[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 <linux/blkdev.h>
#include "blk-cgroup.h"

#define MAX_KEY_LEN 100

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 blkio_group_init(struct blkio_group *blkg)
{
	spin_lock_init(&blkg->stats_lock);
}
EXPORT_SYMBOL_GPL(blkio_group_init);

/*
 * Add to the appropriate stat variable depending on the request type.
 * This should be called with the blkg->stats_lock held.
 */
static void blkio_add_stat(uint64_t *stat, uint64_t add, bool direction,
				bool sync)
{
	if (direction)
		stat[BLKIO_STAT_WRITE] += add;
	else
		stat[BLKIO_STAT_READ] += add;
	if (sync)
		stat[BLKIO_STAT_SYNC] += add;
	else
		stat[BLKIO_STAT_ASYNC] += add;
}

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_update_dispatch_stats(struct blkio_group *blkg,
				uint64_t bytes, bool direction, bool sync)
{
	struct blkio_group_stats *stats;
	unsigned long flags;

	spin_lock_irqsave(&blkg->stats_lock, flags);
	stats = &blkg->stats;
	stats->sectors += bytes >> 9;
	blkio_add_stat(stats->stat_arr[BLKIO_STAT_SERVICED], 1, direction,
			sync);
	blkio_add_stat(stats->stat_arr[BLKIO_STAT_SERVICE_BYTES], bytes,
			direction, sync);
	spin_unlock_irqrestore(&blkg->stats_lock, flags);
}
EXPORT_SYMBOL_GPL(blkiocg_update_dispatch_stats);

void blkiocg_update_completion_stats(struct blkio_group *blkg,
	uint64_t start_time, uint64_t io_start_time, bool direction, bool sync)
{
	struct blkio_group_stats *stats;
	unsigned long flags;
	unsigned long long now = sched_clock();

	spin_lock_irqsave(&blkg->stats_lock, flags);
	stats = &blkg->stats;
	if (time_after64(now, io_start_time))
		blkio_add_stat(stats->stat_arr[BLKIO_STAT_SERVICE_TIME],
				now - io_start_time, direction, sync);
	if (time_after64(io_start_time, start_time))
		blkio_add_stat(stats->stat_arr[BLKIO_STAT_WAIT_TIME],
				io_start_time - start_time, direction, sync);
	spin_unlock_irqrestore(&blkg->stats_lock, flags);
}
EXPORT_SYMBOL_GPL(blkiocg_update_completion_stats);

void blkiocg_update_io_merged_stats(struct blkio_group *blkg, bool direction,
					bool sync)
{
	unsigned long flags;

	spin_lock_irqsave(&blkg->stats_lock, flags);
	blkio_add_stat(blkg->stats.stat_arr[BLKIO_STAT_MERGED], 1, direction,
			sync);
	spin_unlock_irqrestore(&blkg->stats_lock, flags);
}
EXPORT_SYMBOL_GPL(blkiocg_update_io_merged_stats);

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_stats(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;
}

static void blkio_get_key_name(enum stat_sub_type type, dev_t dev, char *str,
				int chars_left, bool diskname_only)
{
	snprintf(str, chars_left, "%d:%d", MAJOR(dev), MINOR(dev));
	chars_left -= strlen(str);
	if (chars_left <= 0) {
		printk(KERN_WARNING
			"Possibly incorrect cgroup stat display format");
		return;
	}
	if (diskname_only)
		return;
	switch (type) {
	case BLKIO_STAT_READ:
		strlcat(str, " Read", chars_left);
		break;
	case BLKIO_STAT_WRITE:
		strlcat(str, " Write", chars_left);
		break;
	case BLKIO_STAT_SYNC:
		strlcat(str, " Sync", chars_left);
		break;
	case BLKIO_STAT_ASYNC:
		strlcat(str, " Async", chars_left);
		break;
	case BLKIO_STAT_TOTAL:
		strlcat(str, " Total", chars_left);
		break;
	default:
		strlcat(str, " Invalid", chars_left);
	}
}

static uint64_t blkio_fill_stat(char *str, int chars_left, uint64_t val,
				struct cgroup_map_cb *cb, dev_t dev)
{
	blkio_get_key_name(0, dev, str, chars_left, true);
	cb->fill(cb, str, val);
	return val;
}

