1 files changed, 464 insertions, 94 deletions

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 5e973efc036e..71c0bb1cdeb0 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -272,7 +272,11 @@ late_initcall(sched_init_debug);
  * Number of tasks to iterate in a single balance run.
  * Limited because this is done with IRQs disabled.
  */
+#ifndef CONFIG_PREEMPT_RT_FULL
 const_debug unsigned int sysctl_sched_nr_migrate = 32;
+#else
+const_debug unsigned int sysctl_sched_nr_migrate = 8;
+#endif
 
 /*
  * period over which we average the RT time consumption, measured
@@ -489,6 +493,7 @@ static void init_rq_hrtick(struct rq *rq)
 
 	hrtimer_init(&rq->hrtick_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
 	rq->hrtick_timer.function = hrtick;
+	rq->hrtick_timer.irqsafe = 1;
 }
 #else	/* CONFIG_SCHED_HRTICK */
 static inline void hrtick_clear(struct rq *rq)
@@ -534,6 +539,37 @@ void resched_task(struct task_struct *p)
 		smp_send_reschedule(cpu);
 }
 
+#ifdef CONFIG_PREEMPT_LAZY
+void resched_task_lazy(struct task_struct *p)
+{
+	int cpu;
+
+	if (!sched_feat(PREEMPT_LAZY)) {
+		resched_task(p);
+		return;
+	}
+
+	lockdep_assert_held(&task_rq(p)->lock);
+
+	if (test_tsk_need_resched(p))
+		return;
+
+	if (test_tsk_need_resched_lazy(p))
+		return;
+
+	set_tsk_need_resched_lazy(p);
+
+	cpu = task_cpu(p);
+	if (cpu == smp_processor_id())
+		return;
+
+	/* NEED_RESCHED_LAZY must be visible before we test polling */
+	smp_mb();
+	if (!tsk_is_polling(p))
+		smp_send_reschedule(cpu);
+}
+#endif
+
 void resched_cpu(int cpu)
 {
 	struct rq *rq = cpu_rq(cpu);
@@ -660,17 +696,34 @@ static inline bool got_nohz_idle_kick(void)
 #endif /* CONFIG_NO_HZ_COMMON */
 
 #ifdef CONFIG_NO_HZ_FULL
+
+static int ksoftirqd_running(void)
+{
+	struct task_struct *softirqd;
+
+	if (!IS_ENABLED(CONFIG_PREEMPT_RT_FULL))
+		return 0;
+	softirqd = this_cpu_ksoftirqd();
+	if (softirqd && softirqd->on_rq)
+		return 1;
+	return 0;
+}
+
 bool sched_can_stop_tick(void)
 {
-       struct rq *rq;
+	struct rq *rq;
 
-       rq = this_rq();
+	rq = this_rq();
 
-       /* Make sure rq->nr_running update is visible after the IPI */
-       smp_rmb();
+	/* Make sure rq->nr_running update is visible after the IPI */
+	smp_rmb();
 
-       /* More than one running task need preemption */
-       if (rq->nr_running > 1)
+	/*
+	 * More than one running task need preemption
+	 *
+	 * NOTE, RT: if ksoftirqd is awake, subtract it.
+	 */
+	if (rq->nr_running - ksoftirqd_running() > 1)
                return false;
 
        return true;
@@ -1122,6 +1175,18 @@ struct migration_arg {
 
 static int migration_cpu_stop(void *data);
 
+static bool check_task_state(struct task_struct *p, long match_state)
+{
+	bool match = false;
+
+	raw_spin_lock_irq(&p->pi_lock);
+	if (p->state == match_state || p->saved_state == match_state)
+		match = true;
+	raw_spin_unlock_irq(&p->pi_lock);
+
+	return match;
+}
+
 /*
  * wait_task_inactive - wait for a thread to unschedule.
  *
@@ -1166,7 +1231,7 @@ unsigned long wait_task_inactive(struct task_struct *p, long match_state)
 		 * is actually now running somewhere else!
 		 */
 		while (task_running(rq, p)) {
-			if (match_state && unlikely(p->state != match_state))
+			if (match_state && !check_task_state(p, match_state))
 				return 0;
 			cpu_relax();
 		}
@@ -1181,7 +1246,8 @@ unsigned long wait_task_inactive(struct task_struct *p, long match_state)
 		running = task_running(rq, p);
 		on_rq = p->on_rq;
 		ncsw = 0;
-		if (!match_state || p->state == match_state)
+		if (!match_state || p->state == match_state
+		    || p->saved_state == match_state)
 			ncsw = p->nvcsw | LONG_MIN; /* sets MSB */
 		task_rq_unlock(rq, p, &flags);
 
