Merge git://git.kernel.org/pub/scm/linux/kernel/git/mingo/linux-2.6-sched

* git://git.kernel.org/pub/scm/linux/kernel/git/mingo/linux-2.6-sched:
  sched: rt-group: refure unrunnable tasks
  sched: rt-group: clean up the ifdeffery
  sched: rt-group: make rt groups scheduling configurable
  sched: rt-group: interface
  sched: rt-group: deal with PI
  sched: fix incorrect irq lock usage in normalize_rt_tasks()
  sched: fair-group: separate tg->shares from task_group_lock
  hrtimer: more hrtimer_init_sleeper() fallout.

Documentation/sched-rt-group.txt

+
+
+Real-Time group scheduling.
+
+The problem space:
+
+In order to schedule multiple groups of realtime tasks each group must
+be assigned a fixed portion of the CPU time available. Without a minimum
+guarantee a realtime group can obviously fall short. A fuzzy upper limit
+is of no use since it cannot be relied upon. Which leaves us with just
+the single fixed portion.
+
+CPU time is divided by means of specifying how much time can be spent
+running in a given period. Say a frame fixed realtime renderer must
+deliver 25 frames a second, which yields a period of 0.04s. Now say
+it will also have to play some music and respond to input, leaving it
+with around 80% for the graphics. We can then give this group a runtime
+of 0.8 * 0.04s = 0.032s.
+
+This way the graphics group will have a 0.04s period with a 0.032s runtime
+limit.
+
+Now if the audio thread needs to refill the DMA buffer every 0.005s, but
+needs only about 3% CPU time to do so, it can do with a 0.03 * 0.005s
+= 0.00015s.
+
+
+The Interface:
+
+system wide:
+
+/proc/sys/kernel/sched_rt_period_ms
+/proc/sys/kernel/sched_rt_runtime_us
+
+CONFIG_FAIR_USER_SCHED
+
+/sys/kernel/uids/<uid>/cpu_rt_runtime_us
+
+or
+
+CONFIG_FAIR_CGROUP_SCHED
+
+/cgroup/<cgroup>/cpu.rt_runtime_us
+
+[ time is specified in us because the interface is s32; this gives an
+  operating range of ~35m to 1us ]
+
+The period takes values in [ 1, INT_MAX ], runtime in [ -1, INT_MAX - 1 ].
+
+A runtime of -1 specifies runtime == period, ie. no limit.
+
+New groups get the period from /proc/sys/kernel/sched_rt_period_us and
+a runtime of 0.
+
+Settings are constrained to:
+
+   \Sum_{i} runtime_{i} / global_period <= global_runtime / global_period
+
+in order to keep the configuration schedulable.

include/linux/cgroup_subsys.h

@@ -25,7 +25,7 @@ SUBSYS(ns)
 
 /* */
 
-#ifdef CONFIG_FAIR_CGROUP_SCHED
+#ifdef CONFIG_CGROUP_SCHED
 SUBSYS(cpu_cgroup)
 #endif
 

include/linux/sched.h

@@ -590,7 +590,7 @@ struct user_struct {
 	struct hlist_node uidhash_node;
 	uid_t uid;
 
-#ifdef CONFIG_FAIR_USER_SCHED
+#ifdef CONFIG_USER_SCHED
 	struct task_group *tg;
 #ifdef CONFIG_SYSFS
 	struct kobject kobj;
@@ -973,7 +973,7 @@ struct sched_rt_entity {
 	unsigned long timeout;
 	int nr_cpus_allowed;
 