/* This should be called with blkg->stats_lock held */
static uint64_t blkio_get_stat(struct blkio_group *blkg,
		struct cgroup_map_cb *cb, dev_t dev, enum stat_type type)
{
	uint64_t disk_total;
	char key_str[MAX_KEY_LEN];
	enum stat_sub_type sub_type;

	if (type == BLKIO_STAT_TIME)
		return blkio_fill_stat(key_str, MAX_KEY_LEN - 1,
					blkg->stats.time, cb, dev);
	if (type == BLKIO_STAT_SECTORS)
		return blkio_fill_stat(key_str, MAX_KEY_LEN - 1,
					blkg->stats.sectors, cb, dev);
#ifdef CONFIG_DEBUG_BLK_CGROUP
	if (type == BLKIO_STAT_DEQUEUE)
		return blkio_fill_stat(key_str, MAX_KEY_LEN - 1,
					blkg->stats.dequeue, cb, dev);
#endif

	for (sub_type = BLKIO_STAT_READ; sub_type < BLKIO_STAT_TOTAL;
			sub_type++) {
		blkio_get_key_name(sub_type, dev, key_str, MAX_KEY_LEN, false);
		cb->fill(cb, key_str, blkg->stats.stat_arr[type][sub_type]);
	}
	disk_total = blkg->stats.stat_arr[type][BLKIO_STAT_READ] +
			blkg->stats.stat_arr[type][BLKIO_STAT_WRITE];
	blkio_get_key_name(BLKIO_STAT_TOTAL, dev, key_str, MAX_KEY_LEN, false);
	cb->fill(cb, key_str, disk_total);
	return disk_total;
}

#define SHOW_FUNCTION_PER_GROUP(__VAR, type, 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;					\
									\
	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);		\
			cgroup_total += blkio_get_stat(blkg, cb,	\
						blkg->dev, type);	\
			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, BLKIO_STAT_TIME, 0);
SHOW_FUNCTION_PER_GROUP(sectors, BLKIO_STAT_SECTORS, 0);
SHOW_FUNCTION_PER_GROUP(io_service_bytes, BLKIO_STAT_SERVICE_BYTES, 1);
SHOW_FUNCTION_PER_GROUP(io_serviced, BLKIO_STAT_SERVICED, 1);
SHOW_FUNCTION_PER_GROUP(io_service_time, BLKIO_STAT_SERVICE_TIME, 1);
SHOW_FUNCTION_PER_GROUP(io_wait_time, BLKIO_STAT_WAIT_TIME, 1);
SHOW_FUNCTION_PER_GROUP(io_merged, BLKIO_STAT_MERGED, 1);
#ifdef CONFIG_DEBUG_BLK_CGROUP
SHOW_FUNCTION_PER_GROUP(dequeue, BLKIO_STAT_DEQUEUE, 0);
#endif
#undef SHOW_FUNCTION_PER_GROUP