@@ -1406,10 +1472,6 @@ static void ttwu_activate(struct rq *rq, struct task_struct *p, int en_flags)
 {
 	activate_task(rq, p, en_flags);
 	p->on_rq = 1;
-
-	/* if a worker is waking up, notify workqueue */
-	if (p->flags & PF_WQ_WORKER)
-		wq_worker_waking_up(p, cpu_of(rq));
 }
 
 /*
@@ -1592,8 +1654,27 @@ try_to_wake_up(struct task_struct *p, unsigned int state, int wake_flags)
 	 */
 	smp_mb__before_spinlock();
 	raw_spin_lock_irqsave(&p->pi_lock, flags);
-	if (!(p->state & state))
+	if (!(p->state & state)) {
+		/*
+		 * The task might be running due to a spinlock sleeper
+		 * wakeup. Check the saved state and set it to running
+		 * if the wakeup condition is true.
+		 */
+		if (!(wake_flags & WF_LOCK_SLEEPER)) {
+			if (p->saved_state & state) {
+				p->saved_state = TASK_RUNNING;
+				success = 1;
+			}
+		}
 		goto out;
+	}
+
+	/*
+	 * If this is a regular wakeup, then we can unconditionally
+	 * clear the saved state of a "lock sleeper".
+	 */
+	if (!(wake_flags & WF_LOCK_SLEEPER))
+		p->saved_state = TASK_RUNNING;
 
 	success = 1; /* we're going to change ->state */
 	cpu = task_cpu(p);
@@ -1636,42 +1717,6 @@ out:
 }
 
 /**
- * try_to_wake_up_local - try to wake up a local task with rq lock held
- * @p: the thread to be awakened
- *
- * Put @p on the run-queue if it's not already there. The caller must
- * ensure that this_rq() is locked, @p is bound to this_rq() and not
- * the current task.
- */
-static void try_to_wake_up_local(struct task_struct *p)
-{
-	struct rq *rq = task_rq(p);
-
-	if (WARN_ON_ONCE(rq != this_rq()) ||
-	    WARN_ON_ONCE(p == current))
-		return;
-
-	lockdep_assert_held(&rq->lock);
-
-	if (!raw_spin_trylock(&p->pi_lock)) {
-		raw_spin_unlock(&rq->lock);
-		raw_spin_lock(&p->pi_lock);
-		raw_spin_lock(&rq->lock);
-	}
-
-	if (!(p->state & TASK_NORMAL))
-		goto out;
-
-	if (!p->on_rq)
-		ttwu_activate(rq, p, ENQUEUE_WAKEUP);
-
-	ttwu_do_wakeup(rq, p, 0);
-	ttwu_stat(p, smp_processor_id(), 0);
-out:
-	raw_spin_unlock(&p->pi_lock);
-}
-
-/**
  * wake_up_process - Wake up a specific process
  * @p: The process to be woken up.
  *
@@ -1685,11 +1730,23 @@ out:
  */
 int wake_up_process(struct task_struct *p)
 {
-	WARN_ON(task_is_stopped_or_traced(p));
+	WARN_ON(__task_is_stopped_or_traced(p));
 	return try_to_wake_up(p, TASK_NORMAL, 0);
 }
 EXPORT_SYMBOL(wake_up_process);
 
