[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>
#include <linux/llist.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 blkcg_gq		       *blkg;
	struct llist_node		lnode;
	int				lqueued;	/* queued in llist */
	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;

	/*
	 * List of updated percpu blkg_iostat_set's since the last flush.
	 */
	struct llist_head __percpu	*lhead;

#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;
	bool				online;
};

/*
 * 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)(struct gendisk *disk,
		struct blkcg *blkcg, gfp_t gfp);
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 gendisk *disk, const struct blkcg_policy *pol);
void blkcg_deactivate_policy(struct gendisk *disk,
			     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 gendisk *disk,
					const struct blkcg_policy *pol) { return 0; }
static inline void blkcg_deactivate_policy(struct gendisk *disk,
					   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>
3b8cc629	21	#include <linux/llist.h>
672fdcf0	22
dec223c9 CH	23	struct blkcg_gq;
	24	struct blkg_policy_data;
	25
	26
672fdcf0 ML	27	/* percpu_counter batch for blkg_[rw]stats, per-cpu drift doesn't matter */
	28	#define BLKG_STAT_CPU_BATCH (INT_MAX / 2)
	29
	30	#ifdef CONFIG_BLK_CGROUP
bbb1ebe7 CH	31
	32	enum blkg_iostat_type {
	33	BLKG_IOSTAT_READ,
	34	BLKG_IOSTAT_WRITE,
	35	BLKG_IOSTAT_DISCARD,
	36
	37	BLKG_IOSTAT_NR,
	38	};
	39
	40	struct blkg_iostat {
	41	u64 bytes[BLKG_IOSTAT_NR];
	42	u64 ios[BLKG_IOSTAT_NR];
	43	};
	44
	45	struct blkg_iostat_set {
	46	struct u64_stats_sync sync;
3b8cc629 WL	47	struct blkcg_gq *blkg;
	48	struct llist_node lnode;
	49	int lqueued; /* queued in llist */
bbb1ebe7 CH	50	struct blkg_iostat cur;
	51	struct blkg_iostat last;
	52	};
	53
	54	/* association between a blk cgroup and a request queue */
	55	struct blkcg_gq {
a06377c5 CH	56	/* Pointer to the associated request_queue */
a06377c5 CH	57	struct request_queue *q;
1231039d	58	struct list_head q_node;
bbb1ebe7 CH	59	struct hlist_node blkcg_node;
	60	struct blkcg *blkcg;
	61
	62	/* all non-root blkcg_gq's are guaranteed to have access to parent */
	63	struct blkcg_gq *parent;
	64
	65	/* reference count */
	66	struct percpu_ref refcnt;
	67
	68	/* is this blkg online? protected by both blkcg and q locks */
	69	bool online;
	70
	71	struct blkg_iostat_set __percpu *iostat_cpu;
	72	struct blkg_iostat_set iostat;
	73
	74	struct blkg_policy_data *pd[BLKCG_MAX_POLS];
	75
	76	spinlock_t async_bio_lock;
	77	struct bio_list async_bios;
	78	union {
	79	struct work_struct async_bio_work;
	80	struct work_struct free_work;
	81	};
	82
	83	atomic_t use_delay;
	84	atomic64_t delay_nsec;
	85	atomic64_t delay_start;
	86	u64 last_delay;
	87	int last_use;
	88
	89	struct rcu_head rcu_head;
	90	};
	91
dec223c9 CH	92	struct blkcg {
	93	struct cgroup_subsys_state css;
	94	spinlock_t lock;
	95	refcount_t online_pin;
	96
	97	struct radix_tree_root blkg_tree;
	98	struct blkcg_gq __rcu *blkg_hint;
	99	struct hlist_head blkg_list;
	100
	101	struct blkcg_policy_data *cpd[BLKCG_MAX_POLS];
	102
	103	struct list_head all_blkcgs_node;
3b8cc629 WL	104
	105	/*
	106	* List of updated percpu blkg_iostat_set's since the last flush.
