[linux-block.git] / block / blk-cgroup.h

/* SPDX-License-Identifier: GPL-2.0 */
#ifndef _BLK_CGROUP_PRIVATE_H
#define _BLK_CGROUP_PRIVATE_H
/*
 * block cgroup private header
 *
 * 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/blk-cgroup.h>
#include <linux/cgroup.h>
#include <linux/kthread.h>
#include <linux/blk-mq.h>

struct blkcg_gq;
struct blkg_policy_data;


/* percpu_counter batch for blkg_[rw]stats, per-cpu drift doesn't matter */
#define BLKG_STAT_CPU_BATCH	(INT_MAX / 2)

#ifdef CONFIG_BLK_CGROUP

enum blkg_iostat_type {
	BLKG_IOSTAT_READ,
	BLKG_IOSTAT_WRITE,
	BLKG_IOSTAT_DISCARD,

	BLKG_IOSTAT_NR,
};

struct blkg_iostat {
	u64				bytes[BLKG_IOSTAT_NR];
	u64				ios[BLKG_IOSTAT_NR];
};

struct blkg_iostat_set {
	struct u64_stats_sync		sync;
	struct blkg_iostat		cur;
	struct blkg_iostat		last;
};

/* association between a blk cgroup and a request queue */
struct blkcg_gq {
	/* Pointer to the associated request_queue */
	struct request_queue		*q;
	struct list_head		q_node;
	struct hlist_node		blkcg_node;
	struct blkcg			*blkcg;

	/* all non-root blkcg_gq's are guaranteed to have access to parent */
	struct blkcg_gq			*parent;

	/* reference count */
	struct percpu_ref		refcnt;

	/* is this blkg online? protected by both blkcg and q locks */
	bool				online;

	struct blkg_iostat_set __percpu	*iostat_cpu;
	struct blkg_iostat_set		iostat;

	struct blkg_policy_data		*pd[BLKCG_MAX_POLS];

	spinlock_t			async_bio_lock;
	struct bio_list			async_bios;
	union {
		struct work_struct	async_bio_work;
		struct work_struct	free_work;
	};

	atomic_t			use_delay;
	atomic64_t			delay_nsec;
	atomic64_t			delay_start;
	u64				last_delay;
	int				last_use;

	struct rcu_head			rcu_head;
};

struct blkcg {
	struct cgroup_subsys_state	css;
	spinlock_t			lock;
	refcount_t			online_pin;

	struct radix_tree_root		blkg_tree;
	struct blkcg_gq	__rcu		*blkg_hint;
	struct hlist_head		blkg_list;

	struct blkcg_policy_data	*cpd[BLKCG_MAX_POLS];

	struct list_head		all_blkcgs_node;
#ifdef CONFIG_BLK_CGROUP_FC_APPID
	char                            fc_app_id[FC_APPID_LEN];
#endif
#ifdef CONFIG_CGROUP_WRITEBACK
	struct list_head		cgwb_list;
#endif
};

static inline struct blkcg *css_to_blkcg(struct cgroup_subsys_state *css)
{
	return css ? container_of(css, struct blkcg, css) : NULL;
}

/*
 * A blkcg_gq (blkg) is association between a block cgroup (blkcg) and a
 * request_queue (q).  This is used by blkcg policies which need to track
 * information per blkcg - q pair.
 *
 * There can be multiple active blkcg policies and each blkg:policy pair is
 * represented by a blkg_policy_data which is allocated and freed by each
 * policy's pd_alloc/free_fn() methods.  A policy can allocate private data
 * area by allocating larger data structure which embeds blkg_policy_data
 * at the beginning.
 */
struct blkg_policy_data {
	/* the blkg and policy id this per-policy data belongs to */
	struct blkcg_gq			*blkg;
	int				plid;
};

/*
 * Policies that need to keep per-blkcg data which is independent from any
 * request_queue associated to it should implement cpd_alloc/free_fn()
 * methods.  A policy can allocate private data area by allocating larger
 * data structure which embeds blkcg_policy_data at the beginning.
 * cpd_init() is invoked to let each policy handle per-blkcg data.
 */
struct blkcg_policy_data {
	/* the blkcg and policy id this per-policy data belongs to */
	struct blkcg			*blkcg;
	int				plid;
};

typedef struct blkcg_policy_data *(blkcg_pol_alloc_cpd_fn)(gfp_t gfp);
typedef void (blkcg_pol_init_cpd_fn)(struct blkcg_policy_data *cpd);
typedef void (blkcg_pol_free_cpd_fn)(struct blkcg_policy_data *cpd);
typedef void (blkcg_pol_bind_cpd_fn)(struct blkcg_policy_data *cpd);
typedef struct blkg_policy_data *(blkcg_pol_alloc_pd_fn)(gfp_t gfp,
				struct request_queue *q, struct blkcg *blkcg);
typedef void (blkcg_pol_init_pd_fn)(struct blkg_policy_data *pd);
typedef void (blkcg_pol_online_pd_fn)(struct blkg_policy_data *pd);
typedef void (blkcg_pol_offline_pd_fn)(struct blkg_policy_data *pd);
typedef void (blkcg_pol_free_pd_fn)(struct blkg_policy_data *pd);
typedef void (blkcg_pol_reset_pd_stats_fn)(struct blkg_policy_data *pd);
typedef void (blkcg_pol_stat_pd_fn)(struct blkg_policy_data *pd,
				struct seq_file *s);

struct blkcg_policy {
	int				plid;
	/* cgroup files for the policy */
	struct cftype			*dfl_cftypes;
	struct cftype			*legacy_cftypes;