+/**
+ * wake_up_lock_sleeper - Wake up a specific process blocked on a "sleeping lock"
+ * @p: The process to be woken up.
+ *
+ * Same as wake_up_process() above, but wake_flags=WF_LOCK_SLEEPER to indicate
+ * the nature of the wakeup.
+ */
+int wake_up_lock_sleeper(struct task_struct *p)
+{
+	return try_to_wake_up(p, TASK_ALL, WF_LOCK_SLEEPER);
+}
+
 int wake_up_state(struct task_struct *p, unsigned int state)
 {
 	return try_to_wake_up(p, state, 0);
@@ -1867,6 +1924,9 @@ int sched_fork(unsigned long clone_flags, struct task_struct *p)
 	p->on_cpu = 0;
 #endif
 	init_task_preempt_count(p);
+#ifdef CONFIG_HAVE_PREEMPT_LAZY
+	task_thread_info(p)->preempt_lazy_count = 0;
+#endif
 #ifdef CONFIG_SMP
 	plist_node_init(&p->pushable_tasks, MAX_PRIO);
 	RB_CLEAR_NODE(&p->pushable_dl_tasks);
@@ -2150,11 +2210,13 @@ static void finish_task_switch(struct rq *rq, struct task_struct *prev)
 	finish_arch_post_lock_switch();
 
 	fire_sched_in_preempt_notifiers(current);
+	/*
+	 * We use mmdrop_delayed() here so we don't have to do the
+	 * full __mmdrop() when we are the last user.
+	 */
 	if (mm)
-		mmdrop(mm);
+		mmdrop_delayed(mm);
 	if (unlikely(prev_state == TASK_DEAD)) {
-		task_numa_free(prev);
-
 		if (prev->sched_class->task_dead)
 			prev->sched_class->task_dead(prev);
 
@@ -2514,8 +2576,13 @@ void __kprobes preempt_count_add(int val)
 	DEBUG_LOCKS_WARN_ON((preempt_count() & PREEMPT_MASK) >=
 				PREEMPT_MASK - 10);
 #endif
-	if (preempt_count() == val)
-		trace_preempt_off(CALLER_ADDR0, get_parent_ip(CALLER_ADDR1));
+	if (preempt_count() == val) {
+		unsigned long ip = get_parent_ip(CALLER_ADDR1);
+#ifdef CONFIG_DEBUG_PREEMPT
+		current->preempt_disable_ip = ip;
+#endif
+		trace_preempt_off(CALLER_ADDR0, ip);
+	}
 }
 EXPORT_SYMBOL(preempt_count_add);
 
@@ -2558,6 +2625,13 @@ static noinline void __schedule_bug(struct task_struct *prev)
 	print_modules();
 	if (irqs_disabled())
 		print_irqtrace_events(prev);
+#ifdef CONFIG_DEBUG_PREEMPT
+	if (in_atomic_preempt_off()) {
+		pr_err("Preemption disabled at:");
+		print_ip_sym(current->preempt_disable_ip);
+		pr_cont("\n");
+	}
+#endif
 	dump_stack();
 	add_taint(TAINT_WARN, LOCKDEP_STILL_OK);
 }
@@ -2581,6 +2655,133 @@ static inline void schedule_debug(struct task_struct *prev)
 	schedstat_inc(this_rq(), sched_count);
 }
 
