Merge branches 'sched/urgent' and 'sched/rt' into sched/devel

kernel/cpuset.c

@@ -843,37 +843,25 @@ static void cpuset_change_cpumask(struct task_struct *tsk,
 /**
  * update_tasks_cpumask - Update the cpumasks of tasks in the cpuset.
  * @cs: the cpuset in which each task's cpus_allowed mask needs to be changed
+ * @heap: if NULL, defer allocating heap memory to cgroup_scan_tasks()
  *
  * Called with cgroup_mutex held
  *
  * The cgroup_scan_tasks() function will scan all the tasks in a cgroup,
  * calling callback functions for each.
  *
- * Return 0 if successful, -errno if not.
+ * No return value. It's guaranteed that cgroup_scan_tasks() always returns 0
+ * if @heap != NULL.
  */
-static int update_tasks_cpumask(struct cpuset *cs)
+static void update_tasks_cpumask(struct cpuset *cs, struct ptr_heap *heap)
 {
 	struct cgroup_scanner scan;
-	struct ptr_heap heap;
-	int retval;
-
-	/*
-	 * cgroup_scan_tasks() will initialize heap->gt for us.
-	 * heap_init() is still needed here for we should not change
-	 * cs->cpus_allowed when heap_init() fails.
-	 */
-	retval = heap_init(&heap, PAGE_SIZE, GFP_KERNEL, NULL);
-	if (retval)
-		return retval;
 
 	scan.cg = cs->css.cgroup;
 	scan.test_task = cpuset_test_cpumask;
 	scan.process_task = cpuset_change_cpumask;
-	scan.heap = &heap;
-	retval = cgroup_scan_tasks(&scan);
-
-	heap_free(&heap);
-	return retval;
+	scan.heap = heap;
+	cgroup_scan_tasks(&scan);
 }
 
 /**
@@ -883,6 +871,7 @@ static int update_tasks_cpumask(struct cpuset *cs)
  */
 static int update_cpumask(struct cpuset *cs, const char *buf)
 {
+	struct ptr_heap heap;
 	struct cpuset trialcs;
 	int retval;
 	int is_load_balanced;
@@ -917,6 +906,10 @@ static int update_cpumask(struct cpuset *cs, const char *buf)
 	if (cpus_equal(cs->cpus_allowed, trialcs.cpus_allowed))
 		return 0;
 
+	retval = heap_init(&heap, PAGE_SIZE, GFP_KERNEL, NULL);
+	if (retval)
+		return retval;
+
 	is_load_balanced = is_sched_load_balance(&trialcs);
 
 	mutex_lock(&callback_mutex);
@@ -927,9 +920,9 @@ static int update_cpumask(struct cpuset *cs, const char *buf)
 	 * Scan tasks in the cpuset, and update the cpumasks of any
 	 * that need an update.
 	 */
-	retval = update_tasks_cpumask(cs);
-	if (retval < 0)
-		return retval;
+	update_tasks_cpumask(cs, &heap);
+
+	heap_free(&heap);
 
 	if (is_load_balanced)
 		async_rebuild_sched_domains();
@@ -1965,7 +1958,7 @@ static void scan_for_empty_cpusets(const struct cpuset *root)
 		     nodes_empty(cp->mems_allowed))
 			remove_tasks_in_empty_cpuset(cp);
 		else {
-			update_tasks_cpumask(cp);
+			update_tasks_cpumask(cp, NULL);
 			update_tasks_nodemask(cp, &oldmems);
 		}
 	}

kernel/sched.c

@@ -204,11 +204,16 @@ void init_rt_bandwidth(struct rt_bandwidth *rt_b, u64 period, u64 runtime)
 	rt_b->rt_period_timer.cb_mode = HRTIMER_CB_IRQSAFE_NO_SOFTIRQ;
 }
 
+static inline int rt_bandwidth_enabled(void)
+{
+	return sysctl_sched_rt_runtime >= 0;
+}
+
 static void start_rt_bandwidth(struct rt_bandwidth *rt_b)
 {
 	ktime_t now;
 