#ifdef CONFIG_DEBUG_BLK_CGROUP
void blkiocg_update_dequeue_stats(struct blkio_group *blkg,
			unsigned long dequeue)
{
	blkg->stats.dequeue += dequeue;
}
EXPORT_SYMBOL_GPL(blkiocg_update_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,
	},
	{
		.name = "sectors",
		.read_map = blkiocg_sectors_read,
	},
	{
		.name = "io_service_bytes",
		.read_map = blkiocg_io_service_bytes_read,
	},
	{
		.name = "io_serviced",
		.read_map = blkiocg_io_serviced_read,
	},
	{
		.name = "io_service_time",
		.read_map = blkiocg_io_service_time_read,
	},
	{
		.name = "io_wait_time",
		.read_map = blkiocg_io_wait_time_read,
	},
	{
		.name = "io_merged",
		.read_map = blkiocg_io_merged_read,
	},
	{
		.name = "reset_stats",
		.write_u64 = blkiocg_reset_stats,
	},
#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>
9195291e	18	#include <linux/blkdev.h>
31e4c28d	19	#include "blk-cgroup.h"
3e252066	20
84c124da DS	21	#define MAX_KEY_LEN 100
84c124da DS	22
3e252066 VG	23	static DEFINE_SPINLOCK(blkio_list_lock);
3e252066 VG	24	static LIST_HEAD(blkio_list);
b1c35769	25
31e4c28d	26	struct blkio_cgroup blkio_root_cgroup = { .weight = 2*BLKIO_WEIGHT_DEFAULT };
9d6a986c VG	27	EXPORT_SYMBOL_GPL(blkio_root_cgroup);
9d6a986c VG	28
67523c48 BB	29	static struct cgroup_subsys_state blkiocg_create(struct cgroup_subsys ,
	30	struct cgroup *);
	31	static int blkiocg_can_attach(struct cgroup_subsys , struct cgroup ,
	32	struct task_struct *, bool);
	33	static void blkiocg_attach(struct cgroup_subsys , struct cgroup ,
	34	struct cgroup , struct task_struct , bool);
	35	static void blkiocg_destroy(struct cgroup_subsys , struct cgroup );
	36	static int blkiocg_populate(struct cgroup_subsys , struct cgroup );
	37
	38	struct cgroup_subsys blkio_subsys = {
	39	.name = "blkio",
	40	.create = blkiocg_create,
	41	.can_attach = blkiocg_can_attach,
	42	.attach = blkiocg_attach,
	43	.destroy = blkiocg_destroy,
	44	.populate = blkiocg_populate,
	45	#ifdef CONFIG_BLK_CGROUP
	46	/* note: blkio_subsys_id is otherwise defined in blk-cgroup.h */
	47	.subsys_id = blkio_subsys_id,
	48	#endif
	49	.use_id = 1,
	50	.module = THIS_MODULE,
	51	};
	52	EXPORT_SYMBOL_GPL(blkio_subsys);
	53
31e4c28d VG	54	struct blkio_cgroup cgroup_to_blkio_cgroup(struct cgroup cgroup)
	55	{
	56	return container_of(cgroup_subsys_state(cgroup, blkio_subsys_id),
	57	struct blkio_cgroup, css);
	58	}
9d6a986c	59	EXPORT_SYMBOL_GPL(cgroup_to_blkio_cgroup);
31e4c28d	60
84c124da DS	61	void blkio_group_init(struct blkio_group *blkg)
	62	{
	63	spin_lock_init(&blkg->stats_lock);
	64	}
	65	EXPORT_SYMBOL_GPL(blkio_group_init);
	66
9195291e DS	67	/*
	68	* Add to the appropriate stat variable depending on the request type.
	69	* This should be called with the blkg->stats_lock held.
	70	*/
84c124da DS	71	static void blkio_add_stat(uint64_t *stat, uint64_t add, bool direction,
84c124da DS	72	bool sync)
9195291e	73	{
84c124da DS	74	if (direction)
84c124da DS	75	stat[BLKIO_STAT_WRITE] += add;
9195291e	76	else
84c124da DS	77	stat[BLKIO_STAT_READ] += add;
	78	if (sync)
	79	stat[BLKIO_STAT_SYNC] += add;
9195291e	80	else
84c124da	81	stat[BLKIO_STAT_ASYNC] += add;
9195291e DS	82	}
9195291e DS	83
303a3acb	84	void blkiocg_update_timeslice_used(struct blkio_group *blkg, unsigned long time)
22084190	85	{
303a3acb DS	86	unsigned long flags;
	87