+#if defined(CONFIG_PREEMPT_RT_FULL) && defined(CONFIG_SMP)
+#define MIGRATE_DISABLE_SET_AFFIN	(1<<30) /* Can't make a negative */
+#define migrate_disabled_updated(p)	((p)->migrate_disable & MIGRATE_DISABLE_SET_AFFIN)
+#define migrate_disable_count(p)	((p)->migrate_disable & ~MIGRATE_DISABLE_SET_AFFIN)
+
+static inline void update_migrate_disable(struct task_struct *p)
+{
+	const struct cpumask *mask;
+
+	if (likely(!p->migrate_disable))
+		return;
+
+	/* Did we already update affinity? */
+	if (unlikely(migrate_disabled_updated(p)))
+		return;
+
+	/*
+	 * Since this is always current we can get away with only locking
+	 * rq->lock, the ->cpus_allowed value can normally only be changed
+	 * while holding both p->pi_lock and rq->lock, but seeing that this
+	 * is current, we cannot actually be waking up, so all code that
+	 * relies on serialization against p->pi_lock is out of scope.
+	 *
+	 * Having rq->lock serializes us against things like
+	 * set_cpus_allowed_ptr() that can still happen concurrently.
+	 */
+	mask = tsk_cpus_allowed(p);
+
+	if (p->sched_class->set_cpus_allowed)
+		p->sched_class->set_cpus_allowed(p, mask);
+	/* mask==cpumask_of(task_cpu(p)) which has a cpumask_weight==1 */
+	p->nr_cpus_allowed = 1;
+
+	/* Let migrate_enable know to fix things back up */
+	p->migrate_disable |= MIGRATE_DISABLE_SET_AFFIN;
+}
+
+void migrate_disable(void)
+{
+	struct task_struct *p = current;
+
+	if (in_atomic()) {
+#ifdef CONFIG_SCHED_DEBUG
+		p->migrate_disable_atomic++;
+#endif
+		return;
+	}
+
+#ifdef CONFIG_SCHED_DEBUG
+	if (unlikely(p->migrate_disable_atomic)) {
+		tracing_off();
+		WARN_ON_ONCE(1);
+	}
+#endif
+
+	if (p->migrate_disable) {
+		p->migrate_disable++;
+		return;
+	}
+
+	preempt_disable();
+	preempt_lazy_disable();
+	pin_current_cpu();
+	p->migrate_disable = 1;
+	preempt_enable();
+}
+EXPORT_SYMBOL(migrate_disable);
+
+void migrate_enable(void)
+{
+	struct task_struct *p = current;
+	const struct cpumask *mask;
+	unsigned long flags;
+	struct rq *rq;
+
+	if (in_atomic()) {
+#ifdef CONFIG_SCHED_DEBUG
+		p->migrate_disable_atomic--;
+#endif
+		return;
+	}
+
+#ifdef CONFIG_SCHED_DEBUG
+	if (unlikely(p->migrate_disable_atomic)) {
+		tracing_off();
+		WARN_ON_ONCE(1);
+	}
+#endif
+	WARN_ON_ONCE(p->migrate_disable <= 0);
+
+	if (migrate_disable_count(p) > 1) {
+		p->migrate_disable--;
+		return;
+	}
+
+	preempt_disable();
+	if (unlikely(migrate_disabled_updated(p))) {
+		/*
+		 * Undo whatever update_migrate_disable() did, also see there
+		 * about locking.
+		 */
+		rq = this_rq();
+		raw_spin_lock_irqsave(&rq->lock, flags);
+
+		/*
+		 * Clearing migrate_disable causes tsk_cpus_allowed to
+		 * show the tasks original cpu affinity.
+		 */
+		p->migrate_disable = 0;
+		mask = tsk_cpus_allowed(p);
+		if (p->sched_class->set_cpus_allowed)
+			p->sched_class->set_cpus_allowed(p, mask);
+		p->nr_cpus_allowed = cpumask_weight(mask);
+		raw_spin_unlock_irqrestore(&rq->lock, flags);
+	} else
+		p->migrate_disable = 0;
+
+	unpin_current_cpu();
+	preempt_enable();
+	preempt_lazy_enable();
+}
+EXPORT_SYMBOL(migrate_enable);
+#else
+static inline void update_migrate_disable(struct task_struct *p) { }
+#define migrate_disabled_updated(p)		0
+#endif
+
 static void put_prev_task(struct rq *rq, struct task_struct *prev)
 {
 	if (prev->on_rq || rq->skip_clock_update < 0)
@@ -2680,6 +2881,8 @@ need_resched:
 	smp_mb__before_spinlock();
 	raw_spin_lock_irq(&rq->lock);
 
