[net-next-2.6.git] / block / blk-cgroup.c

/*
 * Common Block IO controller cgroup interface
 *
 * Based on ideas and code from CFQ, CFS and BFQ:
 * Copyright (C) 2003 Jens Axboe <axboe@kernel.dk>
 *
 * Copyright (C) 2008 Fabio Checconi <fabio@gandalf.sssup.it>
 *		      Paolo Valente <paolo.valente@unimore.it>
 *
 * Copyright (C) 2009 Vivek Goyal <vgoyal@redhat.com>
 * 	              Nauman Rafique <nauman@google.com>
 */
#include <linux/ioprio.h>
#include <linux/seq_file.h>
#include <linux/kdev_t.h>
#include <linux/module.h>
#include <linux/err.h>
#include "blk-cgroup.h"

static DEFINE_SPINLOCK(blkio_list_lock);
static LIST_HEAD(blkio_list);

struct blkio_cgroup blkio_root_cgroup = { .weight = 2*BLKIO_WEIGHT_DEFAULT };
EXPORT_SYMBOL_GPL(blkio_root_cgroup);

static struct cgroup_subsys_state *blkiocg_create(struct cgroup_subsys *,
						  struct cgroup *);
static int blkiocg_can_attach(struct cgroup_subsys *, struct cgroup *,
			      struct task_struct *, bool);
static void blkiocg_attach(struct cgroup_subsys *, struct cgroup *,
			   struct cgroup *, struct task_struct *, bool);
static void blkiocg_destroy(struct cgroup_subsys *, struct cgroup *);
static int blkiocg_populate(struct cgroup_subsys *, struct cgroup *);

struct cgroup_subsys blkio_subsys = {
	.name = "blkio",
	.create = blkiocg_create,
	.can_attach = blkiocg_can_attach,
	.attach = blkiocg_attach,
	.destroy = blkiocg_destroy,
	.populate = blkiocg_populate,
#ifdef CONFIG_BLK_CGROUP
	/* note: blkio_subsys_id is otherwise defined in blk-cgroup.h */
	.subsys_id = blkio_subsys_id,
#endif
	.use_id = 1,
	.module = THIS_MODULE,
};
EXPORT_SYMBOL_GPL(blkio_subsys);

struct blkio_cgroup *cgroup_to_blkio_cgroup(struct cgroup *cgroup)
{
	return container_of(cgroup_subsys_state(cgroup, blkio_subsys_id),
			    struct blkio_cgroup, css);
}
EXPORT_SYMBOL_GPL(cgroup_to_blkio_cgroup);

void blkiocg_update_timeslice_used(struct blkio_group *blkg, unsigned long time)
{
	unsigned long flags;

	spin_lock_irqsave(&blkg->stats_lock, flags);
	blkg->stats.time += time;
	spin_unlock_irqrestore(&blkg->stats_lock, flags);
}
EXPORT_SYMBOL_GPL(blkiocg_update_timeslice_used);

void blkiocg_add_blkio_group(struct blkio_cgroup *blkcg,
			struct blkio_group *blkg, void *key, dev_t dev)
{
	unsigned long flags;

	spin_lock_irqsave(&blkcg->lock, flags);
	rcu_assign_pointer(blkg->key, key);
	blkg->blkcg_id = css_id(&blkcg->css);
	hlist_add_head_rcu(&blkg->blkcg_node, &blkcg->blkg_list);
	spin_unlock_irqrestore(&blkcg->lock, flags);
#ifdef CONFIG_DEBUG_BLK_CGROUP
	/* Need to take css reference ? */
	cgroup_path(blkcg->css.cgroup, blkg->path, sizeof(blkg->path));
#endif
	blkg->dev = dev;
}
EXPORT_SYMBOL_GPL(blkiocg_add_blkio_group);

static void __blkiocg_del_blkio_group(struct blkio_group *blkg)
{
	hlist_del_init_rcu(&blkg->blkcg_node);
	blkg->blkcg_id = 0;
}

/*
 * returns 0 if blkio_group was still on cgroup list. Otherwise returns 1
 * indicating that blk_group was unhashed by the time we got to it.
 */
int blkiocg_del_blkio_group(struct blkio_group *blkg)
{
	struct blkio_cgroup *blkcg;
	unsigned long flags;
	struct cgroup_subsys_state *css;
	int ret = 1;

	rcu_read_lock();
	css = css_lookup(&blkio_subsys, blkg->blkcg_id);
	if (!css)
		goto out;

	blkcg = container_of(css, struct blkio_cgroup, css);
	spin_lock_irqsave(&blkcg->lock, flags);
	if (!hlist_unhashed(&blkg->blkcg_node)) {
		__blkiocg_del_blkio_group(blkg);
		ret = 0;
	}
	spin_unlock_irqrestore(&blkcg->lock, flags);
out:
	rcu_read_unlock();
	return ret;
}
EXPORT_SYMBOL_GPL(blkiocg_del_blkio_group);

/* called under rcu_read_lock(). */
struct blkio_group *blkiocg_lookup_group(struct blkio_cgroup *blkcg, void *key)
{
	struct blkio_group *blkg;
	struct hlist_node *n;
	void *__key;

	hlist_for_each_entry_rcu(blkg, n, &blkcg->blkg_list, blkcg_node) {
		__key = blkg->key;
		if (__key == key)
			return blkg;
	}

	return NULL;
}
EXPORT_SYMBOL_GPL(blkiocg_lookup_group);