-#ifdef CONFIG_FAIR_GROUP_SCHED
+#ifdef CONFIG_RT_GROUP_SCHED
 	struct sched_rt_entity	*parent;
 	/* rq on which this entity is (to be) queued: */
 	struct rt_rq		*rt_rq;
@@ -1541,8 +1541,6 @@ extern unsigned int sysctl_sched_child_runs_first;
 extern unsigned int sysctl_sched_features;
 extern unsigned int sysctl_sched_migration_cost;
 extern unsigned int sysctl_sched_nr_migrate;
-extern unsigned int sysctl_sched_rt_period;
-extern unsigned int sysctl_sched_rt_ratio;
 #if defined(CONFIG_FAIR_GROUP_SCHED) && defined(CONFIG_SMP)
 extern unsigned int sysctl_sched_min_bal_int_shares;
 extern unsigned int sysctl_sched_max_bal_int_shares;
@@ -1552,6 +1550,8 @@ int sched_nr_latency_handler(struct ctl_table *table, int write,
 		struct file *file, void __user *buffer, size_t *length,
 		loff_t *ppos);
 #endif
+extern unsigned int sysctl_sched_rt_period;
+extern int sysctl_sched_rt_runtime;
 
 extern unsigned int sysctl_sched_compat_yield;
 
@@ -2027,16 +2027,22 @@ extern int sched_mc_power_savings, sched_smt_power_savings;
 
 extern void normalize_rt_tasks(void);
 
-#ifdef CONFIG_FAIR_GROUP_SCHED
+#ifdef CONFIG_GROUP_SCHED
 
 extern struct task_group init_task_group;
 
 extern struct task_group *sched_create_group(void);
 extern void sched_destroy_group(struct task_group *tg);
 extern void sched_move_task(struct task_struct *tsk);
+#ifdef CONFIG_FAIR_GROUP_SCHED
 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
 extern unsigned long sched_group_shares(struct task_group *tg);
-
+#endif
+#ifdef CONFIG_RT_GROUP_SCHED
+extern int sched_group_set_rt_runtime(struct task_group *tg,
+				      long rt_runtime_us);
+extern long sched_group_rt_runtime(struct task_group *tg);
+#endif
 #endif
 
 #ifdef CONFIG_TASK_XACCT

init/Kconfig

@@ -311,25 +311,36 @@ config CPUSETS
 
 	  Say N if unsure.
 
-config FAIR_GROUP_SCHED
-	bool "Fair group CPU scheduler"
+config GROUP_SCHED
+	bool "Group CPU scheduler"
 	default y
 	help
 	  This feature lets CPU scheduler recognize task groups and control CPU
 	  bandwidth allocation to such task groups.
 
+config FAIR_GROUP_SCHED
+	bool "Group scheduling for SCHED_OTHER"
+	depends on GROUP_SCHED
+	default y
+
+config RT_GROUP_SCHED
+	bool "Group scheduling for SCHED_RR/FIFO"
+	depends on EXPERIMENTAL
+	depends on GROUP_SCHED
+	default n
+
 choice
-	depends on FAIR_GROUP_SCHED
+	depends on GROUP_SCHED
 	prompt "Basis for grouping tasks"
-	default FAIR_USER_SCHED
+	default USER_SCHED
 
-config FAIR_USER_SCHED
+config USER_SCHED
 	bool "user id"
 	help
 	  This option will choose userid as the basis for grouping
 	  tasks, thus providing equal CPU bandwidth to each user.
 
-config FAIR_CGROUP_SCHED
+config CGROUP_SCHED
 	bool "Control groups"
  	depends on CGROUPS
  	help

kernel/rtmutex.c

@@ -630,9 +630,12 @@ rt_mutex_slowlock(struct rt_mutex *lock, int state,
 	set_current_state(state);
 
 	/* Setup the timer, when timeout != NULL */
-	if (unlikely(timeout))
+	if (unlikely(timeout)) {
 		hrtimer_start(&timeout->timer, timeout->timer.expires,
 			      HRTIMER_MODE_ABS);
+		if (!hrtimer_active(&timeout->timer))
+			timeout->task = NULL;
+	}
 
 	for (;;) {
 		/* Try to acquire the lock: */

kernel/sched_rt.c

@@ -55,14 +55,14 @@ static inline int on_rt_rq(struct sched_rt_entity *rt_se)
 	return !list_empty(&rt_se->run_list);
 }
 