+	update_migrate_disable(prev);
+
 	switch_count = &prev->nivcsw;
 	if (prev->state && !(preempt_count() & PREEMPT_ACTIVE)) {
 		if (unlikely(signal_pending_state(prev->state, prev))) {
@@ -2687,19 +2890,6 @@ need_resched:
 		} else {
 			deactivate_task(rq, prev, DEQUEUE_SLEEP);
 			prev->on_rq = 0;
-
-			/*
-			 * If a worker went to sleep, notify and ask workqueue
-			 * whether it wants to wake up a task to maintain
-			 * concurrency.
-			 */
-			if (prev->flags & PF_WQ_WORKER) {
-				struct task_struct *to_wakeup;
-
-				to_wakeup = wq_worker_sleeping(prev, cpu);
-				if (to_wakeup)
-					try_to_wake_up_local(to_wakeup);
-			}
 		}
 		switch_count = &prev->nvcsw;
 	}
@@ -2712,6 +2902,7 @@ need_resched:
 	put_prev_task(rq, prev);
 	next = pick_next_task(rq);
 	clear_tsk_need_resched(prev);
+	clear_tsk_need_resched_lazy(prev);
 	clear_preempt_need_resched();
 	rq->skip_clock_update = 0;
 
@@ -2741,8 +2932,19 @@ need_resched:
 
 static inline void sched_submit_work(struct task_struct *tsk)
 {
-	if (!tsk->state || tsk_is_pi_blocked(tsk))
+	if (!tsk->state)
+		return;
+	/*
+	 * If a worker went to sleep, notify and ask workqueue whether
+	 * it wants to wake up a task to maintain concurrency.
+	 */
+	if (tsk->flags & PF_WQ_WORKER)
+		wq_worker_sleeping(tsk);
+
+
+	if (tsk_is_pi_blocked(tsk))
 		return;
+
 	/*
 	 * If we are going to sleep and we have plugged IO queued,
 	 * make sure to submit it to avoid deadlocks.
@@ -2751,12 +2953,19 @@ static inline void sched_submit_work(struct task_struct *tsk)
 		blk_schedule_flush_plug(tsk);
 }
 
+static inline void sched_update_worker(struct task_struct *tsk)
+{
+	if (tsk->flags & PF_WQ_WORKER)
+		wq_worker_running(tsk);
+}
+
 asmlinkage void __sched schedule(void)
 {
 	struct task_struct *tsk = current;
 
 	sched_submit_work(tsk);
 	__schedule();
+	sched_update_worker(tsk);
 }
 EXPORT_SYMBOL(schedule);
 
@@ -2802,9 +3011,26 @@ asmlinkage void __sched notrace preempt_schedule(void)
 	if (likely(!preemptible()))
 		return;
 
+#ifdef CONFIG_PREEMPT_LAZY
+	/*
+	 * Check for lazy preemption
+	 */
+	if (current_thread_info()->preempt_lazy_count &&
+			!test_thread_flag(TIF_NEED_RESCHED))
+		return;
+#endif
 	do {
 		__preempt_count_add(PREEMPT_ACTIVE);
+		/*
+		 * The add/subtract must not be traced by the function
+		 * tracer. But we still want to account for the
+		 * preempt off latency tracer. Since the _notrace versions
+		 * of add/subtract skip the accounting for latency tracer
+		 * we must force it manually.
+		 */
+		start_critical_timings();
 		__schedule();
+		stop_critical_timings();
 		__preempt_count_sub(PREEMPT_ACTIVE);
 
 		/*
@@ -2912,7 +3138,8 @@ EXPORT_SYMBOL(sleep_on_timeout);
  * This function changes the 'effective' priority of a task. It does
  * not touch ->normal_prio like __setscheduler().
  *
- * Used by the rt_mutex code to implement priority inheritance logic.
+ * Used by the rt_mutex code to implement priority inheritance
+ * logic. Call site only calls if the priority of the task changed.
  */
 void rt_mutex_setprio(struct task_struct *p, int prio)
 {
@@ -3195,9 +3422,8 @@ __setparam_dl(struct task_struct *p, const struct sched_attr *attr)
 	dl_se->dl_new = 1;
 }
 
-/* Actually do priority change: must hold pi & rq lock. */
-static void __setscheduler(struct rq *rq, struct task_struct *p,
-			   const struct sched_attr *attr)
+static void __setscheduler_params(struct task_struct *p,
+		const struct sched_attr *attr)
 {
 	int policy = attr->sched_policy;
 
