread
- *    off the CPU)
- * 3) it checks whether the migrated task is still in the wrong runqueue.
- * 4) if it's in the wrong runqueue then the migration thread removes
- *    it and puts it into the right queue.
- * 5) stopper completes and stop_one_cpu() returns and the migration
- *    is done.
- */
-
-/*
- * Change a given task's CPU affinity. Migrate the thread to a
- * proper CPU and schedule it away if the CPU it's executing on
- * is removed from the allowed bitmask.
- *
- * NOTE: the caller must have a valid reference to the task, the
- * task must not exit() & deallocate itself prematurely. The
- * call is not atomic; no spinlocks may be held.
- */
-int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask)
-{
-	unsigned long flags;
-	struct rq *rq;
-	unsigned int dest_cpu;
-	int ret = 0;
-
-	rq = task_rq_lock(p, &flags);
-
-	if (cpumask_equal(&p->cpus_allowed, new_mask))
-		goto out;
-
-	if (!cpumask_intersects(new_mask, cpu_active_mask)) {
-		ret = -EINVAL;
-		goto out;
-	}
-
-	do_set_cpus_allowed(p, new_mask);
-
-	/* Can the task run on the task's current CPU? If so, we're done */
-	if (cpumask_test_cpu(task_cpu(p), new_mask))
-		goto out;
-
-	dest_cpu = cpumask_any_and(cpu_active_mask, new_mask);
-	if (task_running(rq, p) || p->state == TASK_WAKING) {
-		struct migration_arg arg = { p, dest_cpu };
-		/* Need help from migration thread: drop lock and wait. */
-		task_rq_unlock(rq, p, &flags);
-		stop_one_cpu(cpu_of(rq), migration_cpu_stop, &arg);
-		tlb_migrate_finish(p->mm);
-		return 0;
-	} else if (task_on_rq_queued(p))
-		rq = move_queued_task(p, dest_cpu);
-out:
-	task_rq_unlock(rq, p, &flags);
-
-	return ret;
-}
-EXPORT_SYMBOL_GPL(set_cpus_allowed_ptr);
-
-/*
- * Move (not current) task off this cpu, onto dest cpu. We're doing
- * this because either it can't run here any more (set_cpus_allowed()
- * away from this CPU, or CPU going down), or because we're
- * attempting to rebalance this task on exec (sched_exec).
- *
- * So we race with normal scheduler movements, but that's OK, as long
- * as the task is no longer on this CPU.
- *
- * Returns non-zero if task was successfully migrated.
- */
-static int __migrate_task(struct task_struct *p, int src_cpu, int dest_cpu)
-{
-	struct rq *rq;
-	int ret = 0;
-
-	if (unlikely(!cpu_active(dest_cpu)))
-		return ret;
-
-	rq = cpu_rq(src_cpu);
-
-	raw_spin_lock(&p->pi_lock);
-	raw_spin_lock(&rq->lock);
-	/* Already moved. */
-	if (task_cpu(p) != src_cpu)
-		goto done;
-
-	/* Affinity changed (again). */
-	if (!cpumask_test_cpu(dest_cpu, tsk_cpus_allowed(p)))
-		goto fail;
-
-	/*
-	 * If we're not on a rq, the next wake-up will ensure we're
-	 * placed properly.
-	 */
-	if (task_on_rq_queued(p))
-		rq = move_queued_task(p, dest_cpu);
-done:
-	ret = 1;
-fail:
-	raw_spin_unlock(&rq->lock);
-	raw_spin_unlock(&p->pi_lock);
-	return ret;
-}
 
 #ifdef CONFIG_NUMA_BALANCING
 /* Migrate current task p to target_cpu */
@@ -4944,35 +5020,9 @@ void sched_setnuma(struct task_struct *p, int nid)
 		enqueue_task(rq, p, 0);
 	task_rq_unlock(rq, p, &flags);
 }
-#endif
-
-/*
- * migration_cpu_stop - this will be executed by a highprio stopper thread
- * and performs thread migration by bumping thread off CPU then
- * 'pushing' onto another runqueue.
- */
-static int migration_cpu_stop(void *data)
-{
-	struct migration_arg *arg = data;
-
-	/*
-	 * The original target cpu might have gone down and we might
-	 * be on another cpu but it doesn't matter.
-	 */
-	local_irq_disable();
-	/*
-	 * We need to explicitly wake pending tasks before running
-	 * __migrate_task() such that we will not miss enforcing cpus_allowed
-	 * during wakeups, see set_cpus_allowed_ptr()'s TASK_WAKING test.
-	 */
-	sched_ttwu_pending();
-	__migrate_task(arg->task, raw_smp_processor_id(), arg->dest_cpu);
-	local_irq_enable();
-	return 0;
-}
+#endif /* CONFIG_NUMA_BALANCING */
 