	88	spin_lock_irqsave(&blkg->stats_lock, flags);
	89	blkg->stats.time += time;
	90	spin_unlock_irqrestore(&blkg->stats_lock, flags);
22084190	91	}
303a3acb	92	EXPORT_SYMBOL_GPL(blkiocg_update_timeslice_used);
22084190	93
84c124da DS	94	void blkiocg_update_dispatch_stats(struct blkio_group *blkg,
84c124da DS	95	uint64_t bytes, bool direction, bool sync)
9195291e DS	96	{
	97	struct blkio_group_stats *stats;
	98	unsigned long flags;
	99
	100	spin_lock_irqsave(&blkg->stats_lock, flags);
	101	stats = &blkg->stats;
84c124da DS	102	stats->sectors += bytes >> 9;
	103	blkio_add_stat(stats->stat_arr[BLKIO_STAT_SERVICED], 1, direction,
	104	sync);
	105	blkio_add_stat(stats->stat_arr[BLKIO_STAT_SERVICE_BYTES], bytes,
	106	direction, sync);
9195291e DS	107	spin_unlock_irqrestore(&blkg->stats_lock, flags);
9195291e DS	108	}
84c124da	109	EXPORT_SYMBOL_GPL(blkiocg_update_dispatch_stats);
9195291e	110
84c124da DS	111	void blkiocg_update_completion_stats(struct blkio_group *blkg,
84c124da DS	112	uint64_t start_time, uint64_t io_start_time, bool direction, bool sync)
9195291e DS	113	{
	114	struct blkio_group_stats *stats;
	115	unsigned long flags;
	116	unsigned long long now = sched_clock();
	117
	118	spin_lock_irqsave(&blkg->stats_lock, flags);
	119	stats = &blkg->stats;
84c124da DS	120	if (time_after64(now, io_start_time))
	121	blkio_add_stat(stats->stat_arr[BLKIO_STAT_SERVICE_TIME],
	122	now - io_start_time, direction, sync);
	123	if (time_after64(io_start_time, start_time))
	124	blkio_add_stat(stats->stat_arr[BLKIO_STAT_WAIT_TIME],
	125	io_start_time - start_time, direction, sync);
9195291e DS	126	spin_unlock_irqrestore(&blkg->stats_lock, flags);
9195291e DS	127	}
84c124da	128	EXPORT_SYMBOL_GPL(blkiocg_update_completion_stats);
9195291e	129
812d4026 DS	130	void blkiocg_update_io_merged_stats(struct blkio_group *blkg, bool direction,
	131	bool sync)
	132	{
	133	unsigned long flags;
	134
	135	spin_lock_irqsave(&blkg->stats_lock, flags);
	136	blkio_add_stat(blkg->stats.stat_arr[BLKIO_STAT_MERGED], 1, direction,
	137	sync);
	138	spin_unlock_irqrestore(&blkg->stats_lock, flags);
	139	}
	140	EXPORT_SYMBOL_GPL(blkiocg_update_io_merged_stats);
	141
31e4c28d	142	void blkiocg_add_blkio_group(struct blkio_cgroup *blkcg,
22084190	143	struct blkio_group blkg, void key, dev_t dev)
31e4c28d VG	144	{
	145	unsigned long flags;
	146
	147	spin_lock_irqsave(&blkcg->lock, flags);
	148	rcu_assign_pointer(blkg->key, key);
b1c35769	149	blkg->blkcg_id = css_id(&blkcg->css);
31e4c28d VG	150	hlist_add_head_rcu(&blkg->blkcg_node, &blkcg->blkg_list);
31e4c28d VG	151	spin_unlock_irqrestore(&blkcg->lock, flags);
2868ef7b VG	152	#ifdef CONFIG_DEBUG_BLK_CGROUP
	153	/* Need to take css reference ? */
	154	cgroup_path(blkcg->css.cgroup, blkg->path, sizeof(blkg->path));
	155	#endif
22084190	156	blkg->dev = dev;
31e4c28d	157	}
9d6a986c	158	EXPORT_SYMBOL_GPL(blkiocg_add_blkio_group);
31e4c28d	159
b1c35769 VG	160	static void __blkiocg_del_blkio_group(struct blkio_group *blkg)
	161	{
	162	hlist_del_init_rcu(&blkg->blkcg_node);
	163	blkg->blkcg_id = 0;
	164	}
	165
	166	/*
	167	* returns 0 if blkio_group was still on cgroup list. Otherwise returns 1
	168	* indicating that blk_group was unhashed by the time we got to it.
	169	*/
31e4c28d VG	170	int blkiocg_del_blkio_group(struct blkio_group *blkg)