#define SHOW_FUNCTION(__VAR)						\
static u64 blkiocg_##__VAR##_read(struct cgroup *cgroup,		\
				       struct cftype *cftype)		\
{									\
	struct blkio_cgroup *blkcg;					\
									\
	blkcg = cgroup_to_blkio_cgroup(cgroup);				\
	return (u64)blkcg->__VAR;					\
}

SHOW_FUNCTION(weight);
#undef SHOW_FUNCTION

static int
blkiocg_weight_write(struct cgroup *cgroup, struct cftype *cftype, u64 val)
{
	struct blkio_cgroup *blkcg;
	struct blkio_group *blkg;
	struct hlist_node *n;
	struct blkio_policy_type *blkiop;

	if (val < BLKIO_WEIGHT_MIN || val > BLKIO_WEIGHT_MAX)
		return -EINVAL;

	blkcg = cgroup_to_blkio_cgroup(cgroup);
	spin_lock(&blkio_list_lock);
	spin_lock_irq(&blkcg->lock);
	blkcg->weight = (unsigned int)val;
	hlist_for_each_entry(blkg, n, &blkcg->blkg_list, blkcg_node) {
		list_for_each_entry(blkiop, &blkio_list, list)
			blkiop->ops.blkio_update_group_weight_fn(blkg,
					blkcg->weight);
	}
	spin_unlock_irq(&blkcg->lock);
	spin_unlock(&blkio_list_lock);
	return 0;
}

static int
blkiocg_reset_write(struct cgroup *cgroup, struct cftype *cftype, u64 val)
{
	struct blkio_cgroup *blkcg;
	struct blkio_group *blkg;
	struct hlist_node *n;
	struct blkio_group_stats *stats;

	blkcg = cgroup_to_blkio_cgroup(cgroup);
	spin_lock_irq(&blkcg->lock);
	hlist_for_each_entry(blkg, n, &blkcg->blkg_list, blkcg_node) {
		spin_lock(&blkg->stats_lock);
		stats = &blkg->stats;
		memset(stats, 0, sizeof(struct blkio_group_stats));
		spin_unlock(&blkg->stats_lock);
	}
	spin_unlock_irq(&blkcg->lock);
	return 0;
}

void get_key_name(int type, char *disk_id, char *str, int chars_left)
{
	strlcpy(str, disk_id, chars_left);
	chars_left -= strlen(str);
	if (chars_left <= 0) {
		printk(KERN_WARNING
			"Possibly incorrect cgroup stat display format");
		return;
	}
	switch (type) {
	case IO_READ:
		strlcat(str, " Read", chars_left);
		break;
	case IO_WRITE:
		strlcat(str, " Write", chars_left);
		break;
	case IO_SYNC:
		strlcat(str, " Sync", chars_left);
		break;
	case IO_ASYNC:
		strlcat(str, " Async", chars_left);
		break;
	case IO_TYPE_MAX:
		strlcat(str, " Total", chars_left);
		break;
	default:
		strlcat(str, " Invalid", chars_left);
	}
}

typedef uint64_t (get_var) (struct blkio_group *, int);

#define MAX_KEY_LEN 100
uint64_t get_typed_stat(struct blkio_group *blkg, struct cgroup_map_cb *cb,
		get_var *getvar, char *disk_id)
{
	uint64_t disk_total;
	char key_str[MAX_KEY_LEN];
	int type;

	for (type = 0; type < IO_TYPE_MAX; type++) {
		get_key_name(type, disk_id, key_str, MAX_KEY_LEN);
		cb->fill(cb, key_str, getvar(blkg, type));
	}
	disk_total = getvar(blkg, IO_READ) + getvar(blkg, IO_WRITE);
	get_key_name(IO_TYPE_MAX, disk_id, key_str, MAX_KEY_LEN);
	cb->fill(cb, key_str, disk_total);
	return disk_total;
}

uint64_t get_stat(struct blkio_group *blkg, struct cgroup_map_cb *cb,
		get_var *getvar, char *disk_id)
{
	uint64_t var = getvar(blkg, 0);
	cb->fill(cb, disk_id, var);
	return var;
}

#define GET_STAT_INDEXED(__VAR)						\
uint64_t get_##__VAR##_stat(struct blkio_group *blkg, int type)		\
{									\
	return blkg->stats.__VAR[type];					\
}									\

GET_STAT_INDEXED(io_service_bytes);
GET_STAT_INDEXED(io_serviced);
GET_STAT_INDEXED(io_service_time);
GET_STAT_INDEXED(io_wait_time);
#undef GET_STAT_INDEXED

#define GET_STAT(__VAR, __CONV)						\
uint64_t get_##__VAR##_stat(struct blkio_group *blkg, int dummy)	\
{									\
	uint64_t data = blkg->stats.__VAR;				\
	if (__CONV)							\
		data = (uint64_t)jiffies_to_msecs(data) * NSEC_PER_MSEC;\
	return data;							\
}

GET_STAT(time, 1);
GET_STAT(sectors, 0);
#ifdef CONFIG_DEBUG_BLK_CGROUP
GET_STAT(dequeue, 0);
#endif
#undef GET_STAT

