[deliverable/linux.git] / kernel / sched_autogroup.c

#ifdef CONFIG_SCHED_AUTOGROUP

#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/kallsyms.h>
#include <linux/utsname.h>

unsigned int __read_mostly sysctl_sched_autogroup_enabled = 1;
static struct autogroup autogroup_default;
static atomic_t autogroup_seq_nr;

static void __init autogroup_init(struct task_struct *init_task)
{
	autogroup_default.tg = &root_task_group;
	kref_init(&autogroup_default.kref);
	init_rwsem(&autogroup_default.lock);
	init_task->signal->autogroup = &autogroup_default;
}

static inline void autogroup_free(struct task_group *tg)
{
	kfree(tg->autogroup);
}

static inline void autogroup_destroy(struct kref *kref)
{
	struct autogroup *ag = container_of(kref, struct autogroup, kref);

#ifdef CONFIG_RT_GROUP_SCHED
	/* We've redirected RT tasks to the root task group... */
	ag->tg->rt_se = NULL;
	ag->tg->rt_rq = NULL;
#endif
	sched_destroy_group(ag->tg);
}

static inline void autogroup_kref_put(struct autogroup *ag)
{
	kref_put(&ag->kref, autogroup_destroy);
}

static inline struct autogroup *autogroup_kref_get(struct autogroup *ag)
{
	kref_get(&ag->kref);
	return ag;
}

static inline struct autogroup *autogroup_task_get(struct task_struct *p)
{
	struct autogroup *ag;
	unsigned long flags;

	if (!lock_task_sighand(p, &flags))
		return autogroup_kref_get(&autogroup_default);

	ag = autogroup_kref_get(p->signal->autogroup);
	unlock_task_sighand(p, &flags);

	return ag;
}

#ifdef CONFIG_RT_GROUP_SCHED
static void free_rt_sched_group(struct task_group *tg);
#endif

static inline struct autogroup *autogroup_create(void)
{
	struct autogroup *ag = kzalloc(sizeof(*ag), GFP_KERNEL);
	struct task_group *tg;

	if (!ag)
		goto out_fail;

	tg = sched_create_group(&root_task_group);

	if (IS_ERR(tg))
		goto out_free;

	kref_init(&ag->kref);
	init_rwsem(&ag->lock);
	ag->id = atomic_inc_return(&autogroup_seq_nr);
	ag->tg = tg;
#ifdef CONFIG_RT_GROUP_SCHED
	/*
	 * Autogroup RT tasks are redirected to the root task group
	 * so we don't have to move tasks around upon policy change,
	 * or flail around trying to allocate bandwidth on the fly.
	 * A bandwidth exception in __sched_setscheduler() allows
	 * the policy change to proceed.  Thereafter, task_group()
	 * returns &root_task_group, so zero bandwidth is required.
	 */
	free_rt_sched_group(tg);
	tg->rt_se = root_task_group.rt_se;
	tg->rt_rq = root_task_group.rt_rq;
#endif
	tg->autogroup = ag;

	return ag;

out_free:
	kfree(ag);
out_fail:
	if (printk_ratelimit()) {
		printk(KERN_WARNING "autogroup_create: %s failure.\n",
			ag ? "sched_create_group()" : "kmalloc()");
	}

	return autogroup_kref_get(&autogroup_default);
}

static inline bool
task_wants_autogroup(struct task_struct *p, struct task_group *tg)
{
	if (tg != &root_task_group)
		return false;

	if (p->sched_class != &fair_sched_class)
		return false;

	/*
	 * We can only assume the task group can't go away on us if
	 * autogroup_move_group() can see us on ->thread_group list.
	 */
	if (p->flags & PF_EXITING)
		return false;

	return true;
}

static inline bool task_group_is_autogroup(struct task_group *tg)
{
	return !!tg->autogroup;
}

static inline struct task_group *
autogroup_task_group(struct task_struct *p, struct task_group *tg)
{
	int enabled = ACCESS_ONCE(sysctl_sched_autogroup_enabled);

	if (enabled && task_wants_autogroup(p, tg))
		return p->signal->autogroup->tg;

	return tg;
}

static void
autogroup_move_group(struct task_struct *p, struct autogroup *ag)
{
	struct autogroup *prev;
	struct task_struct *t;
	unsigned long flags;