@@ -3217,9 +3443,21 @@ static void __setscheduler(struct rq *rq, struct task_struct *p,
 	 * getparam()/getattr() don't report silly values for !rt tasks.
 	 */
 	p->rt_priority = attr->sched_priority;
-
 	p->normal_prio = normal_prio(p);
-	p->prio = rt_mutex_getprio(p);
+	set_load_weight(p);
+}
+
+/* Actually do priority change: must hold pi & rq lock. */
+static void __setscheduler(struct rq *rq, struct task_struct *p,
+			   const struct sched_attr *attr)
+{
+	__setscheduler_params(p, attr);
+
+	/*
+	 * If we get here, there was no pi waiters boosting the
+	 * task. It is safe to use the normal prio.
+	 */
+	p->prio = normal_prio(p);
 
 	if (dl_prio(p->prio))
 		p->sched_class = &dl_sched_class;
@@ -3227,8 +3465,6 @@ static void __setscheduler(struct rq *rq, struct task_struct *p,
 		p->sched_class = &rt_sched_class;
 	else
 		p->sched_class = &fair_sched_class;
-
-	set_load_weight(p);
 }
 
 static void
@@ -3304,6 +3540,8 @@ static int __sched_setscheduler(struct task_struct *p,
 				const struct sched_attr *attr,
 				bool user)
 {
+	int newprio = dl_policy(attr->sched_policy) ? MAX_DL_PRIO - 1 :
+		      MAX_RT_PRIO - 1 - attr->sched_priority;
 	int retval, oldprio, oldpolicy = -1, on_rq, running;
 	int policy = attr->sched_policy;
 	unsigned long flags;
@@ -3428,6 +3666,7 @@ recheck:
 		if (dl_policy(policy))
 			goto change;
 
+		p->sched_reset_on_fork = reset_on_fork;
 		task_rq_unlock(rq, p, &flags);
 		return 0;
 	}
@@ -3481,6 +3720,24 @@ change:
 		return -EBUSY;
 	}
 
+	p->sched_reset_on_fork = reset_on_fork;
+	oldprio = p->prio;
+
+	/*
+	 * Special case for priority boosted tasks.
+	 *
+	 * If the new priority is lower or equal (user space view)
+	 * than the current (boosted) priority, 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.
+	 */
+	if (rt_mutex_check_prio(p, newprio)) {
+		__setscheduler_params(p, attr);
+		task_rq_unlock(rq, p, &flags);
+		return 0;
+	}
+
 	on_rq = p->on_rq;
 	running = task_current(rq, p);
 	if (on_rq)
@@ -3488,16 +3745,18 @@ change:
 	if (running)
 		p->sched_class->put_prev_task(rq, p);
 
-	p->sched_reset_on_fork = reset_on_fork;
-
-	oldprio = p->prio;
 	prev_class = p->sched_class;
 	__setscheduler(rq, p, attr);
 
 	if (running)
 		p->sched_class->set_curr_task(rq);
-	if (on_rq)
-		enqueue_task(rq, p, 0);
+	if (on_rq) {
+		/*
+		 * We enqueue to tail when the priority of a task is
+		 * increased (user space view).
+		 */
+		enqueue_task(rq, p, oldprio <= p->prio ? ENQUEUE_HEAD : 0);
+	}
 
 	check_class_changed(rq, p, prev_class, oldprio);
 	task_rq_unlock(rq, p, &flags);
@@ -4106,9 +4365,16 @@ SYSCALL_DEFINE0(sched_yield)
 
 static void __cond_resched(void)
 {
-	__preempt_count_add(PREEMPT_ACTIVE);
-	__schedule();
-	__preempt_count_sub(PREEMPT_ACTIVE);
+	do {
+		__preempt_count_add(PREEMPT_ACTIVE);
+		__schedule();
+		__preempt_count_sub(PREEMPT_ACTIVE);
+		/*
+		 * Check again in case we missed a preemption
+		 * opportunity between schedule and now.
+		 */
+		barrier();
+	} while (need_resched());
 }
 
 int __sched _cond_resched(void)
@@ -4149,6 +4415,7 @@ int __cond_resched_lock(spinlock_t *lock)
 }
 EXPORT_SYMBOL(__cond_resched_lock);
 