#define SHOW_FUNCTION_PER_GROUP(__VAR, get_stats, getvar, show_total)	\
static int blkiocg_##__VAR##_read(struct cgroup *cgroup,		\
		struct cftype *cftype, struct cgroup_map_cb *cb)	\
{									\
	struct blkio_cgroup *blkcg;					\
	struct blkio_group *blkg;					\
	struct hlist_node *n;						\
	uint64_t cgroup_total = 0;					\
	char disk_id[10];						\
									\
	if (!cgroup_lock_live_group(cgroup))				\
		return -ENODEV;						\
									\
	blkcg = cgroup_to_blkio_cgroup(cgroup);				\
	rcu_read_lock();						\
	hlist_for_each_entry_rcu(blkg, n, &blkcg->blkg_list, blkcg_node) {\
		if (blkg->dev) {					\
			spin_lock_irq(&blkg->stats_lock);		\
			snprintf(disk_id, 10, "%u:%u", MAJOR(blkg->dev),\
					MINOR(blkg->dev));		\
			cgroup_total += get_stats(blkg, cb, getvar,	\
						disk_id);		\
			spin_unlock_irq(&blkg->stats_lock);		\
		}							\
	}								\
	if (show_total)							\
		cb->fill(cb, "Total", cgroup_total);			\
	rcu_read_unlock();						\
	cgroup_unlock();						\
	return 0;							\
}

SHOW_FUNCTION_PER_GROUP(time, get_stat, get_time_stat, 0);
SHOW_FUNCTION_PER_GROUP(sectors, get_stat, get_sectors_stat, 0);
SHOW_FUNCTION_PER_GROUP(io_service_bytes, get_typed_stat,
			get_io_service_bytes_stat, 1);
SHOW_FUNCTION_PER_GROUP(io_serviced, get_typed_stat, get_io_serviced_stat, 1);
SHOW_FUNCTION_PER_GROUP(io_service_time, get_typed_stat,
			get_io_service_time_stat, 1);
SHOW_FUNCTION_PER_GROUP(io_wait_time, get_typed_stat, get_io_wait_time_stat, 1);
#ifdef CONFIG_DEBUG_BLK_CGROUP
SHOW_FUNCTION_PER_GROUP(dequeue, get_stat, get_dequeue_stat, 0);
#endif
#undef SHOW_FUNCTION_PER_GROUP

#ifdef CONFIG_DEBUG_BLK_CGROUP
void blkiocg_update_blkio_group_dequeue_stats(struct blkio_group *blkg,
			unsigned long dequeue)
{
	blkg->stats.dequeue += dequeue;
}
EXPORT_SYMBOL_GPL(blkiocg_update_blkio_group_dequeue_stats);
#endif

struct cftype blkio_files[] = {
	{
		.name = "weight",
		.read_u64 = blkiocg_weight_read,
		.write_u64 = blkiocg_weight_write,
	},
	{
		.name = "time",
		.read_map = blkiocg_time_read,
		.write_u64 = blkiocg_reset_write,
	},
	{
		.name = "sectors",
		.read_map = blkiocg_sectors_read,
		.write_u64 = blkiocg_reset_write,
	},
	{
		.name = "io_service_bytes",
		.read_map = blkiocg_io_service_bytes_read,
		.write_u64 = blkiocg_reset_write,
	},
	{
		.name = "io_serviced",
		.read_map = blkiocg_io_serviced_read,
		.write_u64 = blkiocg_reset_write,
	},
	{
		.name = "io_service_time",
		.read_map = blkiocg_io_service_time_read,
		.write_u64 = blkiocg_reset_write,
	},
	{
		.name = "io_wait_time",
		.read_map = blkiocg_io_wait_time_read,
		.write_u64 = blkiocg_reset_write,
	},
#ifdef CONFIG_DEBUG_BLK_CGROUP
       {
		.name = "dequeue",
		.read_map = blkiocg_dequeue_read,
       },
#endif
};

static int blkiocg_populate(struct cgroup_subsys *subsys, struct cgroup *cgroup)
{
	return cgroup_add_files(cgroup, subsys, blkio_files,
				ARRAY_SIZE(blkio_files));
}

static void blkiocg_destroy(struct cgroup_subsys *subsys, struct cgroup *cgroup)
{
	struct blkio_cgroup *blkcg = cgroup_to_blkio_cgroup(cgroup);
	unsigned long flags;
	struct blkio_group *blkg;
	void *key;
	struct blkio_policy_type *blkiop;

	rcu_read_lock();
remove_entry:
	spin_lock_irqsave(&blkcg->lock, flags);

	if (hlist_empty(&blkcg->blkg_list)) {
		spin_unlock_irqrestore(&blkcg->lock, flags);
		goto done;
	}

	blkg = hlist_entry(blkcg->blkg_list.first, struct blkio_group,
				blkcg_node);
	key = rcu_dereference(blkg->key);
	__blkiocg_del_blkio_group(blkg);

	spin_unlock_irqrestore(&blkcg->lock, flags);