	/* operations */
	blkcg_pol_alloc_cpd_fn		*cpd_alloc_fn;
	blkcg_pol_init_cpd_fn		*cpd_init_fn;
	blkcg_pol_free_cpd_fn		*cpd_free_fn;
	blkcg_pol_bind_cpd_fn		*cpd_bind_fn;

	blkcg_pol_alloc_pd_fn		*pd_alloc_fn;
	blkcg_pol_init_pd_fn		*pd_init_fn;
	blkcg_pol_online_pd_fn		*pd_online_fn;
	blkcg_pol_offline_pd_fn		*pd_offline_fn;
	blkcg_pol_free_pd_fn		*pd_free_fn;
	blkcg_pol_reset_pd_stats_fn	*pd_reset_stats_fn;
	blkcg_pol_stat_pd_fn		*pd_stat_fn;
};

extern struct blkcg blkcg_root;
extern bool blkcg_debug_stats;

int blkcg_init_disk(struct gendisk *disk);
void blkcg_exit_disk(struct gendisk *disk);

/* Blkio controller policy registration */
int blkcg_policy_register(struct blkcg_policy *pol);
void blkcg_policy_unregister(struct blkcg_policy *pol);
int blkcg_activate_policy(struct request_queue *q,
			  const struct blkcg_policy *pol);
void blkcg_deactivate_policy(struct request_queue *q,
			     const struct blkcg_policy *pol);

const char *blkg_dev_name(struct blkcg_gq *blkg);
void blkcg_print_blkgs(struct seq_file *sf, struct blkcg *blkcg,
		       u64 (*prfill)(struct seq_file *,
				     struct blkg_policy_data *, int),
		       const struct blkcg_policy *pol, int data,
		       bool show_total);
u64 __blkg_prfill_u64(struct seq_file *sf, struct blkg_policy_data *pd, u64 v);

struct blkg_conf_ctx {
	struct block_device		*bdev;
	struct blkcg_gq			*blkg;
	char				*body;
};

struct block_device *blkcg_conf_open_bdev(char **inputp);
int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
		   char *input, struct blkg_conf_ctx *ctx);
void blkg_conf_finish(struct blkg_conf_ctx *ctx);

/**
 * bio_issue_as_root_blkg - see if this bio needs to be issued as root blkg
 * @return: true if this bio needs to be submitted with the root blkg context.
 *
 * In order to avoid priority inversions we sometimes need to issue a bio as if
 * it were attached to the root blkg, and then backcharge to the actual owning
 * blkg.  The idea is we do bio_blkcg_css() to look up the actual context for
 * the bio and attach the appropriate blkg to the bio.  Then we call this helper
 * and if it is true run with the root blkg for that queue and then do any
 * backcharging to the originating cgroup once the io is complete.
 */
static inline bool bio_issue_as_root_blkg(struct bio *bio)
{
	return (bio->bi_opf & (REQ_META | REQ_SWAP)) != 0;
}

/**
 * blkg_lookup - lookup blkg for the specified blkcg - q pair
 * @blkcg: blkcg of interest
 * @q: request_queue of interest
 *
 * Lookup blkg for the @blkcg - @q pair.

 * Must be called in a RCU critical section.
 */
static inline struct blkcg_gq *blkg_lookup(struct blkcg *blkcg,
					   struct request_queue *q)
{
	struct blkcg_gq *blkg;

	WARN_ON_ONCE(!rcu_read_lock_held());

	if (blkcg == &blkcg_root)
		return q->root_blkg;

	blkg = rcu_dereference(blkcg->blkg_hint);
	if (blkg && blkg->q == q)
		return blkg;

	blkg = radix_tree_lookup(&blkcg->blkg_tree, q->id);
	if (blkg && blkg->q != q)
		blkg = NULL;
	return blkg;
}

/**
 * blkg_to_pdata - get policy private data
 * @blkg: blkg of interest
 * @pol: policy of interest
 *
 * Return pointer to private data associated with the @blkg-@pol pair.
 */
static inline struct blkg_policy_data *blkg_to_pd(struct blkcg_gq *blkg,
						  struct blkcg_policy *pol)
{
	return blkg ? blkg->pd[pol->plid] : NULL;
}

static inline struct blkcg_policy_data *blkcg_to_cpd(struct blkcg *blkcg,
						     struct blkcg_policy *pol)
{
	return blkcg ? blkcg->cpd[pol->plid] : NULL;
}

/**
 * pdata_to_blkg - get blkg associated with policy private data
 * @pd: policy private data of interest
 *
 * @pd is policy private data.  Determine the blkg it's associated with.
 */
static inline struct blkcg_gq *pd_to_blkg(struct blkg_policy_data *pd)
{
	return pd ? pd->blkg : NULL;
}

static inline struct blkcg *cpd_to_blkcg(struct blkcg_policy_data *cpd)
{
	return cpd ? cpd->blkcg : NULL;
}