-	if (rt_b->rt_runtime == RUNTIME_INF)
+	if (rt_bandwidth_enabled() && rt_b->rt_runtime == RUNTIME_INF)
 		return;
 
 	if (hrtimer_active(&rt_b->rt_period_timer))
@@ -298,9 +303,9 @@ static DEFINE_PER_CPU(struct cfs_rq, init_cfs_rq) ____cacheline_aligned_in_smp;
 static DEFINE_PER_CPU(struct sched_rt_entity, init_sched_rt_entity);
 static DEFINE_PER_CPU(struct rt_rq, init_rt_rq) ____cacheline_aligned_in_smp;
 #endif /* CONFIG_RT_GROUP_SCHED */
-#else /* !CONFIG_FAIR_GROUP_SCHED */
+#else /* !CONFIG_USER_SCHED */
 #define root_task_group init_task_group
-#endif /* CONFIG_FAIR_GROUP_SCHED */
+#endif /* CONFIG_USER_SCHED */
 
 /* task_group_lock serializes add/remove of task groups and also changes to
  * a task group's cpu shares.
@@ -1087,7 +1092,7 @@ hotplug_hrtick(struct notifier_block *nfb, unsigned long action, void *hcpu)
 	return NOTIFY_DONE;
 }
 
-static void init_hrtick(void)
+static __init void init_hrtick(void)
 {
 	hotcpu_notifier(hotplug_hrtick, 0);
 }
@@ -1380,6 +1385,51 @@ static inline void dec_cpu_load(struct rq *rq, unsigned long load)
 	update_load_sub(&rq->load, load);
 }
 
+#if (defined(CONFIG_SMP) && defined(CONFIG_FAIR_GROUP_SCHED)) || defined(CONFIG_RT_GROUP_SCHED)
+typedef int (*tg_visitor)(struct task_group *, void *);
+
+/*
+ * Iterate the full tree, calling @down when first entering a node and @up when
+ * leaving it for the final time.
+ */
+static int walk_tg_tree(tg_visitor down, tg_visitor up, void *data)
+{
+	struct task_group *parent, *child;
+	int ret;
+
+	rcu_read_lock();
+	parent = &root_task_group;
+down:
+	ret = (*down)(parent, data);
+	if (ret)
+		goto out_unlock;
+	list_for_each_entry_rcu(child, &parent->children, siblings) {
+		parent = child;
+		goto down;
+
+up:
+		continue;
+	}
+	ret = (*up)(parent, data);
+	if (ret)
+		goto out_unlock;
+
+	child = parent;
+	parent = parent->parent;
+	if (parent)
+		goto up;
+out_unlock:
+	rcu_read_unlock();
+
+	return ret;
+}
+
+static int tg_nop(struct task_group *tg, void *data)
+{
+	return 0;
+}
+#endif
+
 #ifdef CONFIG_SMP
 static unsigned long source_load(int cpu, int type);
 static unsigned long target_load(int cpu, int type);
@@ -1397,37 +1447,6 @@ static unsigned long cpu_avg_load_per_task(int cpu)
 
 #ifdef CONFIG_FAIR_GROUP_SCHED
 
-typedef void (*tg_visitor)(struct task_group *, int, struct sched_domain *);
-
-/*
- * Iterate the full tree, calling @down when first entering a node and @up when
- * leaving it for the final time.
- */
-static void
-walk_tg_tree(tg_visitor down, tg_visitor up, int cpu, struct sched_domain *sd)
-{
-	struct task_group *parent, *child;
-
-	rcu_read_lock();
-	parent = &root_task_group;
-down:
-	(*down)(parent, cpu, sd);
-	list_for_each_entry_rcu(child, &parent->children, siblings) {
-		parent = child;
-		goto down;
-
-up:
-		continue;
-	}
-	(*up)(parent, cpu, sd);
-
-	child = parent;
-	parent = parent->parent;
-	if (parent)
-		goto up;
-	rcu_read_unlock();
-}
-
 static void __set_se_shares(struct sched_entity *se, unsigned long shares);
 