	/*
	 * This blkio_group is being unlinked as associated cgroup is going
	 * away. Let all the IO controlling policies know about this event.
	 *
	 * Currently this is static call to one io controlling policy. Once
	 * we have more policies in place, we need some dynamic registration
	 * of callback function.
	 */
	spin_lock(&blkio_list_lock);
	list_for_each_entry(blkiop, &blkio_list, list)
		blkiop->ops.blkio_unlink_group_fn(key, blkg);
	spin_unlock(&blkio_list_lock);
	goto remove_entry;
done:
	free_css_id(&blkio_subsys, &blkcg->css);
	rcu_read_unlock();
	if (blkcg != &blkio_root_cgroup)
		kfree(blkcg);
}

static struct cgroup_subsys_state *
blkiocg_create(struct cgroup_subsys *subsys, struct cgroup *cgroup)
{
	struct blkio_cgroup *blkcg, *parent_blkcg;

	if (!cgroup->parent) {
		blkcg = &blkio_root_cgroup;
		goto done;
	}

	/* Currently we do not support hierarchy deeper than two level (0,1) */
	parent_blkcg = cgroup_to_blkio_cgroup(cgroup->parent);
	if (css_depth(&parent_blkcg->css) > 0)
		return ERR_PTR(-EINVAL);

	blkcg = kzalloc(sizeof(*blkcg), GFP_KERNEL);
	if (!blkcg)
		return ERR_PTR(-ENOMEM);

	blkcg->weight = BLKIO_WEIGHT_DEFAULT;
done:
	spin_lock_init(&blkcg->lock);
	INIT_HLIST_HEAD(&blkcg->blkg_list);

	return &blkcg->css;
}

/*
 * We cannot support shared io contexts, as we have no mean to support
 * two tasks with the same ioc in two different groups without major rework
 * of the main cic data structures.  For now we allow a task to change
 * its cgroup only if it's the only owner of its ioc.
 */
static int blkiocg_can_attach(struct cgroup_subsys *subsys,
				struct cgroup *cgroup, struct task_struct *tsk,
				bool threadgroup)
{
	struct io_context *ioc;
	int ret = 0;

	/* task_lock() is needed to avoid races with exit_io_context() */
	task_lock(tsk);
	ioc = tsk->io_context;
	if (ioc && atomic_read(&ioc->nr_tasks) > 1)
		ret = -EINVAL;
	task_unlock(tsk);

	return ret;
}

static void blkiocg_attach(struct cgroup_subsys *subsys, struct cgroup *cgroup,
				struct cgroup *prev, struct task_struct *tsk,
				bool threadgroup)
{
	struct io_context *ioc;

	task_lock(tsk);
	ioc = tsk->io_context;
	if (ioc)
		ioc->cgroup_changed = 1;
	task_unlock(tsk);
}

void blkio_policy_register(struct blkio_policy_type *blkiop)
{
	spin_lock(&blkio_list_lock);
	list_add_tail(&blkiop->list, &blkio_list);
	spin_unlock(&blkio_list_lock);
}
EXPORT_SYMBOL_GPL(blkio_policy_register);

void blkio_policy_unregister(struct blkio_policy_type *blkiop)
{
	spin_lock(&blkio_list_lock);
	list_del_init(&blkiop->list);
	spin_unlock(&blkio_list_lock);
}
EXPORT_SYMBOL_GPL(blkio_policy_unregister);

static int __init init_cgroup_blkio(void)
{
	return cgroup_load_subsys(&blkio_subsys);
}

static void __exit exit_cgroup_blkio(void)
{
	cgroup_unload_subsys(&blkio_subsys);
}