31e4c28d VG	171	{
b1c35769 VG	172	struct blkio_cgroup *blkcg;
	173	unsigned long flags;
	174	struct cgroup_subsys_state *css;
	175	int ret = 1;
	176
	177	rcu_read_lock();
	178	css = css_lookup(&blkio_subsys, blkg->blkcg_id);
	179	if (!css)
	180	goto out;
	181
	182	blkcg = container_of(css, struct blkio_cgroup, css);
	183	spin_lock_irqsave(&blkcg->lock, flags);
	184	if (!hlist_unhashed(&blkg->blkcg_node)) {
	185	__blkiocg_del_blkio_group(blkg);
	186	ret = 0;
	187	}
	188	spin_unlock_irqrestore(&blkcg->lock, flags);
	189	out:
	190	rcu_read_unlock();
	191	return ret;
31e4c28d	192	}
9d6a986c	193	EXPORT_SYMBOL_GPL(blkiocg_del_blkio_group);
31e4c28d VG	194
	195	/* called under rcu_read_lock(). */
	196	struct blkio_group blkiocg_lookup_group(struct blkio_cgroup blkcg, void *key)
	197	{
	198	struct blkio_group *blkg;
	199	struct hlist_node *n;
	200	void *__key;
	201
	202	hlist_for_each_entry_rcu(blkg, n, &blkcg->blkg_list, blkcg_node) {
	203	__key = blkg->key;
	204	if (__key == key)
	205	return blkg;
	206	}
	207
	208	return NULL;
	209	}
9d6a986c	210	EXPORT_SYMBOL_GPL(blkiocg_lookup_group);
31e4c28d VG	211
	212	#define SHOW_FUNCTION(__VAR) \
	213	static u64 blkiocg_##__VAR##_read(struct cgroup *cgroup, \
	214	struct cftype *cftype) \
	215	{ \
	216	struct blkio_cgroup *blkcg; \
	217	\
	218	blkcg = cgroup_to_blkio_cgroup(cgroup); \
	219	return (u64)blkcg->__VAR; \
	220	}
	221
	222	SHOW_FUNCTION(weight);
	223	#undef SHOW_FUNCTION
	224
	225	static int
	226	blkiocg_weight_write(struct cgroup cgroup, struct cftype cftype, u64 val)
	227	{
	228	struct blkio_cgroup *blkcg;
f8d461d6 VG	229	struct blkio_group *blkg;
f8d461d6 VG	230	struct hlist_node *n;
3e252066	231	struct blkio_policy_type *blkiop;
31e4c28d VG	232
	233	if (val < BLKIO_WEIGHT_MIN \|\| val > BLKIO_WEIGHT_MAX)
	234	return -EINVAL;
	235
	236	blkcg = cgroup_to_blkio_cgroup(cgroup);
bcf4dd43	237	spin_lock(&blkio_list_lock);
f8d461d6	238	spin_lock_irq(&blkcg->lock);
31e4c28d	239	blkcg->weight = (unsigned int)val;
3e252066	240	hlist_for_each_entry(blkg, n, &blkcg->blkg_list, blkcg_node) {
3e252066 VG	241	list_for_each_entry(blkiop, &blkio_list, list)
	242	blkiop->ops.blkio_update_group_weight_fn(blkg,
	243	blkcg->weight);
3e252066	244	}
f8d461d6	245	spin_unlock_irq(&blkcg->lock);
bcf4dd43	246	spin_unlock(&blkio_list_lock);
31e4c28d VG	247	return 0;
	248	}
	249
303a3acb	250	static int
84c124da	251	blkiocg_reset_stats(struct cgroup cgroup, struct cftype cftype, u64 val)
303a3acb DS	252	{
	253	struct blkio_cgroup *blkcg;
	254	struct blkio_group *blkg;
	255	struct hlist_node *n;
	256	struct blkio_group_stats *stats;
	257
	258	blkcg = cgroup_to_blkio_cgroup(cgroup);
	259	spin_lock_irq(&blkcg->lock);
	260	hlist_for_each_entry(blkg, n, &blkcg->blkg_list, blkcg_node) {
	261	spin_lock(&blkg->stats_lock);
	262	stats = &blkg->stats;
	263	memset(stats, 0, sizeof(struct blkio_group_stats));
	264	spin_unlock(&blkg->stats_lock);
	265	}
	266	spin_unlock_irq(&blkcg->lock);
	267	return 0;
	268	}
	269
84c124da DS	270	static void blkio_get_key_name(enum stat_sub_type type, dev_t dev, char *str,
84c124da DS	271	int chars_left, bool diskname_only)
303a3acb	272	{
84c124da	273	snprintf(str, chars_left, "%d:%d", MAJOR(dev), MINOR(dev));
303a3acb DS	274	chars_left -= strlen(str);
	275	if (chars_left <= 0) {
	276	printk(KERN_WARNING