	BUG_ON(!lock_task_sighand(p, &flags));

	prev = p->signal->autogroup;
	if (prev == ag) {
		unlock_task_sighand(p, &flags);
		return;
	}

	p->signal->autogroup = autogroup_kref_get(ag);

	if (!ACCESS_ONCE(sysctl_sched_autogroup_enabled))
		goto out;

	t = p;
	do {
		sched_move_task(t);
	} while_each_thread(p, t);

out:
	unlock_task_sighand(p, &flags);
	autogroup_kref_put(prev);
}

/* Allocates GFP_KERNEL, cannot be called under any spinlock */
void sched_autogroup_create_attach(struct task_struct *p)
{
	struct autogroup *ag = autogroup_create();

	autogroup_move_group(p, ag);
	/* drop extra reference added by autogroup_create() */
	autogroup_kref_put(ag);
}
EXPORT_SYMBOL(sched_autogroup_create_attach);

/* Cannot be called under siglock.  Currently has no users */
void sched_autogroup_detach(struct task_struct *p)
{
	autogroup_move_group(p, &autogroup_default);
}
EXPORT_SYMBOL(sched_autogroup_detach);

void sched_autogroup_fork(struct signal_struct *sig)
{
	sig->autogroup = autogroup_task_get(current);
}

void sched_autogroup_exit(struct signal_struct *sig)
{
	autogroup_kref_put(sig->autogroup);
}

static int __init setup_autogroup(char *str)
{
	sysctl_sched_autogroup_enabled = 0;

	return 1;
}

__setup("noautogroup", setup_autogroup);

#ifdef CONFIG_PROC_FS

int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice)
{
	static unsigned long next = INITIAL_JIFFIES;
	struct autogroup *ag;
	int err;

	if (*nice < -20 || *nice > 19)
		return -EINVAL;

	err = security_task_setnice(current, *nice);
	if (err)
		return err;

	if (*nice < 0 && !can_nice(current, *nice))
		return -EPERM;

	/* this is a heavy operation taking global locks.. */
	if (!capable(CAP_SYS_ADMIN) && time_before(jiffies, next))
		return -EAGAIN;

	next = HZ / 10 + jiffies;
	ag = autogroup_task_get(p);

	down_write(&ag->lock);
	err = sched_group_set_shares(ag->tg, prio_to_weight[*nice + 20]);
	if (!err)
		ag->nice = *nice;
	up_write(&ag->lock);

	autogroup_kref_put(ag);

	return err;
}

void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m)
{
	struct autogroup *ag = autogroup_task_get(p);

	if (!task_group_is_autogroup(ag->tg))
		goto out;

	down_read(&ag->lock);
	seq_printf(m, "/autogroup-%ld nice %d\n", ag->id, ag->nice);
	up_read(&ag->lock);

out:
	autogroup_kref_put(ag);
}
#endif /* CONFIG_PROC_FS */

#ifdef CONFIG_SCHED_DEBUG
static inline int autogroup_path(struct task_group *tg, char *buf, int buflen)
{
	if (!task_group_is_autogroup(tg))
		return 0;

	return snprintf(buf, buflen, "%s-%ld", "/autogroup", tg->autogroup->id);
}
#endif /* CONFIG_SCHED_DEBUG */