 #ifdef CONFIG_HOTPLUG_CPU
-
 /*
  * Ensures that the idle task is using init_mm right before its cpu goes
  * offline.
@@ -5028,9 +5078,9 @@ static struct task_struct fake_task = {
  * there's no concurrency possible, we hold the required locks anyway
  * because of lock validation efforts.
  */
-static void migrate_tasks(unsigned int dead_cpu)
+static void migrate_tasks(struct rq *dead_rq)
 {
-	struct rq *rq = cpu_rq(dead_cpu);
+	struct rq *rq = dead_rq;
 	struct task_struct *next, *stop = rq->stop;
 	int dest_cpu;
 
@@ -5052,7 +5102,7 @@ static void migrate_tasks(unsigned int dead_cpu)
 	 */
 	update_rq_clock(rq);
 
-	for ( ; ; ) {
+	for (;;) {
 		/*
 		 * There's this thread running, bail when that's the only
 		 * remaining thread.
@@ -5060,22 +5110,29 @@ static void migrate_tasks(unsigned int dead_cpu)
 		if (rq->nr_running == 1)
 			break;
 
+		/*
+		 * Ensure rq->lock covers the entire task selection
+		 * until the migration.
+		 */
+		lockdep_pin_lock(&rq->lock);
 		next = pick_next_task(rq, &fake_task);
 		BUG_ON(!next);
 		next->sched_class->put_prev_task(rq, next);
 
 		/* Find suitable destination for @next, with force if needed. */
-		dest_cpu = select_fallback_rq(dead_cpu, next);
-		raw_spin_unlock(&rq->lock);
-
-		__migrate_task(next, dead_cpu, dest_cpu);
-
-		raw_spin_lock(&rq->lock);
+		dest_cpu = select_fallback_rq(dead_rq->cpu, next);
+
+		lockdep_unpin_lock(&rq->lock);
+		rq = __migrate_task(rq, next, dest_cpu);
+		if (rq != dead_rq) {
+			raw_spin_unlock(&rq->lock);
+			rq = dead_rq;
+			raw_spin_lock(&rq->lock);
+		}
 	}
 
 	rq->stop = stop;
 }
-
 #endif /* CONFIG_HOTPLUG_CPU */
 
 #if defined(CONFIG_SCHED_DEBUG) && defined(CONFIG_SYSCTL)
@@ -5254,7 +5311,7 @@ static void register_sched_domain_sysctl(void)
 static void unregister_sched_domain_sysctl(void)
 {
 }
-#endif
+#endif /* CONFIG_SCHED_DEBUG && CONFIG_SYSCTL */
 
 static void set_rq_online(struct rq *rq)
 {
@@ -5323,7 +5380,7 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu)
 			BUG_ON(!cpumask_test_cpu(cpu, rq->rd->span));
 			set_rq_offline(rq);
 		}
-		migrate_tasks(cpu);
+		migrate_tasks(rq);
 		BUG_ON(rq->nr_running != 1); /* the migration thread */
 		raw_spin_unlock_irqrestore(&rq->lock, flags);
 		break;
@@ -5401,9 +5458,6 @@ static int __init migration_init(void)
 	return 0;
 }
 early_initcall(migration_init);
-#endif
-
-#ifdef CONFIG_SMP
 
 static cpumask_var_t sched_domains_tmpmask; /* sched_domains_mutex */
 
@@ -6629,7 +6683,7 @@ static int __sdt_alloc(const struct cpumask *cpu_map)
 			struct sched_group *sg;
 			struct sched_group_capacity *sgc;
 
-		       	sd = kzalloc_node(sizeof(struct sched_domain) + cpumask_size(),
+			sd = kzalloc_node(sizeof(struct sched_domain) + cpumask_size(),
 					GFP_KERNEL, cpu_to_node(j));
 			if (!sd)
 				return -ENOMEM;
@@ -7235,7 +7289,7 @@ void __init sched_init(void)
 		rq->sd = NULL;
 		rq->rd = NULL;
 		rq->cpu_capacity = rq->cpu_capacity_orig = SCHED_CAPACITY_SCALE;
-		rq->post_schedule = 0;
+		rq->balance_callback = NULL;
 		rq->active_balance = 0;
 		rq->next_balance = jiffies;
 		rq->push_cpu = 0;
@@ -7365,32 +7419,12 @@ EXPORT_SYMBOL(___might_sleep);
 #endif
 
 #ifdef CONFIG_MAGIC_SYSRQ
-static void normalize_task(struct rq *rq, struct task_struct *p)
+void normalize_rt_tasks(void)
 {
-	const struct sched_class *prev_class = p->sched_class;
+	struct task_struct *g, *p;
 	struct sched_attr attr = {
 		.sched_policy = SCHED_NORMAL,
 	};
-	int old_prio = p->prio;
-	int queued;
-
-	queued = task_on_rq_queued(p);
-	if (queued)
-		dequeue_task(rq, p, 0);
-	__setscheduler(rq, p, &attr, false);
-	if (queued) {
-		enqueue_task(rq, p, 0);
-		resched_curr(rq);
-	}
-
-	check_class_changed(rq, p, prev_class, old_prio);
-}
-
-void normalize_rt_tasks(void)
-{
-	struct task_struct *g, *p;
-	unsigned long flags;
-	struct rq *rq;
 
 	read_lock(&tasklist_lock);
 	for_each_process_thread(g, p) {
@@ -7417,9 +7451,7 @@ void normalize_rt_tasks(void)
 			continue;
 		}
 