	107	*/
	108	struct llist_head __percpu *lhead;
	109
dec223c9 CH	110	#ifdef CONFIG_BLK_CGROUP_FC_APPID
	111	char fc_app_id[FC_APPID_LEN];
	112	#endif
	113	#ifdef CONFIG_CGROUP_WRITEBACK
	114	struct list_head cgwb_list;
	115	#endif
	116	};
	117
	118	static inline struct blkcg css_to_blkcg(struct cgroup_subsys_state css)
	119	{
	120	return css ? container_of(css, struct blkcg, css) : NULL;
	121	}
672fdcf0 ML	122
	123	/*
	124	* A blkcg_gq (blkg) is association between a block cgroup (blkcg) and a
	125	* request_queue (q). This is used by blkcg policies which need to track
	126	* information per blkcg - q pair.
	127	*
	128	* There can be multiple active blkcg policies and each blkg:policy pair is
	129	* represented by a blkg_policy_data which is allocated and freed by each
	130	* policy's pd_alloc/free_fn() methods. A policy can allocate private data
	131	* area by allocating larger data structure which embeds blkg_policy_data
	132	* at the beginning.
	133	*/
	134	struct blkg_policy_data {
	135	/* the blkg and policy id this per-policy data belongs to */
	136	struct blkcg_gq *blkg;
	137	int plid;
dfd6200a	138	bool online;
672fdcf0 ML	139	};
	140
	141	/*
	142	* Policies that need to keep per-blkcg data which is independent from any
	143	* request_queue associated to it should implement cpd_alloc/free_fn()
	144	* methods. A policy can allocate private data area by allocating larger
	145	* data structure which embeds blkcg_policy_data at the beginning.
	146	* cpd_init() is invoked to let each policy handle per-blkcg data.
	147	*/
	148	struct blkcg_policy_data {
	149	/* the blkcg and policy id this per-policy data belongs to */
	150	struct blkcg *blkcg;
	151	int plid;
	152	};
	153
	154	typedef struct blkcg_policy_data *(blkcg_pol_alloc_cpd_fn)(gfp_t gfp);
	155	typedef void (blkcg_pol_init_cpd_fn)(struct blkcg_policy_data *cpd);
	156	typedef void (blkcg_pol_free_cpd_fn)(struct blkcg_policy_data *cpd);
	157	typedef void (blkcg_pol_bind_cpd_fn)(struct blkcg_policy_data *cpd);
0a0b4f79 CH	158	typedef struct blkg_policy_data (blkcg_pol_alloc_pd_fn)(struct gendisk disk,
0a0b4f79 CH	159	struct blkcg *blkcg, gfp_t gfp);
672fdcf0 ML	160	typedef void (blkcg_pol_init_pd_fn)(struct blkg_policy_data *pd);
	161	typedef void (blkcg_pol_online_pd_fn)(struct blkg_policy_data *pd);
	162	typedef void (blkcg_pol_offline_pd_fn)(struct blkg_policy_data *pd);
	163	typedef void (blkcg_pol_free_pd_fn)(struct blkg_policy_data *pd);
	164	typedef void (blkcg_pol_reset_pd_stats_fn)(struct blkg_policy_data *pd);
3607849d	165	typedef void (blkcg_pol_stat_pd_fn)(struct blkg_policy_data *pd,
672fdcf0 ML	166	struct seq_file *s);
	167
	168	struct blkcg_policy {
	169	int plid;
	170	/* cgroup files for the policy */
	171	struct cftype *dfl_cftypes;
	172	struct cftype *legacy_cftypes;
	173
	174	/* operations */
	175	blkcg_pol_alloc_cpd_fn *cpd_alloc_fn;
	176	blkcg_pol_init_cpd_fn *cpd_init_fn;