/**
 * blkg_path - format cgroup path of blkg
 * @blkg: blkg of interest
 * @buf: target buffer
 * @buflen: target buffer length
 *
 * Format the path of the cgroup of @blkg into @buf.
 */
static inline int blkg_path(struct blkcg_gq *blkg, char *buf, int buflen)
{
	return cgroup_path(blkg->blkcg->css.cgroup, buf, buflen);
}

/**
 * blkg_get - get a blkg reference
 * @blkg: blkg to get
 *
 * The caller should be holding an existing reference.
 */
static inline void blkg_get(struct blkcg_gq *blkg)
{
	percpu_ref_get(&blkg->refcnt);
}

/**
 * blkg_tryget - try and get a blkg reference
 * @blkg: blkg to get
 *
 * This is for use when doing an RCU lookup of the blkg.  We may be in the midst
 * of freeing this blkg, so we can only use it if the refcnt is not zero.
 */
static inline bool blkg_tryget(struct blkcg_gq *blkg)
{
	return blkg && percpu_ref_tryget(&blkg->refcnt);
}

/**
 * blkg_put - put a blkg reference
 * @blkg: blkg to put
 */
static inline void blkg_put(struct blkcg_gq *blkg)
{
	percpu_ref_put(&blkg->refcnt);
}

/**
 * blkg_for_each_descendant_pre - pre-order walk of a blkg's descendants
 * @d_blkg: loop cursor pointing to the current descendant
 * @pos_css: used for iteration
 * @p_blkg: target blkg to walk descendants of
 *
 * Walk @c_blkg through the descendants of @p_blkg.  Must be used with RCU
 * read locked.  If called under either blkcg or queue lock, the iteration
 * is guaranteed to include all and only online blkgs.  The caller may
 * update @pos_css by calling css_rightmost_descendant() to skip subtree.
 * @p_blkg is included in the iteration and the first node to be visited.
 */
#define blkg_for_each_descendant_pre(d_blkg, pos_css, p_blkg)		\
	css_for_each_descendant_pre((pos_css), &(p_blkg)->blkcg->css)	\
		if (((d_blkg) = blkg_lookup(css_to_blkcg(pos_css),	\
					    (p_blkg)->q)))

/**
 * blkg_for_each_descendant_post - post-order walk of a blkg's descendants
 * @d_blkg: loop cursor pointing to the current descendant
 * @pos_css: used for iteration
 * @p_blkg: target blkg to walk descendants of
 *
 * Similar to blkg_for_each_descendant_pre() but performs post-order
 * traversal instead.  Synchronization rules are the same.  @p_blkg is
 * included in the iteration and the last node to be visited.
 */
#define blkg_for_each_descendant_post(d_blkg, pos_css, p_blkg)		\
	css_for_each_descendant_post((pos_css), &(p_blkg)->blkcg->css)	\
		if (((d_blkg) = blkg_lookup(css_to_blkcg(pos_css),	\
					    (p_blkg)->q)))

bool __blkcg_punt_bio_submit(struct bio *bio);

static inline bool blkcg_punt_bio_submit(struct bio *bio)
{
	if (bio->bi_opf & REQ_CGROUP_PUNT)
		return __blkcg_punt_bio_submit(bio);
	else
		return false;
}

static inline void blkcg_bio_issue_init(struct bio *bio)
{
	bio_issue_init(&bio->bi_issue, bio_sectors(bio));
}

static inline void blkcg_use_delay(struct blkcg_gq *blkg)
{
	if (WARN_ON_ONCE(atomic_read(&blkg->use_delay) < 0))
		return;
	if (atomic_add_return(1, &blkg->use_delay) == 1)
		atomic_inc(&blkg->blkcg->css.cgroup->congestion_count);
}

static inline int blkcg_unuse_delay(struct blkcg_gq *blkg)
{
	int old = atomic_read(&blkg->use_delay);

	if (WARN_ON_ONCE(old < 0))
		return 0;
	if (old == 0)
		return 0;

	/*
	 * We do this song and dance because we can race with somebody else
	 * adding or removing delay.  If we just did an atomic_dec we'd end up
	 * negative and we'd already be in trouble.  We need to subtract 1 and
	 * then check to see if we were the last delay so we can drop the
	 * congestion count on the cgroup.
	 */
	while (old && !atomic_try_cmpxchg(&blkg->use_delay, &old, old - 1))
		;

	if (old == 0)
		return 0;
	if (old == 1)
		atomic_dec(&blkg->blkcg->css.cgroup->congestion_count);
	return 1;
}

/**
 * blkcg_set_delay - Enable allocator delay mechanism with the specified delay amount
 * @blkg: target blkg
 * @delay: delay duration in nsecs
 *
 * When enabled with this function, the delay is not decayed and must be
 * explicitly cleared with blkcg_clear_delay(). Must not be mixed with
 * blkcg_[un]use_delay() and blkcg_add_delay() usages.
 */
static inline void blkcg_set_delay(struct blkcg_gq *blkg, u64 delay)
{
	int old = atomic_read(&blkg->use_delay);

	/* We only want 1 person setting the congestion count for this blkg. */
	if (!old && atomic_try_cmpxchg(&blkg->use_delay, &old, -1))
		atomic_inc(&blkg->blkcg->css.cgroup->congestion_count);

	atomic64_set(&blkg->delay_nsec, delay);
}

/**
 * blkcg_clear_delay - Disable allocator delay mechanism
 * @blkg: target blkg
 *
 * Disable use_delay mechanism. See blkcg_set_delay().
 */
static inline void blkcg_clear_delay(struct blkcg_gq *blkg)
{
	int old = atomic_read(&blkg->use_delay);