-		rq = task_rq_lock(p, &flags);
-		normalize_task(rq, p);
-		task_rq_unlock(rq, p, &flags);
+		__sched_setscheduler(p, &attr, false, false);
 	}
 	read_unlock(&tasklist_lock);
 }
diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
index eac20c557a55..0a17af35670a 100644
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -213,14 +213,28 @@ static inline bool need_pull_dl_task(struct rq *rq, struct task_struct *prev)
 	return dl_task(prev);
 }
 
-static inline void set_post_schedule(struct rq *rq)
+static DEFINE_PER_CPU(struct callback_head, dl_push_head);
+static DEFINE_PER_CPU(struct callback_head, dl_pull_head);
+
+static void push_dl_tasks(struct rq *);
+static void pull_dl_task(struct rq *);
+
+static inline void queue_push_tasks(struct rq *rq)
 {
-	rq->post_schedule = has_pushable_dl_tasks(rq);
+	if (!has_pushable_dl_tasks(rq))
+		return;
+
+	queue_balance_callback(rq, &per_cpu(dl_push_head, rq->cpu), push_dl_tasks);
+}
+
+static inline void queue_pull_task(struct rq *rq)
+{
+	queue_balance_callback(rq, &per_cpu(dl_pull_head, rq->cpu), pull_dl_task);
 }
 
 static struct rq *find_lock_later_rq(struct task_struct *task, struct rq *rq);
 
-static void dl_task_offline_migration(struct rq *rq, struct task_struct *p)
+static struct rq *dl_task_offline_migration(struct rq *rq, struct task_struct *p)
 {
 	struct rq *later_rq = NULL;
 	bool fallback = false;
@@ -254,14 +268,19 @@ static void dl_task_offline_migration(struct rq *rq, struct task_struct *p)
 		double_lock_balance(rq, later_rq);
 	}
 
+	/*
+	 * By now the task is replenished and enqueued; migrate it.
+	 */
 	deactivate_task(rq, p, 0);
 	set_task_cpu(p, later_rq->cpu);
-	activate_task(later_rq, p, ENQUEUE_REPLENISH);
+	activate_task(later_rq, p, 0);
 
 	if (!fallback)
 		resched_curr(later_rq);
 
-	double_unlock_balance(rq, later_rq);
+	double_unlock_balance(later_rq, rq);
+
+	return later_rq;
 }
 
 #else
@@ -291,12 +310,15 @@ static inline bool need_pull_dl_task(struct rq *rq, struct task_struct *prev)
 	return false;
 }
 
-static inline int pull_dl_task(struct rq *rq)
+static inline void pull_dl_task(struct rq *rq)
+{
+}
+
+static inline void queue_push_tasks(struct rq *rq)
 {
-	return 0;
 }
 
-static inline void set_post_schedule(struct rq *rq)
+static inline void queue_pull_task(struct rq *rq)
 {
 }
 #endif /* CONFIG_SMP */
@@ -498,22 +520,23 @@ static void update_dl_entity(struct sched_dl_entity *dl_se,
  * actually started or not (i.e., the replenishment instant is in
  * the future or in the past).
  */
-static int start_dl_timer(struct sched_dl_entity *dl_se, bool boosted)
+static int start_dl_timer(struct task_struct *p)
 {
-	struct dl_rq *dl_rq = dl_rq_of_se(dl_se);
-	struct rq *rq = rq_of_dl_rq(dl_rq);
+	struct sched_dl_entity *dl_se = &p->dl;
+	struct hrtimer *timer = &dl_se->dl_timer;
+	struct rq *rq = task_rq(p);
 	ktime_t now, act;
 	s64 delta;
 
-	if (boosted)
-		return 0;
+	lockdep_assert_held(&rq->lock);
+
 	/*
 	 * We want the timer to fire at the deadline, but considering
 	 * that it is actually coming from rq->clock and not from
 	 * hrtimer's time base reading.
 	 */
 	act = ns_to_ktime(dl_se->deadline);
-	now = hrtimer_cb_get_time(&dl_se->dl_timer);
+	now = hrtimer_cb_get_time(timer);
 	delta = ktime_to_ns(now) - rq_clock(rq);
 	act = ktime_add_ns(act, delta);
 
@@ -525,7 +548,19 @@ static int start_dl_timer(struct sched_dl_entity *dl_se, bool boosted)
 	if (ktime_us_delta(act, now) < 0)
 		return 0;
 