+#ifndef CONFIG_PREEMPT_RT_FULL
 int __sched __cond_resched_softirq(void)
 {
 	BUG_ON(!in_softirq());
@@ -4162,6 +4429,7 @@ int __sched __cond_resched_softirq(void)
 	return 0;
 }
 EXPORT_SYMBOL(__cond_resched_softirq);
+#endif
 
 /**
  * yield - yield the current processor to other threads.
@@ -4515,6 +4783,7 @@ void init_idle(struct task_struct *idle, int cpu)
 	rcu_read_unlock();
 
 	rq->curr = rq->idle = idle;
+	idle->on_rq = 1;
 #if defined(CONFIG_SMP)
 	idle->on_cpu = 1;
 #endif
@@ -4522,7 +4791,9 @@ void init_idle(struct task_struct *idle, int cpu)
 
 	/* Set the preempt count _outside_ the spinlocks! */
 	init_idle_preempt_count(idle, cpu);
-
+#ifdef CONFIG_HAVE_PREEMPT_LAZY
+	task_thread_info(idle)->preempt_lazy_count = 0;
+#endif
 	/*
 	 * The idle tasks have their own, simple scheduling class:
 	 */
@@ -4537,11 +4808,90 @@ void init_idle(struct task_struct *idle, int cpu)
 #ifdef CONFIG_SMP
 void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask)
 {
-	if (p->sched_class && p->sched_class->set_cpus_allowed)
-		p->sched_class->set_cpus_allowed(p, new_mask);
-
+	if (!migrate_disabled_updated(p)) {
+		if (p->sched_class && p->sched_class->set_cpus_allowed)
+			p->sched_class->set_cpus_allowed(p, new_mask);
+		p->nr_cpus_allowed = cpumask_weight(new_mask);
+	}
 	cpumask_copy(&p->cpus_allowed, new_mask);
-	p->nr_cpus_allowed = cpumask_weight(new_mask);
+}
+
+static DEFINE_PER_CPU(struct cpumask, sched_cpumasks);
+static DEFINE_MUTEX(sched_down_mutex);
+static cpumask_t sched_down_cpumask;
+
+void tell_sched_cpu_down_begin(int cpu)
+{
+	mutex_lock(&sched_down_mutex);
+	cpumask_set_cpu(cpu, &sched_down_cpumask);
+	mutex_unlock(&sched_down_mutex);
+}
+
+void tell_sched_cpu_down_done(int cpu)
+{
+	mutex_lock(&sched_down_mutex);
+	cpumask_clear_cpu(cpu, &sched_down_cpumask);
+	mutex_unlock(&sched_down_mutex);
+}
+
+/**
+ * migrate_me - try to move the current task off this cpu
+ *
+ * Used by the pin_current_cpu() code to try to get tasks
+ * to move off the current CPU as it is going down.
+ * It will only move the task if the task isn't pinned to
+ * the CPU (with migrate_disable, affinity or NO_SETAFFINITY)
+ * and the task has to be in a RUNNING state. Otherwise the
+ * movement of the task will wake it up (change its state
+ * to running) when the task did not expect it.
+ *
+ * Returns 1 if it succeeded in moving the current task
+ *         0 otherwise.
+ */
+int migrate_me(void)
+{
+	struct task_struct *p = current;
+	struct migration_arg arg;
+	struct cpumask *cpumask;
+	struct cpumask *mask;
+	unsigned long flags;
+	unsigned int dest_cpu;
+	struct rq *rq;
+
+	/*
+	 * We can not migrate tasks bounded to a CPU or tasks not
+	 * running. The movement of the task will wake it up.
+	 */
+	if (p->flags & PF_NO_SETAFFINITY || p->state)
+		return 0;
+
+	mutex_lock(&sched_down_mutex);
+	rq = task_rq_lock(p, &flags);
+
+	cpumask = &__get_cpu_var(sched_cpumasks);
+	mask = &p->cpus_allowed;
+
+	cpumask_andnot(cpumask, mask, &sched_down_cpumask);
+
+	if (!cpumask_weight(cpumask)) {
+		/* It's only on this CPU? */
+		task_rq_unlock(rq, p, &flags);
+		mutex_unlock(&sched_down_mutex);
+		return 0;
+	}
+
+	dest_cpu = cpumask_any_and(cpu_active_mask, cpumask);
+
+	arg.task = p;
+	arg.dest_cpu = dest_cpu;
+
+	task_rq_unlock(rq, p, &flags);
+
+	stop_one_cpu(cpu_of(rq), migration_cpu_stop, &arg);
+	tlb_migrate_finish(p->mm);
+	mutex_unlock(&sched_down_mutex);
+
+	return 1;
 }
 