-#ifdef CONFIG_FAIR_GROUP_SCHED
+#ifdef CONFIG_RT_GROUP_SCHED
 
-static inline unsigned int sched_rt_ratio(struct rt_rq *rt_rq)
+static inline u64 sched_rt_runtime(struct rt_rq *rt_rq)
 {
 	if (!rt_rq->tg)
-		return SCHED_RT_FRAC;
+		return RUNTIME_INF;
 
-	return rt_rq->tg->rt_ratio;
+	return rt_rq->tg->rt_runtime;
 }
 
 #define for_each_leaf_rt_rq(rt_rq, rq) \
@@ -89,7 +89,7 @@ static inline struct rt_rq *group_rt_rq(struct sched_rt_entity *rt_se)
 static void enqueue_rt_entity(struct sched_rt_entity *rt_se);
 static void dequeue_rt_entity(struct sched_rt_entity *rt_se);
 
-static void sched_rt_ratio_enqueue(struct rt_rq *rt_rq)
+static void sched_rt_rq_enqueue(struct rt_rq *rt_rq)
 {
 	struct sched_rt_entity *rt_se = rt_rq->rt_se;
 
@@ -102,7 +102,7 @@ static void sched_rt_ratio_enqueue(struct rt_rq *rt_rq)
 	}
 }
 
-static void sched_rt_ratio_dequeue(struct rt_rq *rt_rq)
+static void sched_rt_rq_dequeue(struct rt_rq *rt_rq)
 {
 	struct sched_rt_entity *rt_se = rt_rq->rt_se;
 
@@ -110,11 +110,31 @@ static void sched_rt_ratio_dequeue(struct rt_rq *rt_rq)
 		dequeue_rt_entity(rt_se);
 }
 
+static inline int rt_rq_throttled(struct rt_rq *rt_rq)
+{
+	return rt_rq->rt_throttled && !rt_rq->rt_nr_boosted;
+}
+
+static int rt_se_boosted(struct sched_rt_entity *rt_se)
+{
+	struct rt_rq *rt_rq = group_rt_rq(rt_se);
+	struct task_struct *p;
+
+	if (rt_rq)
+		return !!rt_rq->rt_nr_boosted;
+
+	p = rt_task_of(rt_se);
+	return p->prio != p->normal_prio;
+}
+
 #else
 
-static inline unsigned int sched_rt_ratio(struct rt_rq *rt_rq)
+static inline u64 sched_rt_runtime(struct rt_rq *rt_rq)
 {
-	return sysctl_sched_rt_ratio;
+	if (sysctl_sched_rt_runtime == -1)
+		return RUNTIME_INF;
+
+	return (u64)sysctl_sched_rt_runtime * NSEC_PER_USEC;
 }
 
 #define for_each_leaf_rt_rq(rt_rq, rq) \
@@ -141,19 +161,23 @@ static inline struct rt_rq *group_rt_rq(struct sched_rt_entity *rt_se)
 	return NULL;
 }
 
-static inline void sched_rt_ratio_enqueue(struct rt_rq *rt_rq)
+static inline void sched_rt_rq_enqueue(struct rt_rq *rt_rq)
 {
 }
 
-static inline void sched_rt_ratio_dequeue(struct rt_rq *rt_rq)
+static inline void sched_rt_rq_dequeue(struct rt_rq *rt_rq)
 {
 }
 
+static inline int rt_rq_throttled(struct rt_rq *rt_rq)
+{
+	return rt_rq->rt_throttled;
+}
 #endif
 
 static inline int rt_se_prio(struct sched_rt_entity *rt_se)
 {
-#ifdef CONFIG_FAIR_GROUP_SCHED
+#ifdef CONFIG_RT_GROUP_SCHED
 	struct rt_rq *rt_rq = group_rt_rq(rt_se);
 
 	if (rt_rq)
@@ -163,28 +187,26 @@ static inline int rt_se_prio(struct sched_rt_entity *rt_se)
 	return rt_task_of(rt_se)->prio;
 }
 
-static int sched_rt_ratio_exceeded(struct rt_rq *rt_rq)
+static int sched_rt_runtime_exceeded(struct rt_rq *rt_rq)
 {
-	unsigned int rt_ratio = sched_rt_ratio(rt_rq);
-	u64 period, ratio;
+	u64 runtime = sched_rt_runtime(rt_rq);
 