-	hrtimer_start(&dl_se->dl_timer, act, HRTIMER_MODE_ABS);
+	/*
+	 * !enqueued will guarantee another callback; even if one is already in
+	 * progress. This ensures a balanced {get,put}_task_struct().
+	 *
+	 * The race against __run_timer() clearing the enqueued state is
+	 * harmless because we're holding task_rq()->lock, therefore the timer
+	 * expiring after we've done the check will wait on its task_rq_lock()
+	 * and observe our state.
+	 */
+	if (!hrtimer_is_queued(timer)) {
+		get_task_struct(p);
+		hrtimer_start(timer, act, HRTIMER_MODE_ABS);
+	}
 
 	return 1;
 }
@@ -555,35 +590,40 @@ static enum hrtimer_restart dl_task_timer(struct hrtimer *timer)
 	rq = task_rq_lock(p, &flags);
 
 	/*
-	 * We need to take care of several possible races here:
-	 *
-	 *   - the task might have changed its scheduling policy
-	 *     to something different than SCHED_DEADLINE
-	 *   - the task might have changed its reservation parameters
-	 *     (through sched_setattr())
-	 *   - the task might have been boosted by someone else and
-	 *     might be in the boosting/deboosting path
+	 * The task might have changed its scheduling policy to something
+	 * different than SCHED_DEADLINE (through switched_fromd_dl()).
+	 */
+	if (!dl_task(p)) {
+		__dl_clear_params(p);
+		goto unlock;
+	}
+
+	/*
+	 * This is possible if switched_from_dl() raced against a running
+	 * callback that took the above !dl_task() path and we've since then
+	 * switched back into SCHED_DEADLINE.
 	 *
-	 * In all this cases we bail out, as the task is already
-	 * in the runqueue or is going to be enqueued back anyway.
+	 * There's nothing to do except drop our task reference.
 	 */
-	if (!dl_task(p) || dl_se->dl_new ||
-	    dl_se->dl_boosted || !dl_se->dl_throttled)
+	if (dl_se->dl_new)
 		goto unlock;
 
-	sched_clock_tick();
-	update_rq_clock(rq);
+	/*
+	 * The task might have been boosted by someone else and might be in the
+	 * boosting/deboosting path, its not throttled.
+	 */
+	if (dl_se->dl_boosted)
+		goto unlock;
 
-#ifdef CONFIG_SMP
 	/*
-	 * If we find that the rq the task was on is no longer
-	 * available, we need to select a new rq.
+	 * Spurious timer due to start_dl_timer() race; or we already received
+	 * a replenishment from rt_mutex_setprio().
 	 */
-	if (unlikely(!rq->online)) {
-		dl_task_offline_migration(rq, p);
+	if (!dl_se->dl_throttled)
 		goto unlock;
-	}
-#endif
+
+	sched_clock_tick();
+	update_rq_clock(rq);
 
 	/*
 	 * If the throttle happened during sched-out; like:
@@ -609,17 +649,38 @@ static enum hrtimer_restart dl_task_timer(struct hrtimer *timer)
 		check_preempt_curr_dl(rq, p, 0);
 	else
 		resched_curr(rq);
+
 #ifdef CONFIG_SMP
 	/*
-	 * Queueing this task back might have overloaded rq,
-	 * check if we need to kick someone away.
+	 * Perform balancing operations here; after the replenishments.  We
+	 * cannot drop rq->lock before this, otherwise the assertion in
+	 * start_dl_timer() about not missing updates is not true.
+	 *
+	 * If we find that the rq the task was on is no longer available, we
+	 * need to select a new rq.
+	 *
+	 * XXX figure out if select_task_rq_dl() deals with offline cpus.
+	 */
+	if (unlikely(!rq->online))
+		rq = dl_task_offline_migration(rq, p);
+
+	/*
+	 * Queueing this task back might have overloaded rq, check if we need
+	 * to kick someone away.
 	 */
 	if (has_pushable_dl_tasks(rq))
 		push_dl_task(rq);
 #endif
+
 unlock:
 	task_rq_unlock(rq, p, &flags);
 
+	/*
+	 * This can free the task_struct, including this hrtimer, do not touch
+	 * anything related to that after this.
+	 */
+	put_task_struct(p);
+
 	return HRTIMER_NORESTART;
 }
 
@@ -679,7 +740,7 @@ static void update_curr_dl(struct rq *rq)
 	if (dl_runtime_exceeded(dl_se)) {
 		dl_se->dl_throttled = 1;
 		__dequeue_task_dl(rq, curr, 0);
-		if (unlikely(!start_dl_timer(dl_se, curr->dl.dl_boosted)))
+		if (unlikely(dl_se->dl_boosted || !start_dl_timer(curr)))
 			enqueue_task_dl(rq, curr, ENQUEUE_REPLENISH);
 
 		if (!is_leftmost(curr, &rq->dl))
@@ -1036,8 +1097,6 @@ static void check_preempt_equal_dl(struct rq *rq, struct task_struct *p)
 	resched_curr(rq);
 }
 