module_init(init_cgroup_blkio);
module_exit(exit_cgroup_blkio);
MODULE_LICENSE("GPL");
Commit	Line	Data
31e4c28d VG	1	/*
	2	* Common Block IO controller cgroup interface
	3	*
	4	* Based on ideas and code from CFQ, CFS and BFQ:
	5	* Copyright (C) 2003 Jens Axboe <axboe@kernel.dk>
	6	*
	7	* Copyright (C) 2008 Fabio Checconi <fabio@gandalf.sssup.it>
	8	* Paolo Valente <paolo.valente@unimore.it>
	9	*
	10	* Copyright (C) 2009 Vivek Goyal <vgoyal@redhat.com>
	11	* Nauman Rafique <nauman@google.com>
	12	*/
	13	#include <linux/ioprio.h>
22084190 VG	14	#include <linux/seq_file.h>
22084190 VG	15	#include <linux/kdev_t.h>
9d6a986c	16	#include <linux/module.h>
accee785	17	#include <linux/err.h>
31e4c28d	18	#include "blk-cgroup.h"
3e252066 VG	19
	20	static DEFINE_SPINLOCK(blkio_list_lock);
	21	static LIST_HEAD(blkio_list);
b1c35769	22
31e4c28d	23	struct blkio_cgroup blkio_root_cgroup = { .weight = 2*BLKIO_WEIGHT_DEFAULT };
9d6a986c VG	24	EXPORT_SYMBOL_GPL(blkio_root_cgroup);
9d6a986c VG	25
67523c48 BB	26	static struct cgroup_subsys_state blkiocg_create(struct cgroup_subsys ,
	27	struct cgroup *);
	28	static int blkiocg_can_attach(struct cgroup_subsys , struct cgroup ,
	29	struct task_struct *, bool);
	30	static void blkiocg_attach(struct cgroup_subsys , struct cgroup ,
	31	struct cgroup , struct task_struct , bool);
	32	static void blkiocg_destroy(struct cgroup_subsys , struct cgroup );
	33	static int blkiocg_populate(struct cgroup_subsys , struct cgroup );
	34
	35	struct cgroup_subsys blkio_subsys = {
	36	.name = "blkio",
	37	.create = blkiocg_create,
	38	.can_attach = blkiocg_can_attach,
	39	.attach = blkiocg_attach,
	40	.destroy = blkiocg_destroy,
	41	.populate = blkiocg_populate,
	42	#ifdef CONFIG_BLK_CGROUP
	43	/* note: blkio_subsys_id is otherwise defined in blk-cgroup.h */
	44	.subsys_id = blkio_subsys_id,
	45	#endif
	46	.use_id = 1,
	47	.module = THIS_MODULE,
	48	};
	49	EXPORT_SYMBOL_GPL(blkio_subsys);
	50
31e4c28d VG	51	struct blkio_cgroup cgroup_to_blkio_cgroup(struct cgroup cgroup)
	52	{
	53	return container_of(cgroup_subsys_state(cgroup, blkio_subsys_id),
	54	struct blkio_cgroup, css);
	55	}
9d6a986c	56	EXPORT_SYMBOL_GPL(cgroup_to_blkio_cgroup);
31e4c28d	57
303a3acb	58	void blkiocg_update_timeslice_used(struct blkio_group *blkg, unsigned long time)
22084190	59	{
303a3acb DS	60	unsigned long flags;
	61
	62	spin_lock_irqsave(&blkg->stats_lock, flags);
	63	blkg->stats.time += time;
	64	spin_unlock_irqrestore(&blkg->stats_lock, flags);
22084190	65	}
303a3acb	66	EXPORT_SYMBOL_GPL(blkiocg_update_timeslice_used);
22084190	67
31e4c28d	68	void blkiocg_add_blkio_group(struct blkio_cgroup *blkcg,
22084190	69	struct blkio_group blkg, void key, dev_t dev)
31e4c28d VG	70	{
	71	unsigned long flags;
	72
	73	spin_lock_irqsave(&blkcg->lock, flags);
	74	rcu_assign_pointer(blkg->key, key);
b1c35769	75	blkg->blkcg_id = css_id(&blkcg->css);
31e4c28d VG	76	hlist_add_head_rcu(&blkg->blkcg_node, &blkcg->blkg_list);
31e4c28d VG	77	spin_unlock_irqrestore(&blkcg->lock, flags);
2868ef7b VG	78	#ifdef CONFIG_DEBUG_BLK_CGROUP
	79	/* Need to take css reference ? */
	80	cgroup_path(blkcg->css.cgroup, blkg->path, sizeof(blkg->path));
	81	#endif
22084190	82	blkg->dev = dev;
31e4c28d	83	}
9d6a986c	84	EXPORT_SYMBOL_GPL(blkiocg_add_blkio_group);
31e4c28d	85
b1c35769 VG	86	static void __blkiocg_del_blkio_group(struct blkio_group *blkg)
	87	{
	88	hlist_del_init_rcu(&blkg->blkcg_node);
	89	blkg->blkcg_id = 0;
	90	}
	91
	92	/*
	93	* returns 0 if blkio_group was still on cgroup list. Otherwise returns 1
	94	* indicating that blk_group was unhashed by the time we got to it.
	95	*/
31e4c28d VG	96	int blkiocg_del_blkio_group(struct blkio_group *blkg)
31e4c28d VG	97	{
b1c35769 VG	98	struct blkio_cgroup *blkcg;
	99	unsigned long flags;
	100	struct cgroup_subsys_state *css;
	101	int ret = 1;
	102
	103	rcu_read_lock();
	104	css = css_lookup(&blkio_subsys, blkg->blkcg_id);
	105	if (!css)
	106	goto out;
	107
	108	blkcg = container_of(css, struct blkio_cgroup, css);
	109	spin_lock_irqsave(&blkcg->lock, flags);
	110	if (!hlist_unhashed(&blkg->blkcg_node)) {
	111	__blkiocg_del_blkio_group(blkg);
	112	ret = 0;
	113	}
	114	spin_unlock_irqrestore(&blkcg->lock, flags);