-	if (rt_ratio == SCHED_RT_FRAC)
+	if (runtime == RUNTIME_INF)
 		return 0;
 
 	if (rt_rq->rt_throttled)
-		return 1;
-
-	period = (u64)sysctl_sched_rt_period * NSEC_PER_MSEC;
-	ratio = (period * rt_ratio) >> SCHED_RT_FRAC_SHIFT;
+		return rt_rq_throttled(rt_rq);
 
-	if (rt_rq->rt_time > ratio) {
+	if (rt_rq->rt_time > runtime) {
 		struct rq *rq = rq_of_rt_rq(rt_rq);
 
 		rq->rt_throttled = 1;
 		rt_rq->rt_throttled = 1;
 
-		sched_rt_ratio_dequeue(rt_rq);
-		return 1;
+		if (rt_rq_throttled(rt_rq)) {
+			sched_rt_rq_dequeue(rt_rq);
+			return 1;
+		}
 	}
 
 	return 0;
@@ -196,17 +218,16 @@ static void update_sched_rt_period(struct rq *rq)
 	u64 period;
 
 	while (rq->clock > rq->rt_period_expire) {
-		period = (u64)sysctl_sched_rt_period * NSEC_PER_MSEC;
+		period = (u64)sysctl_sched_rt_period * NSEC_PER_USEC;
 		rq->rt_period_expire += period;
 
 		for_each_leaf_rt_rq(rt_rq, rq) {
-			unsigned long rt_ratio = sched_rt_ratio(rt_rq);
-			u64 ratio = (period * rt_ratio) >> SCHED_RT_FRAC_SHIFT;
+			u64 runtime = sched_rt_runtime(rt_rq);
 
-			rt_rq->rt_time -= min(rt_rq->rt_time, ratio);
-			if (rt_rq->rt_throttled) {
+			rt_rq->rt_time -= min(rt_rq->rt_time, runtime);
+			if (rt_rq->rt_throttled && rt_rq->rt_time < runtime) {
 				rt_rq->rt_throttled = 0;
-				sched_rt_ratio_enqueue(rt_rq);
+				sched_rt_rq_enqueue(rt_rq);
 			}
 		}
 
@@ -239,12 +260,7 @@ static void update_curr_rt(struct rq *rq)
 	cpuacct_charge(curr, delta_exec);
 
 	rt_rq->rt_time += delta_exec;
-	/*
-	 * might make it a tad more accurate:
-	 *
-	 * update_sched_rt_period(rq);
-	 */
-	if (sched_rt_ratio_exceeded(rt_rq))
+	if (sched_rt_runtime_exceeded(rt_rq))
 		resched_task(curr);
 }
 
@@ -253,7 +269,7 @@ void inc_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
 {
 	WARN_ON(!rt_prio(rt_se_prio(rt_se)));
 	rt_rq->rt_nr_running++;
-#if defined CONFIG_SMP || defined CONFIG_FAIR_GROUP_SCHED
+#if defined CONFIG_SMP || defined CONFIG_RT_GROUP_SCHED
 	if (rt_se_prio(rt_se) < rt_rq->highest_prio)
 		rt_rq->highest_prio = rt_se_prio(rt_se);
 #endif
@@ -265,6 +281,10 @@ void inc_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
 
 	update_rt_migration(rq_of_rt_rq(rt_rq));
 #endif
+#ifdef CONFIG_RT_GROUP_SCHED
+	if (rt_se_boosted(rt_se))
+		rt_rq->rt_nr_boosted++;
+#endif
 }
 
 static inline
@@ -273,7 +293,7 @@ void dec_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
 	WARN_ON(!rt_prio(rt_se_prio(rt_se)));
 	WARN_ON(!rt_rq->rt_nr_running);
 	rt_rq->rt_nr_running--;
-#if defined CONFIG_SMP || defined CONFIG_FAIR_GROUP_SCHED
+#if defined CONFIG_SMP || defined CONFIG_RT_GROUP_SCHED
 	if (rt_rq->rt_nr_running) {
 		struct rt_prio_array *array;
 