-static int pull_dl_task(struct rq *this_rq);
-
 #endif /* CONFIG_SMP */
 
 /*
@@ -1094,7 +1153,15 @@ struct task_struct *pick_next_task_dl(struct rq *rq, struct task_struct *prev)
 	dl_rq = &rq->dl;
 
 	if (need_pull_dl_task(rq, prev)) {
+		/*
+		 * This is OK, because current is on_cpu, which avoids it being
+		 * picked for load-balance and preemption/IRQs are still
+		 * disabled avoiding further scheduler activity on it and we're
+		 * being very careful to re-start the picking loop.
+		 */
+		lockdep_unpin_lock(&rq->lock);
 		pull_dl_task(rq);
+		lockdep_pin_lock(&rq->lock);
 		/*
 		 * pull_rt_task() can drop (and re-acquire) rq->lock; this
 		 * means a stop task can slip in, in which case we need to
@@ -1128,7 +1195,7 @@ struct task_struct *pick_next_task_dl(struct rq *rq, struct task_struct *prev)
 	if (hrtick_enabled(rq))
 		start_hrtick_dl(rq, p);
 
-	set_post_schedule(rq);
+	queue_push_tasks(rq);
 
 	return p;
 }
@@ -1165,7 +1232,6 @@ static void task_fork_dl(struct task_struct *p)
 
 static void task_dead_dl(struct task_struct *p)
 {
-	struct hrtimer *timer = &p->dl.dl_timer;
 	struct dl_bw *dl_b = dl_bw_of(task_cpu(p));
 
 	/*
@@ -1175,8 +1241,6 @@ static void task_dead_dl(struct task_struct *p)
 	/* XXX we should retain the bw until 0-lag */
 	dl_b->total_bw -= p->dl.dl_bw;
 	raw_spin_unlock_irq(&dl_b->lock);
-
-	hrtimer_cancel(timer);
 }
 
 static void set_curr_task_dl(struct rq *rq)
@@ -1504,15 +1568,16 @@ static void push_dl_tasks(struct rq *rq)
 		;
 }
 
-static int pull_dl_task(struct rq *this_rq)
+static void pull_dl_task(struct rq *this_rq)
 {
-	int this_cpu = this_rq->cpu, ret = 0, cpu;
+	int this_cpu = this_rq->cpu, cpu;
 	struct task_struct *p;
+	bool resched = false;
 	struct rq *src_rq;
 	u64 dmin = LONG_MAX;
 
 	if (likely(!dl_overloaded(this_rq)))
-		return 0;
+		return;
 
 	/*
 	 * Match the barrier from dl_set_overloaded; this guarantees that if we
@@ -1567,7 +1632,7 @@ static int pull_dl_task(struct rq *this_rq)
 					   src_rq->curr->dl.deadline))
 				goto skip;
 
-			ret = 1;
+			resched = true;
 
 			deactivate_task(src_rq, p, 0);
 			set_task_cpu(p, this_cpu);
@@ -1580,12 +1645,8 @@ skip:
 		double_unlock_balance(this_rq, src_rq);
 	}
 
-	return ret;
-}
-
-static void post_schedule_dl(struct rq *rq)
-{
-	push_dl_tasks(rq);
+	if (resched)
+		resched_curr(this_rq);
 }
 
 /*
@@ -1701,37 +1762,16 @@ void __init init_sched_dl_class(void)
 
 #endif /* CONFIG_SMP */
 
-/*
- *  Ensure p's dl_timer is cancelled. May drop rq->lock for a while.
- */
-static void cancel_dl_timer(struct rq *rq, struct task_struct *p)
-{
-	struct hrtimer *dl_timer = &p->dl.dl_timer;
-
-	/* Nobody will change task's class if pi_lock is held */
-	lockdep_assert_held(&p->pi_lock);
-
-	if (hrtimer_active(dl_timer)) {
-		int ret = hrtimer_try_to_cancel(dl_timer);
-
-		if (unlikely(ret == -1)) {
-			/*
-			 * Note, p may migrate OR new deadline tasks
-			 * may appear in rq when we are unlocking it.
-			 * A caller of us must be fine with that.
-			 */
-			raw_spin_unlock(&rq->lock);
-			hrtimer_cancel(dl_timer);
-			raw_spin_lock(&rq->lock);
-		}
-	}
-}
-
 static void switched_from_dl(struct rq *rq, struct task_struct *p)
 {
-	/* XXX we should retain the bw until 0-lag */
-	cancel_dl_timer(rq, p);
-	__dl_clear_params(p);
+	/*
+	 * Start the deadline timer; if we switch back to dl before this we'll
+	 * continue consuming our current CBS slice. If we stay outside of
+	 * SCHED_DEADLINE until the deadline passes, the timer will reset the
+	 * task.
+	 */
+	if (!start_dl_timer(p))
+		__dl_clear_params(p);
 