	277	"Possibly incorrect cgroup stat display format");
	278	return;
	279	}
84c124da DS	280	if (diskname_only)
84c124da DS	281	return;
303a3acb	282	switch (type) {
84c124da	283	case BLKIO_STAT_READ:
303a3acb DS	284	strlcat(str, " Read", chars_left);
303a3acb DS	285	break;
84c124da	286	case BLKIO_STAT_WRITE:
303a3acb DS	287	strlcat(str, " Write", chars_left);
303a3acb DS	288	break;
84c124da	289	case BLKIO_STAT_SYNC:
303a3acb DS	290	strlcat(str, " Sync", chars_left);
303a3acb DS	291	break;
84c124da	292	case BLKIO_STAT_ASYNC:
303a3acb DS	293	strlcat(str, " Async", chars_left);
303a3acb DS	294	break;
84c124da	295	case BLKIO_STAT_TOTAL:
303a3acb DS	296	strlcat(str, " Total", chars_left);
	297	break;
	298	default:
	299	strlcat(str, " Invalid", chars_left);
	300	}
	301	}
	302
84c124da DS	303	static uint64_t blkio_fill_stat(char *str, int chars_left, uint64_t val,
	304	struct cgroup_map_cb *cb, dev_t dev)
	305	{
	306	blkio_get_key_name(0, dev, str, chars_left, true);
	307	cb->fill(cb, str, val);
	308	return val;
	309	}
303a3acb	310
84c124da DS	311	/* This should be called with blkg->stats_lock held */
	312	static uint64_t blkio_get_stat(struct blkio_group *blkg,
	313	struct cgroup_map_cb *cb, dev_t dev, enum stat_type type)
303a3acb DS	314	{
	315	uint64_t disk_total;
	316	char key_str[MAX_KEY_LEN];
84c124da DS	317	enum stat_sub_type sub_type;
	318
	319	if (type == BLKIO_STAT_TIME)
	320	return blkio_fill_stat(key_str, MAX_KEY_LEN - 1,
	321	blkg->stats.time, cb, dev);
	322	if (type == BLKIO_STAT_SECTORS)
	323	return blkio_fill_stat(key_str, MAX_KEY_LEN - 1,
	324	blkg->stats.sectors, cb, dev);
	325	#ifdef CONFIG_DEBUG_BLK_CGROUP
	326	if (type == BLKIO_STAT_DEQUEUE)
	327	return blkio_fill_stat(key_str, MAX_KEY_LEN - 1,
	328	blkg->stats.dequeue, cb, dev);
	329	#endif
303a3acb	330
84c124da DS	331	for (sub_type = BLKIO_STAT_READ; sub_type < BLKIO_STAT_TOTAL;
	332	sub_type++) {
	333	blkio_get_key_name(sub_type, dev, key_str, MAX_KEY_LEN, false);
	334	cb->fill(cb, key_str, blkg->stats.stat_arr[type][sub_type]);
303a3acb	335	}
84c124da DS	336	disk_total = blkg->stats.stat_arr[type][BLKIO_STAT_READ] +
	337	blkg->stats.stat_arr[type][BLKIO_STAT_WRITE];
	338	blkio_get_key_name(BLKIO_STAT_TOTAL, dev, key_str, MAX_KEY_LEN, false);
303a3acb DS	339	cb->fill(cb, key_str, disk_total);
	340	return disk_total;
	341	}
	342
84c124da	343	#define SHOW_FUNCTION_PER_GROUP(__VAR, type, show_total) \
22084190	344	static int blkiocg_##__VAR##_read(struct cgroup *cgroup, \
303a3acb	345	struct cftype cftype, struct cgroup_map_cb cb) \
22084190 VG	346	{ \
	347	struct blkio_cgroup *blkcg; \
	348	struct blkio_group *blkg; \
	349	struct hlist_node *n; \
303a3acb	350	uint64_t cgroup_total = 0; \
22084190 VG	351	\
	352	if (!cgroup_lock_live_group(cgroup)) \
	353	return -ENODEV; \
	354	\
	355	blkcg = cgroup_to_blkio_cgroup(cgroup); \
	356	rcu_read_lock(); \
	357	hlist_for_each_entry_rcu(blkg, n, &blkcg->blkg_list, blkcg_node) {\
303a3acb DS	358	if (blkg->dev) { \
303a3acb DS	359	spin_lock_irq(&blkg->stats_lock); \
84c124da DS	360	cgroup_total += blkio_get_stat(blkg, cb, \
84c124da DS	361	blkg->dev, type); \
303a3acb DS	362	spin_unlock_irq(&blkg->stats_lock); \
303a3acb DS	363	} \
22084190	364	} \
303a3acb DS	365	if (show_total) \
303a3acb DS	366	cb->fill(cb, "Total", cgroup_total); \