 /*
@@ -1486,11 +1505,11 @@ __update_group_shares_cpu(struct task_group *tg, int cpu,
  * This needs to be done in a bottom-up fashion because the rq weight of a
  * parent group depends on the shares of its child groups.
  */
-static void
-tg_shares_up(struct task_group *tg, int cpu, struct sched_domain *sd)
+static int tg_shares_up(struct task_group *tg, void *data)
 {
 	unsigned long rq_weight = 0;
 	unsigned long shares = 0;
+	struct sched_domain *sd = data;
 	int i;
 
 	for_each_cpu_mask(i, sd->span) {
@@ -1515,6 +1534,8 @@ tg_shares_up(struct task_group *tg, int cpu, struct sched_domain *sd)
 		__update_group_shares_cpu(tg, i, shares, rq_weight);
 		spin_unlock_irqrestore(&rq->lock, flags);
 	}
+
+	return 0;
 }
 
 /*
@@ -1522,10 +1543,10 @@ tg_shares_up(struct task_group *tg, int cpu, struct sched_domain *sd)
  * This needs to be done in a top-down fashion because the load of a child
  * group is a fraction of its parents load.
  */
-static void
-tg_load_down(struct task_group *tg, int cpu, struct sched_domain *sd)
+static int tg_load_down(struct task_group *tg, void *data)
 {
 	unsigned long load;
+	long cpu = (long)data;
 
 	if (!tg->parent) {
 		load = cpu_rq(cpu)->load.weight;
@@ -1536,11 +1557,8 @@ tg_load_down(struct task_group *tg, int cpu, struct sched_domain *sd)
 	}
 
 	tg->cfs_rq[cpu]->h_load = load;
-}
 
-static void
-tg_nop(struct task_group *tg, int cpu, struct sched_domain *sd)
-{
+	return 0;
 }
 
 static void update_shares(struct sched_domain *sd)
@@ -1550,7 +1568,7 @@ static void update_shares(struct sched_domain *sd)
 
 	if (elapsed >= (s64)(u64)sysctl_sched_shares_ratelimit) {
 		sd->last_update = now;
-		walk_tg_tree(tg_nop, tg_shares_up, 0, sd);
+		walk_tg_tree(tg_nop, tg_shares_up, sd);
 	}
 }
 
@@ -1561,9 +1579,9 @@ static void update_shares_locked(struct rq *rq, struct sched_domain *sd)
 	spin_lock(&rq->lock);
 }
 
-static void update_h_load(int cpu)
+static void update_h_load(long cpu)
 {
-	walk_tg_tree(tg_load_down, tg_nop, cpu, NULL);
+	walk_tg_tree(tg_load_down, tg_nop, (void *)cpu);
 }
 
 #else
@@ -5171,7 +5189,8 @@ recheck:
 		 * Do not allow realtime tasks into groups that have no runtime
 		 * assigned.
 		 */
-		if (rt_policy(policy) && task_group(p)->rt_bandwidth.rt_runtime == 0)
+		if (rt_bandwidth_enabled() && rt_policy(policy) &&
+				task_group(p)->rt_bandwidth.rt_runtime == 0)
 			return -EPERM;
 #endif
 
@@ -8808,75 +8827,79 @@ static DEFINE_MUTEX(rt_constraints_mutex);
 static unsigned long to_ratio(u64 period, u64 runtime)
 {
 	if (runtime == RUNTIME_INF)
-		return 1ULL << 16;
+		return 1ULL << 20;
 
-	return div64_u64(runtime << 16, period);
+	return div64_u64(runtime << 20, period);
 }
 