 	/*
 	 * Since this might be the only -deadline task on the rq,
@@ -1741,8 +1781,7 @@ static void switched_from_dl(struct rq *rq, struct task_struct *p)
 	if (!task_on_rq_queued(p) || rq->dl.dl_nr_running)
 		return;
 
-	if (pull_dl_task(rq))
-		resched_curr(rq);
+	queue_pull_task(rq);
 }
 
 /*
@@ -1751,21 +1790,16 @@ static void switched_from_dl(struct rq *rq, struct task_struct *p)
  */
 static void switched_to_dl(struct rq *rq, struct task_struct *p)
 {
-	int check_resched = 1;
-
 	if (task_on_rq_queued(p) && rq->curr != p) {
 #ifdef CONFIG_SMP
-		if (p->nr_cpus_allowed > 1 && rq->dl.overloaded &&
-			push_dl_task(rq) && rq != task_rq(p))
-			/* Only reschedule if pushing failed */
-			check_resched = 0;
-#endif /* CONFIG_SMP */
-		if (check_resched) {
-			if (dl_task(rq->curr))
-				check_preempt_curr_dl(rq, p, 0);
-			else
-				resched_curr(rq);
-		}
+		if (p->nr_cpus_allowed > 1 && rq->dl.overloaded)
+			queue_push_tasks(rq);
+#else
+		if (dl_task(rq->curr))
+			check_preempt_curr_dl(rq, p, 0);
+		else
+			resched_curr(rq);
+#endif
 	}
 }
 
@@ -1785,15 +1819,14 @@ static void prio_changed_dl(struct rq *rq, struct task_struct *p,
 		 * or lowering its prio, so...
 		 */
 		if (!rq->dl.overloaded)
-			pull_dl_task(rq);
+			queue_pull_task(rq);
 
 		/*
 		 * If we now have a earlier deadline task than p,
 		 * then reschedule, provided p is still on this
 		 * runqueue.
 		 */
-		if (dl_time_before(rq->dl.earliest_dl.curr, p->dl.deadline) &&
-		    rq->curr == p)
+		if (dl_time_before(rq->dl.earliest_dl.curr, p->dl.deadline))
 			resched_curr(rq);
 #else
 		/*
@@ -1823,7 +1856,6 @@ const struct sched_class dl_sched_class = {
 	.set_cpus_allowed       = set_cpus_allowed_dl,
 	.rq_online              = rq_online_dl,
 	.rq_offline             = rq_offline_dl,
-	.post_schedule		= post_schedule_dl,
 	.task_woken		= task_woken_dl,
 #endif
 
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 40a7fcbf491e..3d57cc0ca0a6 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -5392,7 +5392,15 @@ simple:
 	return p;
 
 idle:
+	/*
+	 * This is OK, because current is on_cpu, which avoids it being picked
+	 * for load-balance and preemption/IRQs are still disabled avoiding
+	 * further scheduler activity on it and we're being very careful to
+	 * re-start the picking loop.
+	 */
+	lockdep_unpin_lock(&rq->lock);
 	new_tasks = idle_balance(rq);
+	lockdep_pin_lock(&rq->lock);
 	/*
 	 * Because idle_balance() releases (and re-acquires) rq->lock, it is
 	 * possible for any higher priority task to appear. In that case we
@@ -7426,9 +7434,6 @@ static int idle_balance(struct rq *this_rq)
 		goto out;
 	}
 
-	/*
-	 * Drop the rq->lock, but keep IRQ/preempt disabled.
-	 */
 	raw_spin_unlock(&this_rq->lock);
 
 	update_blocked_averages(this_cpu);
diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
index 7d7093c51f8d..0d193a243e96 100644
--- a/kernel/sched/rt.c
+++ b/kernel/sched/rt.c
@@ -260,7 +260,7 @@ int alloc_rt_sched_group(struct task_group *tg, struct task_group *parent)
 
 #ifdef CONFIG_SMP
 
-static int pull_rt_task(struct rq *this_rq);
+static void pull_rt_task(struct rq *this_rq);
 
 static inline bool need_pull_rt_task(struct rq *rq, struct task_struct *prev)
 {
@@ -354,13 +354,23 @@ static inline int has_pushable_tasks(struct rq *rq)
 	return !plist_head_empty(&rq->rt.pushable_tasks);
 }
 
-static inline void set_post_schedule(struct rq *rq)
+static DEFINE_PER_CPU(struct callback_head, rt_push_head);
+static DEFINE_PER_CPU(struct callback_head, rt_pull_head);
+
+static void push_rt_tasks(struct rq *);
+static void pull_rt_task(struct rq *);
+
+static inline void queue_push_tasks(struct rq *rq)
 {
-	/*
-	 * We detect this state here so that we can avoid taking the RQ
-	 * lock again later if there is no need to push
-	 */
-	rq->post_schedule = has_pushable_tasks(rq);
+	if (!has_pushable_tasks(rq))
+		return;
+
+	queue_balance_callback(rq, &per_cpu(rt_push_head, rq->cpu), push_rt_tasks);
+}
+
+static inline void queue_pull_task(struct rq *rq)
+{
+	queue_balance_callback(rq, &per_cpu(rt_pull_head, rq->cpu), pull_rt_task);
 }
 
 static void enqueue_pushable_task(struct rq *rq, struct task_struct *p)
@@ -412,12 +422,11 @@ static inline bool need_pull_rt_task(struct rq *rq, struct task_struct *prev)
 	return false;
 }
 