	/* We only want 1 person clearing the congestion count for this blkg. */
	if (old && atomic_try_cmpxchg(&blkg->use_delay, &old, 0))
		atomic_dec(&blkg->blkcg->css.cgroup->congestion_count);
}

/**
 * blk_cgroup_mergeable - Determine whether to allow or disallow merges
 * @rq: request to merge into
 * @bio: bio to merge
 *
 * @bio and @rq should belong to the same cgroup and their issue_as_root should
 * match. The latter is necessary as we don't want to throttle e.g. a metadata
 * update because it happens to be next to a regular IO.
 */
static inline bool blk_cgroup_mergeable(struct request *rq, struct bio *bio)
{
	return rq->bio->bi_blkg == bio->bi_blkg &&
		bio_issue_as_root_blkg(rq->bio) == bio_issue_as_root_blkg(bio);
}

void blk_cgroup_bio_start(struct bio *bio);
void blkcg_add_delay(struct blkcg_gq *blkg, u64 now, u64 delta);
#else	/* CONFIG_BLK_CGROUP */

struct blkg_policy_data {
};

struct blkcg_policy_data {
};

struct blkcg_policy {
};

struct blkcg {
};

static inline struct blkcg_gq *blkg_lookup(struct blkcg *blkcg, void *key) { return NULL; }
static inline int blkcg_init_disk(struct gendisk *disk) { return 0; }
static inline void blkcg_exit_disk(struct gendisk *disk) { }
static inline int blkcg_policy_register(struct blkcg_policy *pol) { return 0; }
static inline void blkcg_policy_unregister(struct blkcg_policy *pol) { }
static inline int blkcg_activate_policy(struct request_queue *q,
					const struct blkcg_policy *pol) { return 0; }
static inline void blkcg_deactivate_policy(struct request_queue *q,
					   const struct blkcg_policy *pol) { }

static inline struct blkg_policy_data *blkg_to_pd(struct blkcg_gq *blkg,
						  struct blkcg_policy *pol) { return NULL; }
static inline struct blkcg_gq *pd_to_blkg(struct blkg_policy_data *pd) { return NULL; }
static inline char *blkg_path(struct blkcg_gq *blkg) { return NULL; }
static inline void blkg_get(struct blkcg_gq *blkg) { }
static inline void blkg_put(struct blkcg_gq *blkg) { }

static inline bool blkcg_punt_bio_submit(struct bio *bio) { return false; }
static inline void blkcg_bio_issue_init(struct bio *bio) { }
static inline void blk_cgroup_bio_start(struct bio *bio) { }
static inline bool blk_cgroup_mergeable(struct request *rq, struct bio *bio) { return true; }

#define blk_queue_for_each_rl(rl, q)	\
	for ((rl) = &(q)->root_rl; (rl); (rl) = NULL)