-#ifdef CONFIG_CGROUP_SCHED
-static int __rt_schedulable(struct task_group *tg, u64 period, u64 runtime)
-{
-	struct task_group *tgi, *parent = tg->parent;
-	unsigned long total = 0;
-
-	if (!parent) {
-		if (global_rt_period() < period)
-			return 0;
-
-		return to_ratio(period, runtime) <
-			to_ratio(global_rt_period(), global_rt_runtime());
-	}
-
-	if (ktime_to_ns(parent->rt_bandwidth.rt_period) < period)
-		return 0;
-
-	rcu_read_lock();
-	list_for_each_entry_rcu(tgi, &parent->children, siblings) {
-		if (tgi == tg)
-			continue;
-
-		total += to_ratio(ktime_to_ns(tgi->rt_bandwidth.rt_period),
-				tgi->rt_bandwidth.rt_runtime);
-	}
-	rcu_read_unlock();
-
-	return total + to_ratio(period, runtime) <=
-		to_ratio(ktime_to_ns(parent->rt_bandwidth.rt_period),
-				parent->rt_bandwidth.rt_runtime);
-}
-#elif defined CONFIG_USER_SCHED
-static int __rt_schedulable(struct task_group *tg, u64 period, u64 runtime)
-{
-	struct task_group *tgi;
-	unsigned long total = 0;
-	unsigned long global_ratio =
-		to_ratio(global_rt_period(), global_rt_runtime());
-
-	rcu_read_lock();
-	list_for_each_entry_rcu(tgi, &task_groups, list) {
-		if (tgi == tg)
-			continue;
-
-		total += to_ratio(ktime_to_ns(tgi->rt_bandwidth.rt_period),
-				tgi->rt_bandwidth.rt_runtime);
-	}
-	rcu_read_unlock();
-
-	return total + to_ratio(period, runtime) < global_ratio;
-}
-#endif
-
 /* Must be called with tasklist_lock held */
 static inline int tg_has_rt_tasks(struct task_group *tg)
 {
 	struct task_struct *g, *p;
+
 	do_each_thread(g, p) {
 		if (rt_task(p) && rt_rq_of_se(&p->rt)->tg == tg)
 			return 1;
 	} while_each_thread(g, p);
+
 	return 0;
 }
 
+struct rt_schedulable_data {
+	struct task_group *tg;
+	u64 rt_period;
+	u64 rt_runtime;
+};
+
+static int tg_schedulable(struct task_group *tg, void *data)
+{
+	struct rt_schedulable_data *d = data;
+	struct task_group *child;
+	unsigned long total, sum = 0;
+	u64 period, runtime;
+
+	period = ktime_to_ns(tg->rt_bandwidth.rt_period);
+	runtime = tg->rt_bandwidth.rt_runtime;
+
+	if (tg == d->tg) {
+		period = d->rt_period;
+		runtime = d->rt_runtime;
+	}
+
+	if (rt_bandwidth_enabled() && !runtime && tg_has_rt_tasks(tg))
+		return -EBUSY;
+
+	total = to_ratio(period, runtime);
+
+	list_for_each_entry_rcu(child, &tg->children, siblings) {
+		period = ktime_to_ns(child->rt_bandwidth.rt_period);
+		runtime = child->rt_bandwidth.rt_runtime;
+
+		if (child == d->tg) {
+			period = d->rt_period;
+			runtime = d->rt_runtime;
+		}
+
+		sum += to_ratio(period, runtime);
+	}
+
+	if (sum > total)
+		return -EINVAL;
+
+	return 0;
+}
+
+static int __rt_schedulable(struct task_group *tg, u64 period, u64 runtime)
+{
+	struct rt_schedulable_data data = {
+		.tg = tg,
+		.rt_period = period,
+		.rt_runtime = runtime,
+	};
+
+	return walk_tg_tree(tg_schedulable, tg_nop, &data);
+}
+
 static int tg_set_bandwidth(struct task_group *tg,
 		u64 rt_period, u64 rt_runtime)
 {
@@ -8884,14 +8907,9 @@ static int tg_set_bandwidth(struct task_group *tg,
 
 	mutex_lock(&rt_constraints_mutex);
 	read_lock(&tasklist_lock);
-	if (rt_runtime == 0 && tg_has_rt_tasks(tg)) {
-		err = -EBUSY;
+	err = __rt_schedulable(tg, rt_period, rt_runtime);
+	if (err)
 		goto unlock;
-	}
-	if (!__rt_schedulable(tg, rt_period, rt_runtime)) {
-		err = -EINVAL;
-		goto unlock;
-	}
 
 	spin_lock_irq(&tg->rt_bandwidth.rt_runtime_lock);
 	tg->rt_bandwidth.rt_period = ns_to_ktime(rt_period);
@@ -8964,12 +8982,16 @@ static int sched_rt_global_constraints(void)
 	u64 rt_runtime, rt_period;
 	int ret = 0;
 