#endif /* CONFIG_SCHED_AUTOGROUP */
Commit	Line	Data
5091faa4 MG	1	#ifdef CONFIG_SCHED_AUTOGROUP
	2
	3	#include <linux/proc_fs.h>
	4	#include <linux/seq_file.h>
	5	#include <linux/kallsyms.h>
	6	#include <linux/utsname.h>
	7
	8	unsigned int __read_mostly sysctl_sched_autogroup_enabled = 1;
	9	static struct autogroup autogroup_default;
	10	static atomic_t autogroup_seq_nr;
	11
0ca08735	12	static void __init autogroup_init(struct task_struct *init_task)
5091faa4	13	{
07e06b01	14	autogroup_default.tg = &root_task_group;
5091faa4 MG	15	kref_init(&autogroup_default.kref);
	16	init_rwsem(&autogroup_default.lock);
	17	init_task->signal->autogroup = &autogroup_default;
	18	}
	19
	20	static inline void autogroup_free(struct task_group *tg)
	21	{
	22	kfree(tg->autogroup);
	23	}
	24
	25	static inline void autogroup_destroy(struct kref *kref)
	26	{
	27	struct autogroup *ag = container_of(kref, struct autogroup, kref);
	28
f4493771 MG	29	#ifdef CONFIG_RT_GROUP_SCHED
	30	/* We've redirected RT tasks to the root task group... */
	31	ag->tg->rt_se = NULL;
	32	ag->tg->rt_rq = NULL;
	33	#endif
5091faa4 MG	34	sched_destroy_group(ag->tg);
	35	}
	36
	37	static inline void autogroup_kref_put(struct autogroup *ag)
	38	{
	39	kref_put(&ag->kref, autogroup_destroy);
	40	}
	41
	42	static inline struct autogroup autogroup_kref_get(struct autogroup ag)
	43	{
	44	kref_get(&ag->kref);
	45	return ag;
	46	}
	47
4f821987 MG	48	static inline struct autogroup autogroup_task_get(struct task_struct p)
	49	{
	50	struct autogroup *ag;
	51	unsigned long flags;
	52
	53	if (!lock_task_sighand(p, &flags))
	54	return autogroup_kref_get(&autogroup_default);
	55
	56	ag = autogroup_kref_get(p->signal->autogroup);
	57	unlock_task_sighand(p, &flags);
	58
	59	return ag;
	60	}
	61
f4493771 MG	62	#ifdef CONFIG_RT_GROUP_SCHED
	63	static void free_rt_sched_group(struct task_group *tg);
	64	#endif
	65
5091faa4 MG	66	static inline struct autogroup *autogroup_create(void)
	67	{
	68	struct autogroup ag = kzalloc(sizeof(ag), GFP_KERNEL);
	69	struct task_group *tg;
	70
	71	if (!ag)
	72	goto out_fail;
	73
07e06b01	74	tg = sched_create_group(&root_task_group);
5091faa4 MG	75
	76	if (IS_ERR(tg))
	77	goto out_free;
	78
	79	kref_init(&ag->kref);
	80	init_rwsem(&ag->lock);
	81	ag->id = atomic_inc_return(&autogroup_seq_nr);
	82	ag->tg = tg;
f4493771 MG	83	#ifdef CONFIG_RT_GROUP_SCHED
	84	/*
	85	* Autogroup RT tasks are redirected to the root task group
	86	* so we don't have to move tasks around upon policy change,
	87	* or flail around trying to allocate bandwidth on the fly.
	88	* A bandwidth exception in __sched_setscheduler() allows
	89	* the policy change to proceed. Thereafter, task_group()
	90	* returns &root_task_group, so zero bandwidth is required.
	91	*/
	92	free_rt_sched_group(tg);
	93	tg->rt_se = root_task_group.rt_se;
	94	tg->rt_rq = root_task_group.rt_rq;
	95	#endif
5091faa4 MG	96	tg->autogroup = ag;
	97
	98	return ag;
	99
	100	out_free:
	101	kfree(ag);
	102	out_fail:
	103	if (printk_ratelimit()) {
	104	printk(KERN_WARNING "autogroup_create: %s failure.\n",
	105	ag ? "sched_create_group()" : "kmalloc()");
	106	}
	107
	108	return autogroup_kref_get(&autogroup_default);
	109	}
	110
	111	static inline bool
	112	task_wants_autogroup(struct task_struct p, struct task_group tg)
	113	{
	114	if (tg != &root_task_group)
	115	return false;
	116
	117	if (p->sched_class != &fair_sched_class)
	118	return false;
	119
	120	/*
	121	* We can only assume the task group can't go away on us if
	122	* autogroup_move_group() can see us on ->thread_group list.
	123	*/
	124	if (p->flags & PF_EXITING)
	125	return false;
	126
	127	return true;
	128	}
	129
f4493771 MG	130	static inline bool task_group_is_autogroup(struct task_group *tg)
f4493771 MG	131	{
511f67a5	132	return !!tg->autogroup;
f4493771 MG	133	}
f4493771 MG	134
