* 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
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)
#endif
#endif
+void wake_q_add(struct wake_q_head *head, struct task_struct *task)
+{
+ struct wake_q_node *node = &task->wake_q;
+
+ /*
+ * Atomically grab the task, if ->wake_q is !nil already it means
+ * its already queued (either by us or someone else) and will get the
+ * wakeup due to that.
+ *
+ * This cmpxchg() implies a full barrier, which pairs with the write
+ * barrier implied by the wakeup in wake_up_list().
+ */
+ if (cmpxchg(&node->next, NULL, WAKE_Q_TAIL))
+ return;
+
+ get_task_struct(task);
+
+ /*
+ * The head is context local, there can be no concurrency.
+ */
+ *head->lastp = node;
+ head->lastp = &node->next;
+}
+
+void wake_up_q(struct wake_q_head *head)
+{
+ struct wake_q_node *node = head->first;
+
+ while (node != WAKE_Q_TAIL) {
+ struct task_struct *task;
+
+ task = container_of(node, struct task_struct, wake_q);
+ BUG_ON(!task);
+ /* task can safely be re-inserted now */
+ node = node->next;
+ task->wake_q.next = NULL;
+
+ /*
+ * wake_up_process() implies a wmb() to pair with the queueing
+ * in wake_q_add() so as not to miss wakeups.
+ */
+ wake_up_process(task);
+ put_task_struct(task);
+ }
+}
+
/*
* resched_curr - mark rq's current task 'to be rescheduled now'.
*
trace_sched_wake_idle_without_ipi(cpu);
}
+#ifdef CONFIG_PREEMPT_LAZY
+void resched_curr_lazy(struct rq *rq)
+{
+ struct task_struct *curr = rq->curr;
+ int cpu;
+
+ if (!sched_feat(PREEMPT_LAZY)) {
+ resched_curr(rq);
+ return;
+ }
+
+ lockdep_assert_held(&rq->lock);
+
+ if (test_tsk_need_resched(curr))
+ return;
+
+ if (test_tsk_need_resched_lazy(curr))
+ return;
+
+ set_tsk_need_resched_lazy(curr);
+
+ cpu = cpu_of(rq);
+ if (cpu == smp_processor_id())
+ return;
+
+ /* NEED_RESCHED_LAZY must be visible before we test polling */
+ smp_mb();
+ if (!tsk_is_polling(curr))
+ smp_send_reschedule(cpu);
+}
+#endif
+
void resched_cpu(int cpu)
{
struct rq *rq = cpu_rq(cpu);
*/
int get_nohz_timer_target(int pinned)
{
- int cpu = smp_processor_id();
+ int cpu;
int i;
struct sched_domain *sd;
+ preempt_disable_rt();
+ cpu = smp_processor_id();
if (pinned || !get_sysctl_timer_migration() || !idle_cpu(cpu))
- return cpu;
+ goto preempt_en_rt;
rcu_read_lock();
for_each_domain(cpu, sd) {
}
unlock:
rcu_read_unlock();
+preempt_en_rt:
+ preempt_enable_rt();
return cpu;
}
/*
#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)
{
/*
* More than one running task need preemption.
* nr_running update is assumed to be visible
* after IPI is sent from wakers.
+ *
+ * NOTE, RT: if ksoftirqd is awake, subtract it.
*/
- if (this_rq()->nr_running > 1)
+ if (this_rq()->nr_running - ksoftirqd_running() > 1)
return false;
return true;
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.
*
* 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();
}
running = task_running(rq, p);
queued = task_on_rq_queued(p);
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);
{
activate_task(rq, p, en_flags);
p->on_rq = TASK_ON_RQ_QUEUED;
-
- /* if a worker is waking up, notify workqueue */
- if (p->flags & PF_WQ_WORKER)
- wq_worker_waking_up(p, cpu_of(rq));
}
/*
*/
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);
return success;
}
-/**
- * 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 (!task_on_rq_queued(p))
- 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.
*/
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);
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);
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)) {
if (prev->sched_class->task_dead)
prev->sched_class->task_dead(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
+
/*
* Pick up the highest-prio task:
*/
smp_mb__before_spinlock();
raw_spin_lock_irq(&rq->lock);
+ update_migrate_disable(prev);
+
rq->clock_skip_update <<= 1; /* promote REQ to ACT */
switch_count = &prev->nivcsw;
} 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;
}
next = pick_next_task(rq, prev);
clear_tsk_need_resched(prev);
+ clear_tsk_need_resched_lazy(prev);
clear_preempt_need_resched();
rq->clock_skip_update = 0;
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.
blk_schedule_flush_plug(tsk);
}
+static void sched_update_worker(struct task_struct *tsk)
+{
+ if (tsk->flags & PF_WQ_WORKER)
+ wq_worker_running(tsk);
+}
+
asmlinkage __visible void __sched schedule(void)
{
struct task_struct *tsk = current;
do {
__schedule();
} while (need_resched());
+ sched_update_worker(tsk);
}
EXPORT_SYMBOL(schedule);
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);
/*
* an infinite recursion.
*/
prev_ctx = exception_enter();
+ /*
+ * 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();
exception_exit(prev_ctx);
__preempt_count_sub(PREEMPT_ACTIVE);
}
EXPORT_SYMBOL(__cond_resched_lock);
+#ifndef CONFIG_PREEMPT_RT_FULL
int __sched __cond_resched_softirq(void)
{
BUG_ON(!in_softirq());
return 0;
}
EXPORT_SYMBOL(__cond_resched_softirq);
+#endif
/**
* yield - yield the current processor to other threads.
/* 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:
*/
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);
+ if (!migrate_disabled_updated(p)) {
+ if (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 = this_cpu_ptr(&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;
}
/*
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);
#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.
switch_mm(mm, &init_mm, current);
finish_arch_post_lock_switch();
}
- 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;
}
/*
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
}
#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);
}