	177	blkcg_pol_free_cpd_fn *cpd_free_fn;
	178	blkcg_pol_bind_cpd_fn *cpd_bind_fn;
	179
	180	blkcg_pol_alloc_pd_fn *pd_alloc_fn;
	181	blkcg_pol_init_pd_fn *pd_init_fn;
	182	blkcg_pol_online_pd_fn *pd_online_fn;
	183	blkcg_pol_offline_pd_fn *pd_offline_fn;
	184	blkcg_pol_free_pd_fn *pd_free_fn;
	185	blkcg_pol_reset_pd_stats_fn *pd_reset_stats_fn;
	186	blkcg_pol_stat_pd_fn *pd_stat_fn;
	187	};
	188
	189	extern struct blkcg blkcg_root;
	190	extern bool blkcg_debug_stats;
	191
9823538f CH	192	int blkcg_init_disk(struct gendisk *disk);
9823538f CH	193	void blkcg_exit_disk(struct gendisk *disk);
672fdcf0 ML	194
	195	/* Blkio controller policy registration */
	196	int blkcg_policy_register(struct blkcg_policy *pol);
	197	void blkcg_policy_unregister(struct blkcg_policy *pol);
40e4996e CH	198	int blkcg_activate_policy(struct gendisk disk, const struct blkcg_policy pol);
40e4996e CH	199	void blkcg_deactivate_policy(struct gendisk *disk,
672fdcf0 ML	200	const struct blkcg_policy *pol);
	201
	202	const char blkg_dev_name(struct blkcg_gq blkg);
	203	void blkcg_print_blkgs(struct seq_file sf, struct blkcg blkcg,
	204	u64 (prfill)(struct seq_file ,
	205	struct blkg_policy_data *, int),
	206	const struct blkcg_policy *pol, int data,
	207	bool show_total);
	208	u64 __blkg_prfill_u64(struct seq_file sf, struct blkg_policy_data pd, u64 v);
	209
	210	struct blkg_conf_ctx {
	211	struct block_device *bdev;
	212	struct blkcg_gq *blkg;
	213	char *body;
	214	};
	215
	216	struct block_device blkcg_conf_open_bdev(char *inputp);
	217	int blkg_conf_prep(struct blkcg blkcg, const struct blkcg_policy pol,
	218	char input, struct blkg_conf_ctx ctx);
	219	void blkg_conf_finish(struct blkg_conf_ctx *ctx);
	220
672fdcf0 ML	221	/**
	222	* bio_issue_as_root_blkg - see if this bio needs to be issued as root blkg
	223	* @return: true if this bio needs to be submitted with the root blkg context.
	224	*
	225	* In order to avoid priority inversions we sometimes need to issue a bio as if
	226	* it were attached to the root blkg, and then backcharge to the actual owning
bbb1ebe7 CH	227	* blkg. The idea is we do bio_blkcg_css() to look up the actual context for
	228	* the bio and attach the appropriate blkg to the bio. Then we call this helper
	229	* and if it is true run with the root blkg for that queue and then do any
672fdcf0 ML	230	* backcharging to the originating cgroup once the io is complete.
	231	*/
	232	static inline bool bio_issue_as_root_blkg(struct bio *bio)
	233	{
	234	return (bio->bi_opf & (REQ_META \| REQ_SWAP)) != 0;
	235	}
	236
	237	/**
9a9c261e	238	* blkg_lookup - lookup blkg for the specified blkcg - q pair
672fdcf0	239	* @blkcg: blkcg of interest
9a9c261e	240	* @q: request_queue of interest
672fdcf0	241	*
9a9c261e	242	* Lookup blkg for the @blkcg - @q pair.
4a69f325 CH	243
4a69f325 CH	244	* Must be called in a RCU critical section.