@@ -295,6 +315,12 @@ void dec_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
 
 	update_rt_migration(rq_of_rt_rq(rt_rq));
 #endif /* CONFIG_SMP */
+#ifdef CONFIG_RT_GROUP_SCHED
+	if (rt_se_boosted(rt_se))
+		rt_rq->rt_nr_boosted--;
+
+	WARN_ON(!rt_rq->rt_nr_running && rt_rq->rt_nr_boosted);
+#endif
 }
 
 static void enqueue_rt_entity(struct sched_rt_entity *rt_se)
@@ -303,7 +329,7 @@ static void enqueue_rt_entity(struct sched_rt_entity *rt_se)
 	struct rt_prio_array *array = &rt_rq->active;
 	struct rt_rq *group_rq = group_rt_rq(rt_se);
 
-	if (group_rq && group_rq->rt_throttled)
+	if (group_rq && rt_rq_throttled(group_rq))
 		return;
 
 	list_add_tail(&rt_se->run_list, array->queue + rt_se_prio(rt_se));
@@ -496,7 +522,7 @@ static struct task_struct *pick_next_task_rt(struct rq *rq)
 	if (unlikely(!rt_rq->rt_nr_running))
 		return NULL;
 
-	if (sched_rt_ratio_exceeded(rt_rq))
+	if (rt_rq_throttled(rt_rq))
 		return NULL;
 
 	do {

kernel/sysctl.c

@@ -311,22 +311,6 @@ static struct ctl_table kern_table[] = {
 		.mode		= 0644,
 		.proc_handler	= &proc_dointvec,
 	},
-	{
-		.ctl_name	= CTL_UNNUMBERED,
-		.procname	= "sched_rt_period_ms",
-		.data		= &sysctl_sched_rt_period,
-		.maxlen		= sizeof(unsigned int),
-		.mode		= 0644,
-		.proc_handler	= &proc_dointvec,
-	},
-	{
-		.ctl_name	= CTL_UNNUMBERED,
-		.procname	= "sched_rt_ratio",
-		.data		= &sysctl_sched_rt_ratio,
-		.maxlen		= sizeof(unsigned int),
-		.mode		= 0644,
-		.proc_handler	= &proc_dointvec,
-	},
 #if defined(CONFIG_FAIR_GROUP_SCHED) && defined(CONFIG_SMP)
 	{
 		.ctl_name       = CTL_UNNUMBERED,
@@ -346,6 +330,22 @@ static struct ctl_table kern_table[] = {
 	},
 #endif
 #endif
+	{
+		.ctl_name	= CTL_UNNUMBERED,
+		.procname	= "sched_rt_period_us",
+		.data		= &sysctl_sched_rt_period,
+		.maxlen		= sizeof(unsigned int),
+		.mode		= 0644,
+		.proc_handler	= &proc_dointvec,
+	},
+	{
+		.ctl_name	= CTL_UNNUMBERED,
+		.procname	= "sched_rt_runtime_us",
+		.data		= &sysctl_sched_rt_runtime,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= &proc_dointvec,
+	},
 	{
 		.ctl_name	= CTL_UNNUMBERED,
 		.procname	= "sched_compat_yield",

kernel/user.c

@@ -57,7 +57,7 @@ struct user_struct root_user = {
 	.uid_keyring	= &root_user_keyring,
 	.session_keyring = &root_session_keyring,
 #endif
-#ifdef CONFIG_FAIR_USER_SCHED
+#ifdef CONFIG_USER_SCHED
 	.tg		= &init_task_group,
 #endif
 };
@@ -90,7 +90,7 @@ static struct user_struct *uid_hash_find(uid_t uid, struct hlist_head *hashent)
 	return NULL;
 }
 