22084190 VG	367	rcu_read_unlock(); \
	368	cgroup_unlock(); \
	369	return 0; \
	370	}
	371
84c124da DS	372	SHOW_FUNCTION_PER_GROUP(time, BLKIO_STAT_TIME, 0);
	373	SHOW_FUNCTION_PER_GROUP(sectors, BLKIO_STAT_SECTORS, 0);
	374	SHOW_FUNCTION_PER_GROUP(io_service_bytes, BLKIO_STAT_SERVICE_BYTES, 1);
	375	SHOW_FUNCTION_PER_GROUP(io_serviced, BLKIO_STAT_SERVICED, 1);
	376	SHOW_FUNCTION_PER_GROUP(io_service_time, BLKIO_STAT_SERVICE_TIME, 1);
	377	SHOW_FUNCTION_PER_GROUP(io_wait_time, BLKIO_STAT_WAIT_TIME, 1);
812d4026	378	SHOW_FUNCTION_PER_GROUP(io_merged, BLKIO_STAT_MERGED, 1);
22084190	379	#ifdef CONFIG_DEBUG_BLK_CGROUP
84c124da	380	SHOW_FUNCTION_PER_GROUP(dequeue, BLKIO_STAT_DEQUEUE, 0);
22084190 VG	381	#endif
	382	#undef SHOW_FUNCTION_PER_GROUP
	383
	384	#ifdef CONFIG_DEBUG_BLK_CGROUP
9195291e	385	void blkiocg_update_dequeue_stats(struct blkio_group *blkg,
22084190 VG	386	unsigned long dequeue)
22084190 VG	387	{
303a3acb	388	blkg->stats.dequeue += dequeue;
22084190	389	}
9195291e	390	EXPORT_SYMBOL_GPL(blkiocg_update_dequeue_stats);
22084190 VG	391	#endif
22084190 VG	392
31e4c28d VG	393	struct cftype blkio_files[] = {
	394	{
	395	.name = "weight",
	396	.read_u64 = blkiocg_weight_read,
	397	.write_u64 = blkiocg_weight_write,
	398	},
22084190 VG	399	{
22084190 VG	400	.name = "time",
303a3acb	401	.read_map = blkiocg_time_read,
22084190 VG	402	},
	403	{
	404	.name = "sectors",
303a3acb	405	.read_map = blkiocg_sectors_read,
303a3acb DS	406	},
	407	{
	408	.name = "io_service_bytes",
	409	.read_map = blkiocg_io_service_bytes_read,
303a3acb DS	410	},
	411	{
	412	.name = "io_serviced",
	413	.read_map = blkiocg_io_serviced_read,
303a3acb DS	414	},
	415	{
	416	.name = "io_service_time",
	417	.read_map = blkiocg_io_service_time_read,
303a3acb DS	418	},
	419	{
	420	.name = "io_wait_time",
	421	.read_map = blkiocg_io_wait_time_read,
84c124da	422	},
812d4026 DS	423	{
	424	.name = "io_merged",
	425	.read_map = blkiocg_io_merged_read,
	426	},
84c124da DS	427	{
	428	.name = "reset_stats",
	429	.write_u64 = blkiocg_reset_stats,
22084190 VG	430	},
	431	#ifdef CONFIG_DEBUG_BLK_CGROUP
	432	{
	433	.name = "dequeue",
303a3acb	434	.read_map = blkiocg_dequeue_read,
22084190 VG	435	},
22084190 VG	436	#endif
31e4c28d VG	437	};
	438
	439	static int blkiocg_populate(struct cgroup_subsys subsys, struct cgroup cgroup)
	440	{
	441	return cgroup_add_files(cgroup, subsys, blkio_files,
	442	ARRAY_SIZE(blkio_files));
	443	}
	444
	445	static void blkiocg_destroy(struct cgroup_subsys subsys, struct cgroup cgroup)
	446	{
	447	struct blkio_cgroup *blkcg = cgroup_to_blkio_cgroup(cgroup);
b1c35769 VG	448	unsigned long flags;
	449	struct blkio_group *blkg;
	450	void *key;
3e252066	451	struct blkio_policy_type *blkiop;
b1c35769 VG	452
	453	rcu_read_lock();
	454	remove_entry:
	455	spin_lock_irqsave(&blkcg->lock, flags);
	456
	457	if (hlist_empty(&blkcg->blkg_list)) {
	458	spin_unlock_irqrestore(&blkcg->lock, flags);
	459	goto done;
	460	}
	461
	462	blkg = hlist_entry(blkcg->blkg_list.first, struct blkio_group,
	463	blkcg_node);
	464	key = rcu_dereference(blkg->key);
	465	__blkiocg_del_blkio_group(blkg);
31e4c28d	466
b1c35769 VG	467	spin_unlock_irqrestore(&blkcg->lock, flags);
	468
	469	/*
	470	* This blkio_group is being unlinked as associated cgroup is going
	471	* away. Let all the IO controlling policies know about this event.