+	if (sysctl_sched_rt_period <= 0)
+		return -EINVAL;
+
 	rt_period = ktime_to_ns(tg->rt_bandwidth.rt_period);
 	rt_runtime = tg->rt_bandwidth.rt_runtime;
 
 	mutex_lock(&rt_constraints_mutex);
-	if (!__rt_schedulable(tg, rt_period, rt_runtime))
-		ret = -EINVAL;
+	read_lock(&tasklist_lock);
+	ret = __rt_schedulable(tg, rt_period, rt_runtime);
+	read_unlock(&tasklist_lock);
 	mutex_unlock(&rt_constraints_mutex);
 
 	return ret;
@@ -8980,6 +9002,9 @@ static int sched_rt_global_constraints(void)
 	unsigned long flags;
 	int i;
 
+	if (sysctl_sched_rt_period <= 0)
+		return -EINVAL;
+
 	spin_lock_irqsave(&def_rt_bandwidth.rt_runtime_lock, flags);
 	for_each_possible_cpu(i) {
 		struct rt_rq *rt_rq = &cpu_rq(i)->rt;

kernel/sched_rt.c

@@ -350,6 +350,7 @@ static void __enable_runtime(struct rq *rq)
 		spin_lock(&rt_rq->rt_runtime_lock);
 		rt_rq->rt_runtime = rt_b->rt_runtime;
 		rt_rq->rt_time = 0;
+		rt_rq->rt_throttled = 0;
 		spin_unlock(&rt_rq->rt_runtime_lock);
 		spin_unlock(&rt_b->rt_runtime_lock);
 	}
@@ -388,7 +389,7 @@ static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun)
 	int i, idle = 1;
 	cpumask_t span;
 
-	if (rt_b->rt_runtime == RUNTIME_INF)
+	if (!rt_bandwidth_enabled() || rt_b->rt_runtime == RUNTIME_INF)
 		return 1;
 
 	span = sched_rt_period_mask();
@@ -486,6 +487,9 @@ static void update_curr_rt(struct rq *rq)
 	curr->se.exec_start = rq->clock;
 	cpuacct_charge(curr, delta_exec);
 
+	if (!rt_bandwidth_enabled())
+		return;
+
 	for_each_sched_rt_entity(rt_se) {
 		rt_rq = rt_rq_of_se(rt_se);
 

kernel/time/clockevents.c

@@ -71,6 +71,16 @@ void clockevents_set_mode(struct clock_event_device *dev,
 	}
 }
 
+/**
+ * clockevents_shutdown - shutdown the device and clear next_event
+ * @dev:	device to shutdown
+ */
+void clockevents_shutdown(struct clock_event_device *dev)
+{
+	clockevents_set_mode(dev, CLOCK_EVT_MODE_SHUTDOWN);
+	dev->next_event.tv64 = KTIME_MAX;
+}
+
 /**
  * clockevents_program_event - Reprogram the clock event device.
  * @expires:	absolute expiry time (monotonic clock)
@@ -206,7 +216,7 @@ void clockevents_exchange_device(struct clock_event_device *old,
 
 	if (new) {
 		BUG_ON(new->mode != CLOCK_EVT_MODE_UNUSED);
-		clockevents_set_mode(new, CLOCK_EVT_MODE_SHUTDOWN);
+		clockevents_shutdown(new);
 	}
 	local_irq_restore(flags);
 }