#endif	/* CONFIG_BLK_CGROUP */

#endif /* _BLK_CGROUP_PRIVATE_H */
Commit	Line	Data
672fdcf0 ML	1	/* SPDX-License-Identifier: GPL-2.0 */
	2	#ifndef _BLK_CGROUP_PRIVATE_H
	3	#define _BLK_CGROUP_PRIVATE_H
	4	/*
	5	* block cgroup private header
	6	*
	7	* Based on ideas and code from CFQ, CFS and BFQ:
	8	* Copyright (C) 2003 Jens Axboe <axboe@kernel.dk>
	9	*
	10	* Copyright (C) 2008 Fabio Checconi <fabio@gandalf.sssup.it>
	11	* Paolo Valente <paolo.valente@unimore.it>
	12	*
	13	* Copyright (C) 2009 Vivek Goyal <vgoyal@redhat.com>
	14	* Nauman Rafique <nauman@google.com>
	15	*/
	16
	17	#include <linux/blk-cgroup.h>
c97ab271 CH	18	#include <linux/cgroup.h>
c97ab271 CH	19	#include <linux/kthread.h>
6b2b0459	20	#include <linux/blk-mq.h>
672fdcf0	21
dec223c9 CH	22	struct blkcg_gq;
	23	struct blkg_policy_data;
	24
	25
672fdcf0 ML	26	/* percpu_counter batch for blkg_[rw]stats, per-cpu drift doesn't matter */
	27	#define BLKG_STAT_CPU_BATCH (INT_MAX / 2)
	28
	29	#ifdef CONFIG_BLK_CGROUP
bbb1ebe7 CH	30
	31	enum blkg_iostat_type {
	32	BLKG_IOSTAT_READ,
	33	BLKG_IOSTAT_WRITE,
	34	BLKG_IOSTAT_DISCARD,
	35
	36	BLKG_IOSTAT_NR,
	37	};
	38
	39	struct blkg_iostat {
	40	u64 bytes[BLKG_IOSTAT_NR];
	41	u64 ios[BLKG_IOSTAT_NR];
	42	};
	43
	44	struct blkg_iostat_set {
	45	struct u64_stats_sync sync;
	46	struct blkg_iostat cur;
	47	struct blkg_iostat last;
	48	};
	49
	50	/* association between a blk cgroup and a request queue */
	51	struct blkcg_gq {
	52	/* Pointer to the associated request_queue */
	53	struct request_queue *q;
	54	struct list_head q_node;
	55	struct hlist_node blkcg_node;
	56	struct blkcg *blkcg;
	57
	58	/* all non-root blkcg_gq's are guaranteed to have access to parent */
	59	struct blkcg_gq *parent;
	60
	61	/* reference count */
	62	struct percpu_ref refcnt;
	63
	64	/* is this blkg online? protected by both blkcg and q locks */
	65	bool online;
	66
	67	struct blkg_iostat_set __percpu *iostat_cpu;
	68	struct blkg_iostat_set iostat;
	69
	70	struct blkg_policy_data *pd[BLKCG_MAX_POLS];
	71
	72	spinlock_t async_bio_lock;
	73	struct bio_list async_bios;
	74	union {
	75	struct work_struct async_bio_work;
	76	struct work_struct free_work;
	77	};
	78
	79	atomic_t use_delay;
	80	atomic64_t delay_nsec;
	81	atomic64_t delay_start;
	82	u64 last_delay;
	83	int last_use;
	84
	85	struct rcu_head rcu_head;
	86	};
	87
dec223c9 CH	88	struct blkcg {
	89	struct cgroup_subsys_state css;
	90	spinlock_t lock;
	91	refcount_t online_pin;
	92
	93	struct radix_tree_root blkg_tree;
	94	struct blkcg_gq __rcu *blkg_hint;
	95	struct hlist_head blkg_list;
	96
	97	struct blkcg_policy_data *cpd[BLKCG_MAX_POLS];
	98
	99	struct list_head all_blkcgs_node;
	100	#ifdef CONFIG_BLK_CGROUP_FC_APPID
	101	char fc_app_id[FC_APPID_LEN];
	102	#endif
	103	#ifdef CONFIG_CGROUP_WRITEBACK
	104	struct list_head cgwb_list;
	105	#endif
	106	};
	107
	108	static inline struct blkcg css_to_blkcg(struct cgroup_subsys_state css)
	109	{
	110	return css ? container_of(css, struct blkcg, css) : NULL;
	111	}
672fdcf0 ML	112
	113	/*
	114	* A blkcg_gq (blkg) is association between a block cgroup (blkcg) and a
	115	* request_queue (q). This is used by blkcg policies which need to track
	116	* information per blkcg - q pair.
	117	*
	118	* There can be multiple active blkcg policies and each blkg:policy pair is
	119	* represented by a blkg_policy_data which is allocated and freed by each
	120	* policy's pd_alloc/free_fn() methods. A policy can allocate private data
	121	* area by allocating larger data structure which embeds blkg_policy_data
	122	* at the beginning.
	123	*/
	124	struct blkg_policy_data {
	125	/* the blkg and policy id this per-policy data belongs to */
	126	struct blkcg_gq *blkg;
	127	int plid;
	128	};
	129
	130	/*
	131	* Policies that need to keep per-blkcg data which is independent from any
	132	* request_queue associated to it should implement cpd_alloc/free_fn()
	133	* methods. A policy can allocate private data area by allocating larger
	134	* data structure which embeds blkcg_policy_data at the beginning.
	135	* cpd_init() is invoked to let each policy handle per-blkcg data.
	136	*/
	137	struct blkcg_policy_data {
	138	/* the blkcg and policy id this per-policy data belongs to */
	139	struct blkcg *blkcg;
	140	int plid;
	141	};
	142
	143	typedef struct blkcg_policy_data *(blkcg_pol_alloc_cpd_fn)(gfp_t gfp);
	144	typedef void (blkcg_pol_init_cpd_fn)(struct blkcg_policy_data *cpd);
	145	typedef void (blkcg_pol_free_cpd_fn)(struct blkcg_policy_data *cpd);
	146	typedef void (blkcg_pol_bind_cpd_fn)(struct blkcg_policy_data *cpd);