	472	*
	473	* Currently this is static call to one io controlling policy. Once
	474	* we have more policies in place, we need some dynamic registration
	475	* of callback function.
	476	*/
3e252066 VG	477	spin_lock(&blkio_list_lock);
	478	list_for_each_entry(blkiop, &blkio_list, list)
	479	blkiop->ops.blkio_unlink_group_fn(key, blkg);
	480	spin_unlock(&blkio_list_lock);
b1c35769 VG	481	goto remove_entry;
b1c35769 VG	482	done:
31e4c28d	483	free_css_id(&blkio_subsys, &blkcg->css);
b1c35769	484	rcu_read_unlock();
67523c48 BB	485	if (blkcg != &blkio_root_cgroup)
67523c48 BB	486	kfree(blkcg);
31e4c28d VG	487	}
	488
	489	static struct cgroup_subsys_state *
	490	blkiocg_create(struct cgroup_subsys subsys, struct cgroup cgroup)
	491	{
	492	struct blkio_cgroup blkcg, parent_blkcg;
	493
	494	if (!cgroup->parent) {
	495	blkcg = &blkio_root_cgroup;
	496	goto done;
	497	}
	498
	499	/* Currently we do not support hierarchy deeper than two level (0,1) */
	500	parent_blkcg = cgroup_to_blkio_cgroup(cgroup->parent);
	501	if (css_depth(&parent_blkcg->css) > 0)
	502	return ERR_PTR(-EINVAL);
	503
	504	blkcg = kzalloc(sizeof(*blkcg), GFP_KERNEL);
	505	if (!blkcg)
	506	return ERR_PTR(-ENOMEM);
	507
	508	blkcg->weight = BLKIO_WEIGHT_DEFAULT;
	509	done:
	510	spin_lock_init(&blkcg->lock);
	511	INIT_HLIST_HEAD(&blkcg->blkg_list);
	512
	513	return &blkcg->css;
	514	}
	515
	516	/*
	517	* We cannot support shared io contexts, as we have no mean to support
	518	* two tasks with the same ioc in two different groups without major rework
	519	* of the main cic data structures. For now we allow a task to change
	520	* its cgroup only if it's the only owner of its ioc.
	521	*/
	522	static int blkiocg_can_attach(struct cgroup_subsys *subsys,
	523	struct cgroup cgroup, struct task_struct tsk,
	524	bool threadgroup)
	525	{
	526	struct io_context *ioc;
	527	int ret = 0;
	528
	529	/* task_lock() is needed to avoid races with exit_io_context() */
	530	task_lock(tsk);
	531	ioc = tsk->io_context;
	532	if (ioc && atomic_read(&ioc->nr_tasks) > 1)
	533	ret = -EINVAL;
	534	task_unlock(tsk);
	535
	536	return ret;
	537	}
	538
	539	static void blkiocg_attach(struct cgroup_subsys subsys, struct cgroup cgroup,
	540	struct cgroup prev, struct task_struct tsk,
	541	bool threadgroup)
	542	{
	543	struct io_context *ioc;
	544
	545	task_lock(tsk);
	546	ioc = tsk->io_context;
	547	if (ioc)
	548	ioc->cgroup_changed = 1;
	549	task_unlock(tsk);
	550	}
551
3e252066 VG	552	void blkio_policy_register(struct blkio_policy_type *blkiop)
	553	{
	554	spin_lock(&blkio_list_lock);
	555	list_add_tail(&blkiop->list, &blkio_list);
	556	spin_unlock(&blkio_list_lock);
	557	}
	558	EXPORT_SYMBOL_GPL(blkio_policy_register);
	559
	560	void blkio_policy_unregister(struct blkio_policy_type *blkiop)
	561	{
	562	spin_lock(&blkio_list_lock);
	563	list_del_init(&blkiop->list);
	564	spin_unlock(&blkio_list_lock);
	565	}
	566	EXPORT_SYMBOL_GPL(blkio_policy_unregister);
67523c48 BB	567
	568	static int __init init_cgroup_blkio(void)
	569	{
	570	return cgroup_load_subsys(&blkio_subsys);
	571	}
	572
	573	static void __exit exit_cgroup_blkio(void)
	574	{
	575	cgroup_unload_subsys(&blkio_subsys);
	576	}
	577
	578	module_init(init_cgroup_blkio);
	579	module_exit(exit_cgroup_blkio);
	580	MODULE_LICENSE("GPL");