-static inline int pull_rt_task(struct rq *this_rq)
+static inline void pull_rt_task(struct rq *this_rq)
 {
-	return 0;
 }
 
-static inline void set_post_schedule(struct rq *rq)
+static inline void queue_push_tasks(struct rq *rq)
 {
 }
 #endif /* CONFIG_SMP */
@@ -1469,7 +1478,15 @@ pick_next_task_rt(struct rq *rq, struct task_struct *prev)
 	struct rt_rq *rt_rq = &rq->rt;
 
 	if (need_pull_rt_task(rq, prev)) {
+		/*
+		 * This is OK, because current is on_cpu, which avoids it being
+		 * picked for load-balance and preemption/IRQs are still
+		 * disabled avoiding further scheduler activity on it and we're
+		 * being very careful to re-start the picking loop.
+		 */
+		lockdep_unpin_lock(&rq->lock);
 		pull_rt_task(rq);
+		lockdep_pin_lock(&rq->lock);
 		/*
 		 * pull_rt_task() can drop (and re-acquire) rq->lock; this
 		 * means a dl or stop task can slip in, in which case we need
@@ -1497,7 +1514,7 @@ pick_next_task_rt(struct rq *rq, struct task_struct *prev)
 	/* The running task is never eligible for pushing */
 	dequeue_pushable_task(rq, p);
 
-	set_post_schedule(rq);
+	queue_push_tasks(rq);
 
 	return p;
 }
@@ -1952,14 +1969,15 @@ static void push_irq_work_func(struct irq_work *work)
 }
 #endif /* HAVE_RT_PUSH_IPI */
 
-static int pull_rt_task(struct rq *this_rq)
+static void pull_rt_task(struct rq *this_rq)
 {
-	int this_cpu = this_rq->cpu, ret = 0, cpu;
+	int this_cpu = this_rq->cpu, cpu;
+	bool resched = false;
 	struct task_struct *p;
 	struct rq *src_rq;
 
 	if (likely(!rt_overloaded(this_rq)))
-		return 0;
+		return;
 
 	/*
 	 * Match the barrier from rt_set_overloaded; this guarantees that if we
@@ -1970,7 +1988,7 @@ static int pull_rt_task(struct rq *this_rq)
 #ifdef HAVE_RT_PUSH_IPI
 	if (sched_feat(RT_PUSH_IPI)) {
 		tell_cpu_to_push(this_rq);
-		return 0;
+		return;
 	}
 #endif
 
@@ -2023,7 +2041,7 @@ static int pull_rt_task(struct rq *this_rq)
 			if (p->prio < src_rq->curr->prio)
 				goto skip;
 
-			ret = 1;
+			resched = true;
 
 			deactivate_task(src_rq, p, 0);
 			set_task_cpu(p, this_cpu);
@@ -2039,12 +2057,8 @@ skip:
 		double_unlock_balance(this_rq, src_rq);
 	}
 
-	return ret;
-}
-
-static void post_schedule_rt(struct rq *rq)
-{
-	push_rt_tasks(rq);
+	if (resched)
+		resched_curr(this_rq);
 }
 
 /*
@@ -2140,8 +2154,7 @@ static void switched_from_rt(struct rq *rq, struct task_struct *p)
 	if (!task_on_rq_queued(p) || rq->rt.rt_nr_running)
 		return;
 
-	if (pull_rt_task(rq))
-		resched_curr(rq);
+	queue_pull_task(rq);
 }
 
 void __init init_sched_rt_class(void)
@@ -2162,8 +2175,6 @@ void __init init_sched_rt_class(void)
  */
 static void switched_to_rt(struct rq *rq, struct task_struct *p)
 {
-	int check_resched = 1;
-
 	/*
 	 * If we are already running, then there's nothing
 	 * that needs to be done. But if we are not running
@@ -2173,13 +2184,12 @@ static void switched_to_rt(struct rq *rq, struct task_struct *p)
 	 */
 	if (task_on_rq_queued(p) && rq->curr != p) {
 #ifdef CONFIG_SMP
-		if (p->nr_cpus_allowed > 1 && rq->rt.overloaded &&
-		    /* Don't resched if we changed runqueues */
-		    push_rt_task(rq) && rq != task_rq(p))
-			check_resched = 0;
-#endif /* CONFIG_SMP */
-		if (check_resched && p->prio < rq->curr->prio)
+		if (p->nr_cpus_allowed > 1 && rq->rt.overloaded)
+			queue_push_tasks(rq);
+#else
+		if (p->prio < rq->curr->prio)
 			resched_curr(rq);
+#endif /* CONFIG_SMP */
 	}
 }
 