	147	typedef struct blkg_policy_data *(blkcg_pol_alloc_pd_fn)(gfp_t gfp,
	148	struct request_queue q, struct blkcg blkcg);
	149	typedef void (blkcg_pol_init_pd_fn)(struct blkg_policy_data *pd);
	150	typedef void (blkcg_pol_online_pd_fn)(struct blkg_policy_data *pd);
	151	typedef void (blkcg_pol_offline_pd_fn)(struct blkg_policy_data *pd);
	152	typedef void (blkcg_pol_free_pd_fn)(struct blkg_policy_data *pd);
	153	typedef void (blkcg_pol_reset_pd_stats_fn)(struct blkg_policy_data *pd);
3607849d	154	typedef void (blkcg_pol_stat_pd_fn)(struct blkg_policy_data *pd,
672fdcf0 ML	155	struct seq_file *s);
	156
	157	struct blkcg_policy {
	158	int plid;
	159	/* cgroup files for the policy */
	160	struct cftype *dfl_cftypes;
	161	struct cftype *legacy_cftypes;
	162
	163	/* operations */
	164	blkcg_pol_alloc_cpd_fn *cpd_alloc_fn;
	165	blkcg_pol_init_cpd_fn *cpd_init_fn;
	166	blkcg_pol_free_cpd_fn *cpd_free_fn;
	167	blkcg_pol_bind_cpd_fn *cpd_bind_fn;
	168
	169	blkcg_pol_alloc_pd_fn *pd_alloc_fn;
	170	blkcg_pol_init_pd_fn *pd_init_fn;
	171	blkcg_pol_online_pd_fn *pd_online_fn;
	172	blkcg_pol_offline_pd_fn *pd_offline_fn;
	173	blkcg_pol_free_pd_fn *pd_free_fn;
	174	blkcg_pol_reset_pd_stats_fn *pd_reset_stats_fn;
	175	blkcg_pol_stat_pd_fn *pd_stat_fn;
	176	};
	177
	178	extern struct blkcg blkcg_root;
	179	extern bool blkcg_debug_stats;
	180
9823538f CH	181	int blkcg_init_disk(struct gendisk *disk);
9823538f CH	182	void blkcg_exit_disk(struct gendisk *disk);
672fdcf0 ML	183
	184	/* Blkio controller policy registration */
	185	int blkcg_policy_register(struct blkcg_policy *pol);
	186	void blkcg_policy_unregister(struct blkcg_policy *pol);
	187	int blkcg_activate_policy(struct request_queue *q,
	188	const struct blkcg_policy *pol);
	189	void blkcg_deactivate_policy(struct request_queue *q,
	190	const struct blkcg_policy *pol);
	191
	192	const char blkg_dev_name(struct blkcg_gq blkg);
	193	void blkcg_print_blkgs(struct seq_file sf, struct blkcg blkcg,
	194	u64 (prfill)(struct seq_file ,
	195	struct blkg_policy_data *, int),
	196	const struct blkcg_policy *pol, int data,
	197	bool show_total);
	198	u64 __blkg_prfill_u64(struct seq_file sf, struct blkg_policy_data pd, u64 v);
	199
	200	struct blkg_conf_ctx {
	201	struct block_device *bdev;
	202	struct blkcg_gq *blkg;
	203	char *body;
	204	};
	205
	206	struct block_device blkcg_conf_open_bdev(char *inputp);
	207	int blkg_conf_prep(struct blkcg blkcg, const struct blkcg_policy pol,
	208	char input, struct blkg_conf_ctx ctx);
	209	void blkg_conf_finish(struct blkg_conf_ctx *ctx);
	210
672fdcf0 ML	211	/**
	212	* bio_issue_as_root_blkg - see if this bio needs to be issued as root blkg
	213	* @return: true if this bio needs to be submitted with the root blkg context.
	214	*
	215	* In order to avoid priority inversions we sometimes need to issue a bio as if
	216	* it were attached to the root blkg, and then backcharge to the actual owning
bbb1ebe7 CH	217	* blkg. The idea is we do bio_blkcg_css() to look up the actual context for
	218	* the bio and attach the appropriate blkg to the bio. Then we call this helper
	219	* and if it is true run with the root blkg for that queue and then do any
672fdcf0 ML	220	* backcharging to the originating cgroup once the io is complete.
	221	*/
	222	static inline bool bio_issue_as_root_blkg(struct bio *bio)
	223	{
	224	return (bio->bi_opf & (REQ_META \| REQ_SWAP)) != 0;
	225	}
	226
	227	/**
4a69f325	228	* blkg_lookup - lookup blkg for the specified blkcg - q pair
672fdcf0 ML	229	* @blkcg: blkcg of interest
672fdcf0 ML	230	* @q: request_queue of interest
672fdcf0	231	*
4a69f325 CH	232	* Lookup blkg for the @blkcg - @q pair.
	233
	234	* Must be called in a RCU critical section.
672fdcf0	235	*/
4a69f325 CH	236	static inline struct blkcg_gq blkg_lookup(struct blkcg blkcg,
4a69f325 CH	237	struct request_queue *q)
672fdcf0 ML	238	{
	239	struct blkcg_gq *blkg;
	240
4a69f325 CH	241	WARN_ON_ONCE(!rcu_read_lock_held());
4a69f325 CH	242
672fdcf0 ML	243	if (blkcg == &blkcg_root)
	244	return q->root_blkg;
	245
	246	blkg = rcu_dereference(blkcg->blkg_hint);
	247	if (blkg && blkg->q == q)
	248	return blkg;
	249
4a69f325 CH	250	blkg = radix_tree_lookup(&blkcg->blkg_tree, q->id);
	251	if (blkg && blkg->q != q)
	252	blkg = NULL;
	253	return blkg;
672fdcf0 ML	254	}
672fdcf0 ML	255
672fdcf0 ML	256	/**
	257	* blkg_to_pdata - get policy private data
	258	* @blkg: blkg of interest
	259	* @pol: policy of interest
	260	*
	261	* Return pointer to private data associated with the @blkg-@pol pair.