-#ifdef CONFIG_FAIR_USER_SCHED
+#ifdef CONFIG_USER_SCHED
 
 static void sched_destroy_user(struct user_struct *up)
 {
@@ -113,15 +113,15 @@ static void sched_switch_user(struct task_struct *p)
 	sched_move_task(p);
 }
 
-#else	/* CONFIG_FAIR_USER_SCHED */
+#else	/* CONFIG_USER_SCHED */
 
 static void sched_destroy_user(struct user_struct *up) { }
 static int sched_create_user(struct user_struct *up) { return 0; }
 static void sched_switch_user(struct task_struct *p) { }
 
-#endif	/* CONFIG_FAIR_USER_SCHED */
+#endif	/* CONFIG_USER_SCHED */
 
-#if defined(CONFIG_FAIR_USER_SCHED) && defined(CONFIG_SYSFS)
+#if defined(CONFIG_USER_SCHED) && defined(CONFIG_SYSFS)
 
 static struct kset *uids_kset; /* represents the /sys/kernel/uids/ directory */
 static DEFINE_MUTEX(uids_mutex);
@@ -137,6 +137,7 @@ static inline void uids_mutex_unlock(void)
 }
 
 /* uid directory attributes */
+#ifdef CONFIG_FAIR_GROUP_SCHED
 static ssize_t cpu_shares_show(struct kobject *kobj,
 			       struct kobj_attribute *attr,
 			       char *buf)
@@ -163,10 +164,45 @@ static ssize_t cpu_shares_store(struct kobject *kobj,
 
 static struct kobj_attribute cpu_share_attr =
 	__ATTR(cpu_share, 0644, cpu_shares_show, cpu_shares_store);
+#endif
+
+#ifdef CONFIG_RT_GROUP_SCHED
+static ssize_t cpu_rt_runtime_show(struct kobject *kobj,
+				   struct kobj_attribute *attr,
+				   char *buf)
+{
+	struct user_struct *up = container_of(kobj, struct user_struct, kobj);
+
+	return sprintf(buf, "%lu\n", sched_group_rt_runtime(up->tg));
+}
+
+static ssize_t cpu_rt_runtime_store(struct kobject *kobj,
+				    struct kobj_attribute *attr,
+				    const char *buf, size_t size)
+{
+	struct user_struct *up = container_of(kobj, struct user_struct, kobj);
+	unsigned long rt_runtime;
+	int rc;
+
+	sscanf(buf, "%lu", &rt_runtime);
+
+	rc = sched_group_set_rt_runtime(up->tg, rt_runtime);
+
+	return (rc ? rc : size);
+}
+
+static struct kobj_attribute cpu_rt_runtime_attr =
+	__ATTR(cpu_rt_runtime, 0644, cpu_rt_runtime_show, cpu_rt_runtime_store);
+#endif
 
 /* default attributes per uid directory */
 static struct attribute *uids_attributes[] = {
+#ifdef CONFIG_FAIR_GROUP_SCHED
 	&cpu_share_attr.attr,
+#endif
+#ifdef CONFIG_RT_GROUP_SCHED
+	&cpu_rt_runtime_attr.attr,
+#endif
 	NULL
 };
 
@@ -269,7 +305,7 @@ static inline void free_user(struct user_struct *up, unsigned long flags)
 	schedule_work(&up->work);
 }
 
-#else	/* CONFIG_FAIR_USER_SCHED && CONFIG_SYSFS */
+#else	/* CONFIG_USER_SCHED && CONFIG_SYSFS */
 
 int uids_sysfs_init(void) { return 0; }
 static inline int uids_user_create(struct user_struct *up) { return 0; }
@@ -373,7 +409,7 @@ struct user_struct * alloc_uid(struct user_namespace *ns, uid_t uid)
 		spin_lock_irq(&uidhash_lock);
 		up = uid_hash_find(uid, hashent);
 		if (up) {
-			/* This case is not possible when CONFIG_FAIR_USER_SCHED
+			/* This case is not possible when CONFIG_USER_SCHED
 			 * is defined, since we serialize alloc_uid() using
 			 * uids_mutex. Hence no need to call
 			 * sched_destroy_user() or remove_user_sysfs_dir().