5091faa4 MG	135	static inline struct task_group *
	136	autogroup_task_group(struct task_struct p, struct task_group tg)
	137	{
	138	int enabled = ACCESS_ONCE(sysctl_sched_autogroup_enabled);
	139
	140	if (enabled && task_wants_autogroup(p, tg))
	141	return p->signal->autogroup->tg;
	142
	143	return tg;
	144	}
	145
	146	static void
	147	autogroup_move_group(struct task_struct p, struct autogroup ag)
	148	{
	149	struct autogroup *prev;
	150	struct task_struct *t;
	151	unsigned long flags;
	152
	153	BUG_ON(!lock_task_sighand(p, &flags));
	154
	155	prev = p->signal->autogroup;
	156	if (prev == ag) {
	157	unlock_task_sighand(p, &flags);
	158	return;
	159	}
	160
	161	p->signal->autogroup = autogroup_kref_get(ag);
	162
800d4d30 YZ	163	if (!ACCESS_ONCE(sysctl_sched_autogroup_enabled))
	164	goto out;
	165
5091faa4 MG	166	t = p;
	167	do {
	168	sched_move_task(t);
	169	} while_each_thread(p, t);
	170
800d4d30	171	out:
5091faa4 MG	172	unlock_task_sighand(p, &flags);
	173	autogroup_kref_put(prev);
	174	}
	175
	176	/* Allocates GFP_KERNEL, cannot be called under any spinlock */
	177	void sched_autogroup_create_attach(struct task_struct *p)
	178	{
	179	struct autogroup *ag = autogroup_create();
	180
	181	autogroup_move_group(p, ag);
25985edc	182	/* drop extra reference added by autogroup_create() */
5091faa4 MG	183	autogroup_kref_put(ag);
	184	}
	185	EXPORT_SYMBOL(sched_autogroup_create_attach);
	186
	187	/* Cannot be called under siglock. Currently has no users */
	188	void sched_autogroup_detach(struct task_struct *p)
	189	{
	190	autogroup_move_group(p, &autogroup_default);
	191	}
	192	EXPORT_SYMBOL(sched_autogroup_detach);
	193
	194	void sched_autogroup_fork(struct signal_struct *sig)
	195	{
4f821987	196	sig->autogroup = autogroup_task_get(current);
5091faa4 MG	197	}
	198
	199	void sched_autogroup_exit(struct signal_struct *sig)
	200	{
	201	autogroup_kref_put(sig->autogroup);
	202	}
	203
	204	static int __init setup_autogroup(char *str)
	205	{
	206	sysctl_sched_autogroup_enabled = 0;
	207
	208	return 1;
	209	}
	210
	211	__setup("noautogroup", setup_autogroup);
	212
	213	#ifdef CONFIG_PROC_FS
	214
5091faa4 MG	215	int proc_sched_autogroup_set_nice(struct task_struct p, int nice)
	216	{
	217	static unsigned long next = INITIAL_JIFFIES;
	218	struct autogroup *ag;
	219	int err;
	220
	221	if (nice < -20 \|\| nice > 19)
	222	return -EINVAL;
	223
	224	err = security_task_setnice(current, *nice);
	225	if (err)
	226	return err;
	227
	228	if (nice < 0 && !can_nice(current, nice))
	229	return -EPERM;
	230
	231	/* this is a heavy operation taking global locks.. */
	232	if (!capable(CAP_SYS_ADMIN) && time_before(jiffies, next))
	233	return -EAGAIN;
	234
	235	next = HZ / 10 + jiffies;
4f821987	236	ag = autogroup_task_get(p);
5091faa4 MG	237
	238	down_write(&ag->lock);
	239	err = sched_group_set_shares(ag->tg, prio_to_weight[*nice + 20]);
	240	if (!err)
	241	ag->nice = *nice;
	242	up_write(&ag->lock);
	243
	244	autogroup_kref_put(ag);
	245
	246	return err;
	247	}
	248
	249	void proc_sched_autogroup_show_task(struct task_struct p, struct seq_file m)
	250	{
4f821987	251	struct autogroup *ag = autogroup_task_get(p);
5091faa4	252
511f67a5 MG	253	if (!task_group_is_autogroup(ag->tg))
	254	goto out;
	255
5091faa4 MG	256	down_read(&ag->lock);
	257	seq_printf(m, "/autogroup-%ld nice %d\n", ag->id, ag->nice);
	258	up_read(&ag->lock);
	259
511f67a5	260	out:
5091faa4 MG	261	autogroup_kref_put(ag);
	262	}
	263	#endif /* CONFIG_PROC_FS */
	264
	265	#ifdef CONFIG_SCHED_DEBUG
	266	static inline int autogroup_path(struct task_group tg, char buf, int buflen)
	267	{
511f67a5	268	if (!task_group_is_autogroup(tg))
8ecedd7a BR	269	return 0;
8ecedd7a BR	270
5091faa4 MG	271	return snprintf(buf, buflen, "%s-%ld", "/autogroup", tg->autogroup->id);
	272	}
	273	#endif /* CONFIG_SCHED_DEBUG */
	274
	275	#endif /* CONFIG_SCHED_AUTOGROUP */