	262	*/
	263	static inline struct blkg_policy_data blkg_to_pd(struct blkcg_gq blkg,
	264	struct blkcg_policy *pol)
	265	{
	266	return blkg ? blkg->pd[pol->plid] : NULL;
	267	}
	268
	269	static inline struct blkcg_policy_data blkcg_to_cpd(struct blkcg blkcg,
	270	struct blkcg_policy *pol)
	271	{
	272	return blkcg ? blkcg->cpd[pol->plid] : NULL;
	273	}
	274
	275	/**
	276	* pdata_to_blkg - get blkg associated with policy private data
	277	* @pd: policy private data of interest
	278	*
	279	* @pd is policy private data. Determine the blkg it's associated with.
	280	*/
	281	static inline struct blkcg_gq pd_to_blkg(struct blkg_policy_data pd)
	282	{
	283	return pd ? pd->blkg : NULL;
	284	}
	285
	286	static inline struct blkcg cpd_to_blkcg(struct blkcg_policy_data cpd)
	287	{
	288	return cpd ? cpd->blkcg : NULL;
	289	}
	290
	291	/**
	292	* blkg_path - format cgroup path of blkg
	293	* @blkg: blkg of interest
	294	* @buf: target buffer
	295	* @buflen: target buffer length
	296	*
	297	* Format the path of the cgroup of @blkg into @buf.
	298	*/
	299	static inline int blkg_path(struct blkcg_gq blkg, char buf, int buflen)
	300	{
	301	return cgroup_path(blkg->blkcg->css.cgroup, buf, buflen);
	302	}
	303
	304	/**
	305	* blkg_get - get a blkg reference
	306	* @blkg: blkg to get
	307	*
	308	* The caller should be holding an existing reference.
	309	*/
	310	static inline void blkg_get(struct blkcg_gq *blkg)
	311	{
	312	percpu_ref_get(&blkg->refcnt);
	313	}
	314
	315	/**
	316	* blkg_tryget - try and get a blkg reference
	317	* @blkg: blkg to get
	318	*
	319	* This is for use when doing an RCU lookup of the blkg. We may be in the midst
320	* of freeing this blkg, so we can only use it if the refcnt is not zero.
321	*/
322	static inline bool blkg_tryget(struct blkcg_gq *blkg)
323	{
324	return blkg && percpu_ref_tryget(&blkg->refcnt);
325	}
326
327	/**
328	* blkg_put - put a blkg reference
329	* @blkg: blkg to put
330	*/
331	static inline void blkg_put(struct blkcg_gq *blkg)
332	{
333	percpu_ref_put(&blkg->refcnt);
334	}
335
336	/**
337	* blkg_for_each_descendant_pre - pre-order walk of a blkg's descendants
338	* @d_blkg: loop cursor pointing to the current descendant
339	* @pos_css: used for iteration
340	* @p_blkg: target blkg to walk descendants of
341	*
342	* Walk @c_blkg through the descendants of @p_blkg. Must be used with RCU
343	* read locked. If called under either blkcg or queue lock, the iteration
344	* is guaranteed to include all and only online blkgs. The caller may
345	* update @pos_css by calling css_rightmost_descendant() to skip subtree.
346	* @p_blkg is included in the iteration and the first node to be visited.
347	*/
348	#define blkg_for_each_descendant_pre(d_blkg, pos_css, p_blkg) \
349	css_for_each_descendant_pre((pos_css), &(p_blkg)->blkcg->css) \
79fcc5be CH	350	if (((d_blkg) = blkg_lookup(css_to_blkcg(pos_css), \
79fcc5be CH	351	(p_blkg)->q)))
672fdcf0 ML	352
	353	/**
	354	* blkg_for_each_descendant_post - post-order walk of a blkg's descendants
	355	* @d_blkg: loop cursor pointing to the current descendant
	356	* @pos_css: used for iteration
	357	* @p_blkg: target blkg to walk descendants of
	358	*
	359	* Similar to blkg_for_each_descendant_pre() but performs post-order
	360	* traversal instead. Synchronization rules are the same. @p_blkg is
	361	* included in the iteration and the last node to be visited.
	362	*/
	363	#define blkg_for_each_descendant_post(d_blkg, pos_css, p_blkg) \
	364	css_for_each_descendant_post((pos_css), &(p_blkg)->blkcg->css) \
79fcc5be CH	365	if (((d_blkg) = blkg_lookup(css_to_blkcg(pos_css), \
79fcc5be CH	366	(p_blkg)->q)))
672fdcf0 ML	367
	368	bool __blkcg_punt_bio_submit(struct bio *bio);
	369
	370	static inline bool blkcg_punt_bio_submit(struct bio *bio)
	371	{
	372	if (bio->bi_opf & REQ_CGROUP_PUNT)
	373	return __blkcg_punt_bio_submit(bio);
	374	else
	375	return false;
	376	}
	377
	378	static inline void blkcg_bio_issue_init(struct bio *bio)
	379	{
	380	bio_issue_init(&bio->bi_issue, bio_sectors(bio));
	381	}
	382
	383	static inline void blkcg_use_delay(struct blkcg_gq *blkg)
	384	{
	385	if (WARN_ON_ONCE(atomic_read(&blkg->use_delay) < 0))
	386	return;
	387	if (atomic_add_return(1, &blkg->use_delay) == 1)
	388	atomic_inc(&blkg->blkcg->css.cgroup->congestion_count);
	389	}
	390
	391	static inline int blkcg_unuse_delay(struct blkcg_gq *blkg)
	392	{
	393	int old = atomic_read(&blkg->use_delay);
	394
	395	if (WARN_ON_ONCE(old < 0))
	396	return 0;
	397	if (old == 0)
	398	return 0;
	399