kernel/time/tick-broadcast.c

@@ -236,8 +236,7 @@ static void tick_do_broadcast_on_off(void *why)
 		if (!cpu_isset(cpu, tick_broadcast_mask)) {
 			cpu_set(cpu, tick_broadcast_mask);
 			if (td->mode == TICKDEV_MODE_PERIODIC)
-				clockevents_set_mode(dev,
-						     CLOCK_EVT_MODE_SHUTDOWN);
+				clockevents_shutdown(dev);
 		}
 		if (*reason == CLOCK_EVT_NOTIFY_BROADCAST_FORCE)
 			tick_broadcast_force = 1;
@@ -254,7 +253,7 @@ static void tick_do_broadcast_on_off(void *why)
 
 	if (cpus_empty(tick_broadcast_mask)) {
 		if (!bc_stopped)
-			clockevents_set_mode(bc, CLOCK_EVT_MODE_SHUTDOWN);
+			clockevents_shutdown(bc);
 	} else if (bc_stopped) {
 		if (tick_broadcast_device.mode == TICKDEV_MODE_PERIODIC)
 			tick_broadcast_start_periodic(bc);
@@ -306,7 +305,7 @@ void tick_shutdown_broadcast(unsigned int *cpup)
 
 	if (tick_broadcast_device.mode == TICKDEV_MODE_PERIODIC) {
 		if (bc && cpus_empty(tick_broadcast_mask))
-			clockevents_set_mode(bc, CLOCK_EVT_MODE_SHUTDOWN);
+			clockevents_shutdown(bc);
 	}
 
 	spin_unlock_irqrestore(&tick_broadcast_lock, flags);
@@ -321,7 +320,7 @@ void tick_suspend_broadcast(void)
 
 	bc = tick_broadcast_device.evtdev;
 	if (bc)
-		clockevents_set_mode(bc, CLOCK_EVT_MODE_SHUTDOWN);
+		clockevents_shutdown(bc);
 
 	spin_unlock_irqrestore(&tick_broadcast_lock, flags);
 }

kernel/time/tick-common.c

@@ -249,7 +249,7 @@ static int tick_check_new_device(struct clock_event_device *newdev)
 	 * not give it back to the clockevents layer !
 	 */
 	if (tick_is_broadcast_device(curdev)) {
-		clockevents_set_mode(curdev, CLOCK_EVT_MODE_SHUTDOWN);
+		clockevents_shutdown(curdev);
 		curdev = NULL;
 	}
 	clockevents_exchange_device(curdev, newdev);
@@ -311,7 +311,7 @@ static void tick_suspend(void)
 	unsigned long flags;
 
 	spin_lock_irqsave(&tick_device_lock, flags);
-	clockevents_set_mode(td->evtdev, CLOCK_EVT_MODE_SHUTDOWN);
+	clockevents_shutdown(td->evtdev);
 	spin_unlock_irqrestore(&tick_device_lock, flags);
 }
 

kernel/time/tick-internal.h

@@ -10,6 +10,8 @@ extern int tick_do_timer_cpu __read_mostly;
 extern void tick_setup_periodic(struct clock_event_device *dev, int broadcast);
 extern void tick_handle_periodic(struct clock_event_device *dev);
 
+extern void clockevents_shutdown(struct clock_event_device *dev);
+
 /*
  * NO_HZ / high resolution timer shared code
  */

kernel/user.c

@@ -169,7 +169,7 @@ static ssize_t cpu_rt_runtime_show(struct kobject *kobj,
 {
 	struct user_struct *up = container_of(kobj, struct user_struct, kobj);
 
-	return sprintf(buf, "%lu\n", sched_group_rt_runtime(up->tg));
+	return sprintf(buf, "%ld\n", sched_group_rt_runtime(up->tg));
 }
 
 static ssize_t cpu_rt_runtime_store(struct kobject *kobj,
@@ -180,7 +180,7 @@ static ssize_t cpu_rt_runtime_store(struct kobject *kobj,
 	unsigned long rt_runtime;
 	int rc;
 
-	sscanf(buf, "%lu", &rt_runtime);
+	sscanf(buf, "%ld", &rt_runtime);
 
 	rc = sched_group_set_rt_runtime(up->tg, rt_runtime);