@@ -2200,14 +2210,13 @@ prio_changed_rt(struct rq *rq, struct task_struct *p, int oldprio)
 		 * may need to pull tasks to this runqueue.
 		 */
 		if (oldprio < p->prio)
-			pull_rt_task(rq);
+			queue_pull_task(rq);
+
 		/*
 		 * If there's a higher priority task waiting to run
-		 * then reschedule. Note, the above pull_rt_task
-		 * can release the rq lock and p could migrate.
-		 * Only reschedule if p is still on the same runqueue.
+		 * then reschedule.
 		 */
-		if (p->prio > rq->rt.highest_prio.curr && rq->curr == p)
+		if (p->prio > rq->rt.highest_prio.curr)
 			resched_curr(rq);
 #else
 		/* For UP simply resched on drop of prio */
@@ -2318,7 +2327,6 @@ const struct sched_class rt_sched_class = {
 	.set_cpus_allowed       = set_cpus_allowed_rt,
 	.rq_online              = rq_online_rt,
 	.rq_offline             = rq_offline_rt,
-	.post_schedule		= post_schedule_rt,
 	.task_woken		= task_woken_rt,
 	.switched_from		= switched_from_rt,
 #endif
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index aea7c1f393cb..885889190a1f 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -624,9 +624,10 @@ struct rq {
 	unsigned long cpu_capacity;
 	unsigned long cpu_capacity_orig;
 
+	struct callback_head *balance_callback;
+
 	unsigned char idle_balance;
 	/* For active balancing */
-	int post_schedule;
 	int active_balance;
 	int push_cpu;
 	struct cpu_stop_work active_balance_work;
@@ -767,6 +768,21 @@ extern int migrate_swap(struct task_struct *, struct task_struct *);
 
 #ifdef CONFIG_SMP
 
+static inline void
+queue_balance_callback(struct rq *rq,
+		       struct callback_head *head,
+		       void (*func)(struct rq *rq))
+{
+	lockdep_assert_held(&rq->lock);
+
+	if (unlikely(head->next))
+		return;
+
+	head->func = (void (*)(struct callback_head *))func;
+	head->next = rq->balance_callback;
+	rq->balance_callback = head;
+}
+
 extern void sched_ttwu_pending(void);
 
 #define rcu_dereference_check_sched_domain(p) \
@@ -1192,7 +1208,6 @@ struct sched_class {
 	int  (*select_task_rq)(struct task_struct *p, int task_cpu, int sd_flag, int flags);
 	void (*migrate_task_rq)(struct task_struct *p, int next_cpu);
 
-	void (*post_schedule) (struct rq *this_rq);
 	void (*task_waking) (struct task_struct *task);
 	void (*task_woken) (struct rq *this_rq, struct task_struct *task);
 
@@ -1423,8 +1438,10 @@ static inline struct rq *__task_rq_lock(struct task_struct *p)
 	for (;;) {
 		rq = task_rq(p);
 		raw_spin_lock(&rq->lock);
-		if (likely(rq == task_rq(p) && !task_on_rq_migrating(p)))
+		if (likely(rq == task_rq(p) && !task_on_rq_migrating(p))) {
+			lockdep_pin_lock(&rq->lock);
 			return rq;
+		}
 		raw_spin_unlock(&rq->lock);
 
 		while (unlikely(task_on_rq_migrating(p)))
@@ -1461,8 +1478,10 @@ static inline struct rq *task_rq_lock(struct task_struct *p, unsigned long *flag
 		 * If we observe the new cpu in task_rq_lock, the acquire will
 		 * pair with the WMB to ensure we must then also see migrating.
 		 */
-		if (likely(rq == task_rq(p) && !task_on_rq_migrating(p)))
+		if (likely(rq == task_rq(p) && !task_on_rq_migrating(p))) {
+			lockdep_pin_lock(&rq->lock);
 			return rq;
+		}
 		raw_spin_unlock(&rq->lock);
 		raw_spin_unlock_irqrestore(&p->pi_lock, *flags);
 
@@ -1474,6 +1493,7 @@ static inline struct rq *task_rq_lock(struct task_struct *p, unsigned long *flag
 static inline void __task_rq_unlock(struct rq *rq)
 	__releases(rq->lock)
 {
+	lockdep_unpin_lock(&rq->lock);
 	raw_spin_unlock(&rq->lock);
 }
 
@@ -1482,6 +1502,7 @@ task_rq_unlock(struct rq *rq, struct task_struct *p, unsigned long *flags)
 	__releases(rq->lock)
 	__releases(p->pi_lock)
 {
+	lockdep_unpin_lock(&rq->lock);
 	raw_spin_unlock(&rq->lock);
 	raw_spin_unlock_irqrestore(&p->pi_lock, *flags);
 }