	400	/*
	401	* We do this song and dance because we can race with somebody else
	402	* adding or removing delay. If we just did an atomic_dec we'd end up
	403	* negative and we'd already be in trouble. We need to subtract 1 and
	404	* then check to see if we were the last delay so we can drop the
	405	* congestion count on the cgroup.
	406	*/
96388f57 UB	407	while (old && !atomic_try_cmpxchg(&blkg->use_delay, &old, old - 1))
96388f57 UB	408	;
672fdcf0 ML	409
	410	if (old == 0)
	411	return 0;
	412	if (old == 1)
	413	atomic_dec(&blkg->blkcg->css.cgroup->congestion_count);
	414	return 1;
	415	}
	416
	417	/**
	418	* blkcg_set_delay - Enable allocator delay mechanism with the specified delay amount
	419	* @blkg: target blkg
	420	* @delay: delay duration in nsecs
	421	*
	422	* When enabled with this function, the delay is not decayed and must be
	423	* explicitly cleared with blkcg_clear_delay(). Must not be mixed with
	424	* blkcg_[un]use_delay() and blkcg_add_delay() usages.
	425	*/
	426	static inline void blkcg_set_delay(struct blkcg_gq *blkg, u64 delay)
	427	{
	428	int old = atomic_read(&blkg->use_delay);
	429
	430	/* We only want 1 person setting the congestion count for this blkg. */
96388f57	431	if (!old && atomic_try_cmpxchg(&blkg->use_delay, &old, -1))
672fdcf0 ML	432	atomic_inc(&blkg->blkcg->css.cgroup->congestion_count);
	433
	434	atomic64_set(&blkg->delay_nsec, delay);
	435	}
	436
	437	/**
	438	* blkcg_clear_delay - Disable allocator delay mechanism
	439	* @blkg: target blkg
	440	*
	441	* Disable use_delay mechanism. See blkcg_set_delay().
	442	*/
	443	static inline void blkcg_clear_delay(struct blkcg_gq *blkg)
	444	{
	445	int old = atomic_read(&blkg->use_delay);
	446
	447	/* We only want 1 person clearing the congestion count for this blkg. */
96388f57	448	if (old && atomic_try_cmpxchg(&blkg->use_delay, &old, 0))
672fdcf0 ML	449	atomic_dec(&blkg->blkcg->css.cgroup->congestion_count);
	450	}
	451
6b2b0459 TH	452	/**
	453	* blk_cgroup_mergeable - Determine whether to allow or disallow merges
	454	* @rq: request to merge into
	455	* @bio: bio to merge
	456	*
	457	* @bio and @rq should belong to the same cgroup and their issue_as_root should
	458	* match. The latter is necessary as we don't want to throttle e.g. a metadata
	459	* update because it happens to be next to a regular IO.
	460	*/
	461	static inline bool blk_cgroup_mergeable(struct request rq, struct bio bio)
	462	{
	463	return rq->bio->bi_blkg == bio->bi_blkg &&
	464	bio_issue_as_root_blkg(rq->bio) == bio_issue_as_root_blkg(bio);
	465	}
	466
672fdcf0 ML	467	void blk_cgroup_bio_start(struct bio *bio);
	468	void blkcg_add_delay(struct blkcg_gq *blkg, u64 now, u64 delta);
	469	#else /* CONFIG_BLK_CGROUP */
	470
	471	struct blkg_policy_data {
	472	};
	473
	474	struct blkcg_policy_data {
	475	};
	476
	477	struct blkcg_policy {
	478	};
	479
bbb1ebe7 CH	480	struct blkcg {
	481	};
	482
672fdcf0	483	static inline struct blkcg_gq blkg_lookup(struct blkcg blkcg, void *key) { return NULL; }
9823538f CH	484	static inline int blkcg_init_disk(struct gendisk *disk) { return 0; }
9823538f CH	485	static inline void blkcg_exit_disk(struct gendisk *disk) { }
672fdcf0 ML	486	static inline int blkcg_policy_register(struct blkcg_policy *pol) { return 0; }
	487	static inline void blkcg_policy_unregister(struct blkcg_policy *pol) { }
	488	static inline int blkcg_activate_policy(struct request_queue *q,
	489	const struct blkcg_policy *pol) { return 0; }
	490	static inline void blkcg_deactivate_policy(struct request_queue *q,
	491	const struct blkcg_policy *pol) { }
	492
672fdcf0 ML	493	static inline struct blkg_policy_data blkg_to_pd(struct blkcg_gq blkg,
	494	struct blkcg_policy *pol) { return NULL; }
	495	static inline struct blkcg_gq pd_to_blkg(struct blkg_policy_data pd) { return NULL; }
	496	static inline char blkg_path(struct blkcg_gq blkg) { return NULL; }
	497	static inline void blkg_get(struct blkcg_gq *blkg) { }
	498	static inline void blkg_put(struct blkcg_gq *blkg) { }
	499
	500	static inline bool blkcg_punt_bio_submit(struct bio *bio) { return false; }
	501	static inline void blkcg_bio_issue_init(struct bio *bio) { }
	502	static inline void blk_cgroup_bio_start(struct bio *bio) { }
6b2b0459	503	static inline bool blk_cgroup_mergeable(struct request rq, struct bio bio) { return true; }
672fdcf0 ML	504
	505	#define blk_queue_for_each_rl(rl, q) \
	506	for ((rl) = &(q)->root_rl; (rl); (rl) = NULL)
	507
672fdcf0 ML	508	#endif /* CONFIG_BLK_CGROUP */
	509
	510	#endif /* _BLK_CGROUP_PRIVATE_H */