 /*
@@ -4587,7 +4937,7 @@ int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask)
 	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))
+	if (cpumask_test_cpu(task_cpu(p), new_mask) || __migrate_disabled(p))
 		goto out;
 
 	dest_cpu = cpumask_any_and(cpu_active_mask, new_mask);
@@ -4724,6 +5074,8 @@ static int migration_cpu_stop(void *data)
 
 #ifdef CONFIG_HOTPLUG_CPU
 
+static DEFINE_PER_CPU(struct mm_struct *, idle_last_mm);
+
 /*
  * Ensures that the idle task is using init_mm right before its cpu goes
  * offline.
@@ -4736,7 +5088,12 @@ void idle_task_exit(void)
 
 	if (mm != &init_mm)
 		switch_mm(mm, &init_mm, current);
-	mmdrop(mm);
+
+	/*
+	 * Defer the cleanup to an alive cpu. On RT we can neither
+	 * call mmdrop() nor mmdrop_delayed() from here.
+	 */
+	per_cpu(idle_last_mm, smp_processor_id()) = mm;
 }
 
 /*
@@ -5060,6 +5417,10 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu)
 
 	case CPU_DEAD:
 		calc_load_migrate(rq);
+		if (per_cpu(idle_last_mm, cpu)) {
+			mmdrop(per_cpu(idle_last_mm, cpu));
+			per_cpu(idle_last_mm, cpu) = NULL;
+		}
 		break;
 #endif
 	}
@@ -6968,7 +7329,8 @@ void __init sched_init(void)
 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
 static inline int preempt_count_equals(int preempt_offset)
 {
-	int nested = (preempt_count() & ~PREEMPT_ACTIVE) + rcu_preempt_depth();
+	int nested = (preempt_count() & ~PREEMPT_ACTIVE) +
+		sched_rcu_preempt_depth();
 
 	return (nested == preempt_offset);
 }
@@ -6978,7 +7340,8 @@ void __might_sleep(const char *file, int line, int preempt_offset)
 	static unsigned long prev_jiffy;	/* ratelimiting */
 
 	rcu_sleep_check(); /* WARN_ON_ONCE() by default, no rate limit reqd. */
-	if ((preempt_count_equals(preempt_offset) && !irqs_disabled()) ||
+	if ((preempt_count_equals(preempt_offset) && !irqs_disabled() &&
+	     !is_idle_task(current)) ||
 	    system_state != SYSTEM_RUNNING || oops_in_progress)
 		return;
 	if (time_before(jiffies, prev_jiffy + HZ) && prev_jiffy)
@@ -6996,6 +7359,13 @@ void __might_sleep(const char *file, int line, int preempt_offset)
 	debug_show_held_locks(current);
 	if (irqs_disabled())
 		print_irqtrace_events(current);
+#ifdef CONFIG_DEBUG_PREEMPT
+	if (!preempt_count_equals(preempt_offset)) {
+		pr_err("Preemption disabled at:");
+		print_ip_sym(current->preempt_disable_ip);
+		pr_cont("\n");
+	}
+#endif
 	dump_stack();
 }
 EXPORT_SYMBOL(__might_sleep);