	115	out:
	116	rcu_read_unlock();
	117	return ret;
31e4c28d	118	}
9d6a986c	119	EXPORT_SYMBOL_GPL(blkiocg_del_blkio_group);
31e4c28d VG	120
	121	/* called under rcu_read_lock(). */
	122	struct blkio_group blkiocg_lookup_group(struct blkio_cgroup blkcg, void *key)
	123	{
	124	struct blkio_group *blkg;
	125	struct hlist_node *n;
	126	void *__key;
	127
	128	hlist_for_each_entry_rcu(blkg, n, &blkcg->blkg_list, blkcg_node) {
	129	__key = blkg->key;
	130	if (__key == key)
	131	return blkg;
	132	}
	133
	134	return NULL;
	135	}
9d6a986c	136	EXPORT_SYMBOL_GPL(blkiocg_lookup_group);
31e4c28d VG	137
	138	#define SHOW_FUNCTION(__VAR) \
	139	static u64 blkiocg_##__VAR##_read(struct cgroup *cgroup, \
	140	struct cftype *cftype) \
	141	{ \
	142	struct blkio_cgroup *blkcg; \
	143	\
	144	blkcg = cgroup_to_blkio_cgroup(cgroup); \
	145	return (u64)blkcg->__VAR; \
	146	}
	147
	148	SHOW_FUNCTION(weight);
	149	#undef SHOW_FUNCTION
	150
	151	static int
	152	blkiocg_weight_write(struct cgroup cgroup, struct cftype cftype, u64 val)
	153	{
	154	struct blkio_cgroup *blkcg;
f8d461d6 VG	155	struct blkio_group *blkg;
f8d461d6 VG	156	struct hlist_node *n;
3e252066	157	struct blkio_policy_type *blkiop;
31e4c28d VG	158
	159	if (val < BLKIO_WEIGHT_MIN \|\| val > BLKIO_WEIGHT_MAX)
	160	return -EINVAL;
	161
	162	blkcg = cgroup_to_blkio_cgroup(cgroup);
bcf4dd43	163	spin_lock(&blkio_list_lock);
f8d461d6	164	spin_lock_irq(&blkcg->lock);
31e4c28d	165	blkcg->weight = (unsigned int)val;
3e252066	166	hlist_for_each_entry(blkg, n, &blkcg->blkg_list, blkcg_node) {
3e252066 VG	167	list_for_each_entry(blkiop, &blkio_list, list)
	168	blkiop->ops.blkio_update_group_weight_fn(blkg,
	169	blkcg->weight);
3e252066	170	}
f8d461d6	171	spin_unlock_irq(&blkcg->lock);
bcf4dd43	172	spin_unlock(&blkio_list_lock);
31e4c28d VG	173	return 0;
	174	}
	175
303a3acb DS	176	static int
	177	blkiocg_reset_write(struct cgroup cgroup, struct cftype cftype, u64 val)
	178	{
	179	struct blkio_cgroup *blkcg;
	180	struct blkio_group *blkg;
	181	struct hlist_node *n;
	182	struct blkio_group_stats *stats;
	183
	184	blkcg = cgroup_to_blkio_cgroup(cgroup);
	185	spin_lock_irq(&blkcg->lock);
	186	hlist_for_each_entry(blkg, n, &blkcg->blkg_list, blkcg_node) {
	187	spin_lock(&blkg->stats_lock);
	188	stats = &blkg->stats;
	189	memset(stats, 0, sizeof(struct blkio_group_stats));
	190	spin_unlock(&blkg->stats_lock);
	191	}
	192	spin_unlock_irq(&blkcg->lock);
	193	return 0;
	194	}
	195
	196	void get_key_name(int type, char disk_id, char str, int chars_left)
	197	{
	198	strlcpy(str, disk_id, chars_left);
	199	chars_left -= strlen(str);
	200	if (chars_left <= 0) {
	201	printk(KERN_WARNING
	202	"Possibly incorrect cgroup stat display format");
	203	return;
	204	}
	205	switch (type) {
	206	case IO_READ:
	207	strlcat(str, " Read", chars_left);
	208	break;
	209	case IO_WRITE:
	210	strlcat(str, " Write", chars_left);
	211	break;
	212	case IO_SYNC:
	213	strlcat(str, " Sync", chars_left);
	214	break;
	215	case IO_ASYNC:
	216	strlcat(str, " Async", chars_left);
	217	break;
	218	case IO_TYPE_MAX:
	219	strlcat(str, " Total", chars_left);
	220	break;
	221	default:
	222	strlcat(str, " Invalid", chars_left);
	223	}
	224	}
	225
	226	typedef uint64_t (get_var) (struct blkio_group *, int);
	227
	228	#define MAX_KEY_LEN 100
	229	uint64_t get_typed_stat(struct blkio_group blkg, struct cgroup_map_cb cb,
	230	get_var getvar, char disk_id)
	231	{
	232	uint64_t disk_total;
	233	char key_str[MAX_KEY_LEN];
	234	int type;
	235
	236	for (type = 0; type < IO_TYPE_MAX; type++) {
	237	get_key_name(type, disk_id, key_str, MAX_KEY_LEN);
	238	cb->fill(cb, key_str, getvar(blkg, type));
	239	}
240	disk_total = getvar(blkg, IO_READ) + getvar(blkg, IO_WRITE);
241	get_key_name(IO_TYPE_MAX, disk_id, key_str, MAX_KEY_LEN);
242	cb->fill(cb, key_str, disk_total);
243	return disk_total;
244	}
245
246	uint64_t get_stat(struct blkio_group blkg, struct cgroup_map_cb cb,
247	get_var getvar, char disk_id)
248	{
249	uint64_t var = getvar(blkg, 0);
250	cb->fill(cb, disk_id, var);
251	return var;
252	}
253
254	#define GET_STAT_INDEXED(__VAR) \
255	uint64_t get_##__VAR##_stat(struct blkio_group *blkg, int type) \
256	{ \
257	return blkg->stats.__VAR[type]; \
258	} \
259
260	GET_STAT_INDEXED(io_service_bytes);
261	GET_STAT_INDEXED(io_serviced);
262	GET_STAT_INDEXED(io_service_time);
263	GET_STAT_INDEXED(io_wait_time);
264	#undef GET_STAT_INDEXED
265
266	#define GET_STAT(__VAR, __CONV) \
267	uint64_t get_##__VAR##_stat(struct blkio_group *blkg, int dummy) \
268	{ \
269	uint64_t data = blkg->stats.__VAR; \
270	if (__CONV) \
271	data = (uint64_t)jiffies_to_msecs(data) * NSEC_PER_MSEC;\
272	return data; \
273	}
274
275	GET_STAT(time, 1);
276	GET_STAT(sectors, 0);
277	#ifdef CONFIG_DEBUG_BLK_CGROUP
278	GET_STAT(dequeue, 0);
279	#endif
280	#undef GET_STAT
281
282	#define SHOW_FUNCTION_PER_GROUP(__VAR, get_stats, getvar, show_total) \
22084190	283	static int blkiocg_##__VAR##_read(struct cgroup *cgroup, \
303a3acb	284	struct cftype cftype, struct cgroup_map_cb cb) \
22084190 VG	285	{ \
	286	struct blkio_cgroup *blkcg; \
	287	struct blkio_group *blkg; \
	288	struct hlist_node *n; \
303a3acb DS	289	uint64_t cgroup_total = 0; \
303a3acb DS	290	char disk_id[10]; \
22084190 VG	291	\
	292	if (!cgroup_lock_live_group(cgroup)) \
	293	return -ENODEV; \
	294	\
	295	blkcg = cgroup_to_blkio_cgroup(cgroup); \
	296	rcu_read_lock(); \
	297	hlist_for_each_entry_rcu(blkg, n, &blkcg->blkg_list, blkcg_node) {\
303a3acb DS	298	if (blkg->dev) { \
	299	spin_lock_irq(&blkg->stats_lock); \
	300	snprintf(disk_id, 10, "%u:%u", MAJOR(blkg->dev),\
	301	MINOR(blkg->dev)); \
	302	cgroup_total += get_stats(blkg, cb, getvar, \
	303	disk_id); \
	304	spin_unlock_irq(&blkg->stats_lock); \
	305	} \
22084190	306	} \
303a3acb DS	307	if (show_total) \
303a3acb DS	308	cb->fill(cb, "Total", cgroup_total); \
22084190 VG	309	rcu_read_unlock(); \
	310	cgroup_unlock(); \
	311	return 0; \
	312	}
	313
303a3acb DS	314	SHOW_FUNCTION_PER_GROUP(time, get_stat, get_time_stat, 0);
	315	SHOW_FUNCTION_PER_GROUP(sectors, get_stat, get_sectors_stat, 0);
	316	SHOW_FUNCTION_PER_GROUP(io_service_bytes, get_typed_stat,
	317	get_io_service_bytes_stat, 1);
	318	SHOW_FUNCTION_PER_GROUP(io_serviced, get_typed_stat, get_io_serviced_stat, 1);
	319	SHOW_FUNCTION_PER_GROUP(io_service_time, get_typed_stat,
	320	get_io_service_time_stat, 1);
	321	SHOW_FUNCTION_PER_GROUP(io_wait_time, get_typed_stat, get_io_wait_time_stat, 1);
22084190	322	#ifdef CONFIG_DEBUG_BLK_CGROUP
303a3acb	323	SHOW_FUNCTION_PER_GROUP(dequeue, get_stat, get_dequeue_stat, 0);
22084190 VG	324	#endif
	325	#undef SHOW_FUNCTION_PER_GROUP
	326
	327	#ifdef CONFIG_DEBUG_BLK_CGROUP
	328	void blkiocg_update_blkio_group_dequeue_stats(struct blkio_group *blkg,
	329	unsigned long dequeue)
	330	{
303a3acb	331	blkg->stats.dequeue += dequeue;
22084190	332	}
9d6a986c	333	EXPORT_SYMBOL_GPL(blkiocg_update_blkio_group_dequeue_stats);
22084190 VG	334	#endif
22084190 VG	335
31e4c28d VG	336	struct cftype blkio_files[] = {
	337	{
	338	.name = "weight",
	339	.read_u64 = blkiocg_weight_read,
	340	.write_u64 = blkiocg_weight_write,
	341	},
22084190 VG	342	{
22084190 VG	343	.name = "time",
303a3acb DS	344	.read_map = blkiocg_time_read,
303a3acb DS	345	.write_u64 = blkiocg_reset_write,
22084190 VG	346	},
	347	{
	348	.name = "sectors",
303a3acb DS	349	.read_map = blkiocg_sectors_read,
	350	.write_u64 = blkiocg_reset_write,
	351	},
	352	{
	353	.name = "io_service_bytes",
	354	.read_map = blkiocg_io_service_bytes_read,
	355	.write_u64 = blkiocg_reset_write,
	356	},
	357	{
	358	.name = "io_serviced",
	359	.read_map = blkiocg_io_serviced_read,
	360	.write_u64 = blkiocg_reset_write,
	361	},
	362	{
	363	.name = "io_service_time",
	364	.read_map = blkiocg_io_service_time_read,
	365	.write_u64 = blkiocg_reset_write,
	366	},
	367	{
	368	.name = "io_wait_time",
	369	.read_map = blkiocg_io_wait_time_read,
	370	.write_u64 = blkiocg_reset_write,
22084190 VG	371	},
	372	#ifdef CONFIG_DEBUG_BLK_CGROUP
	373	{
	374	.name = "dequeue",
303a3acb	375	.read_map = blkiocg_dequeue_read,
22084190 VG	376	},
22084190 VG	377	#endif
31e4c28d VG	378	};
	379
	380	static int blkiocg_populate(struct cgroup_subsys subsys, struct cgroup cgroup)
	381	{
	382	return cgroup_add_files(cgroup, subsys, blkio_files,
	383	ARRAY_SIZE(blkio_files));
	384	}
	385
	386	static void blkiocg_destroy(struct cgroup_subsys subsys, struct cgroup cgroup)
	387	{
	388	struct blkio_cgroup *blkcg = cgroup_to_blkio_cgroup(cgroup);
b1c35769 VG	389	unsigned long flags;
	390	struct blkio_group *blkg;
	391	void *key;
3e252066	392	struct blkio_policy_type *blkiop;
b1c35769 VG	393
	394	rcu_read_lock();
	395	remove_entry:
	396	spin_lock_irqsave(&blkcg->lock, flags);
	397
	398	if (hlist_empty(&blkcg->blkg_list)) {
	399	spin_unlock_irqrestore(&blkcg->lock, flags);
	400	goto done;
	401	}
	402
	403	blkg = hlist_entry(blkcg->blkg_list.first, struct blkio_group,
	404	blkcg_node);
	405	key = rcu_dereference(blkg->key);
	406	__blkiocg_del_blkio_group(blkg);
31e4c28d	407
b1c35769 VG	408	spin_unlock_irqrestore(&blkcg->lock, flags);
	409
	410	/*
	411	* This blkio_group is being unlinked as associated cgroup is going
	412	* away. Let all the IO controlling policies know about this event.
	413	*
	414	* Currently this is static call to one io controlling policy. Once
	415	* we have more policies in place, we need some dynamic registration
	416	* of callback function.
	417	*/
3e252066 VG	418	spin_lock(&blkio_list_lock);
	419	list_for_each_entry(blkiop, &blkio_list, list)
	420	blkiop->ops.blkio_unlink_group_fn(key, blkg);
	421	spin_unlock(&blkio_list_lock);
b1c35769 VG	422	goto remove_entry;
b1c35769 VG	423	done:
31e4c28d	424	free_css_id(&blkio_subsys, &blkcg->css);
b1c35769	425	rcu_read_unlock();
67523c48 BB	426	if (blkcg != &blkio_root_cgroup)
67523c48 BB	427	kfree(blkcg);
31e4c28d VG	428	}
	429
	430	static struct cgroup_subsys_state *
	431	blkiocg_create(struct cgroup_subsys subsys, struct cgroup cgroup)
	432	{
	433	struct blkio_cgroup blkcg, parent_blkcg;
	434
	435	if (!cgroup->parent) {
	436	blkcg = &blkio_root_cgroup;
	437	goto done;
	438	}
	439
	440	/* Currently we do not support hierarchy deeper than two level (0,1) */
	441	parent_blkcg = cgroup_to_blkio_cgroup(cgroup->parent);
	442	if (css_depth(&parent_blkcg->css) > 0)
	443	return ERR_PTR(-EINVAL);
	444
	445	blkcg = kzalloc(sizeof(*blkcg), GFP_KERNEL);
	446	if (!blkcg)
	447	return ERR_PTR(-ENOMEM);
	448
	449	blkcg->weight = BLKIO_WEIGHT_DEFAULT;
	450	done:
	451	spin_lock_init(&blkcg->lock);
	452	INIT_HLIST_HEAD(&blkcg->blkg_list);
	453
	454	return &blkcg->css;
	455	}
	456
	457	/*
	458	* We cannot support shared io contexts, as we have no mean to support
	459	* two tasks with the same ioc in two different groups without major rework
	460	* of the main cic data structures. For now we allow a task to change
	461	* its cgroup only if it's the only owner of its ioc.
	462	*/
	463	static int blkiocg_can_attach(struct cgroup_subsys *subsys,
	464	struct cgroup cgroup, struct task_struct tsk,
	465	bool threadgroup)
	466	{
	467	struct io_context *ioc;
	468	int ret = 0;
	469
	470	/* task_lock() is needed to avoid races with exit_io_context() */
	471	task_lock(tsk);
	472	ioc = tsk->io_context;
	473	if (ioc && atomic_read(&ioc->nr_tasks) > 1)
	474	ret = -EINVAL;
	475	task_unlock(tsk);
	476
	477	return ret;
	478	}
	479
	480	static void blkiocg_attach(struct cgroup_subsys subsys, struct cgroup cgroup,
	481	struct cgroup prev, struct task_struct tsk,
	482	bool threadgroup)
	483	{
	484	struct io_context *ioc;
	485
	486	task_lock(tsk);
	487	ioc = tsk->io_context;
	488	if (ioc)
	489	ioc->cgroup_changed = 1;
	490	task_unlock(tsk);
	491	}
492
3e252066 VG	493	void blkio_policy_register(struct blkio_policy_type *blkiop)
	494	{
	495	spin_lock(&blkio_list_lock);
	496	list_add_tail(&blkiop->list, &blkio_list);
	497	spin_unlock(&blkio_list_lock);
	498	}
	499	EXPORT_SYMBOL_GPL(blkio_policy_register);
	500
	501	void blkio_policy_unregister(struct blkio_policy_type *blkiop)
	502	{
	503	spin_lock(&blkio_list_lock);
	504	list_del_init(&blkiop->list);
	505	spin_unlock(&blkio_list_lock);
	506	}
	507	EXPORT_SYMBOL_GPL(blkio_policy_unregister);
67523c48 BB	508
	509	static int __init init_cgroup_blkio(void)
	510	{
	511	return cgroup_load_subsys(&blkio_subsys);
	512	}
	513
	514	static void __exit exit_cgroup_blkio(void)
	515	{
	516	cgroup_unload_subsys(&blkio_subsys);
	517	}
	518
	519	module_init(init_cgroup_blkio);
	520	module_exit(exit_cgroup_blkio);
	521	MODULE_LICENSE("GPL");