672fdcf0	245	*/
4a69f325	246	static inline struct blkcg_gq blkg_lookup(struct blkcg blkcg,
9a9c261e	247	struct request_queue *q)
672fdcf0 ML	248	{
	249	struct blkcg_gq *blkg;
	250
4a69f325 CH	251	WARN_ON_ONCE(!rcu_read_lock_held());
4a69f325 CH	252
672fdcf0	253	if (blkcg == &blkcg_root)
9a9c261e	254	return q->root_blkg;
672fdcf0 ML	255
672fdcf0 ML	256	blkg = rcu_dereference(blkcg->blkg_hint);
a06377c5	257	if (blkg && blkg->q == q)
672fdcf0 ML	258	return blkg;
672fdcf0 ML	259
9a9c261e	260	blkg = radix_tree_lookup(&blkcg->blkg_tree, q->id);
a06377c5	261	if (blkg && blkg->q != q)
4a69f325 CH	262	blkg = NULL;
4a69f325 CH	263	return blkg;
672fdcf0 ML	264	}
672fdcf0 ML	265
672fdcf0 ML	266	/**
	267	* blkg_to_pdata - get policy private data
	268	* @blkg: blkg of interest
	269	* @pol: policy of interest
	270	*
	271	* Return pointer to private data associated with the @blkg-@pol pair.
	272	*/
	273	static inline struct blkg_policy_data blkg_to_pd(struct blkcg_gq blkg,
	274	struct blkcg_policy *pol)
	275	{
	276	return blkg ? blkg->pd[pol->plid] : NULL;
	277	}
	278
	279	static inline struct blkcg_policy_data blkcg_to_cpd(struct blkcg blkcg,
	280	struct blkcg_policy *pol)
	281	{
	282	return blkcg ? blkcg->cpd[pol->plid] : NULL;
	283	}
	284
	285	/**
	286	* pdata_to_blkg - get blkg associated with policy private data
	287	* @pd: policy private data of interest
	288	*
	289	* @pd is policy private data. Determine the blkg it's associated with.
	290	*/
	291	static inline struct blkcg_gq pd_to_blkg(struct blkg_policy_data pd)
	292	{
	293	return pd ? pd->blkg : NULL;
	294	}
	295
	296	static inline struct blkcg cpd_to_blkcg(struct blkcg_policy_data cpd)
	297	{
	298	return cpd ? cpd->blkcg : NULL;
	299	}
	300
	301	/**
	302	* blkg_path - format cgroup path of blkg
	303	* @blkg: blkg of interest
	304	* @buf: target buffer
	305	* @buflen: target buffer length
	306	*
	307	* Format the path of the cgroup of @blkg into @buf.
	308	*/
	309	static inline int blkg_path(struct blkcg_gq blkg, char buf, int buflen)
	310	{
	311	return cgroup_path(blkg->blkcg->css.cgroup, buf, buflen);
	312	}
	313
	314	/**
	315	* blkg_get - get a blkg reference
	316	* @blkg: blkg to get
	317	*
	318	* The caller should be holding an existing reference.
	319	*/
	320	static inline void blkg_get(struct blkcg_gq *blkg)
	321	{
	322	percpu_ref_get(&blkg->refcnt);
	323	}
	324
	325	/**
	326	* blkg_tryget - try and get a blkg reference
	327	* @blkg: blkg to get
	328	*
	329	* This is for use when doing an RCU lookup of the blkg. We may be in the midst
330	* of freeing this blkg, so we can only use it if the refcnt is not zero.
331	*/
332	static inline bool blkg_tryget(struct blkcg_gq *blkg)
333	{
334	return blkg && percpu_ref_tryget(&blkg->refcnt);
335	}
336
337	/**
338	* blkg_put - put a blkg reference
339	* @blkg: blkg to put
340	*/
341	static inline void blkg_put(struct blkcg_gq *blkg)
342	{
343	percpu_ref_put(&blkg->refcnt);
344	}
345
346	/**
347	* blkg_for_each_descendant_pre - pre-order walk of a blkg's descendants
348	* @d_blkg: loop cursor pointing to the current descendant
349	* @pos_css: used for iteration
350	* @p_blkg: target blkg to walk descendants of
351	*
352	* Walk @c_blkg through the descendants of @p_blkg. Must be used with RCU
353	* read locked. If called under either blkcg or queue lock, the iteration
354	* is guaranteed to include all and only online blkgs. The caller may
355	* update @pos_css by calling css_rightmost_descendant() to skip subtree.
356	* @p_blkg is included in the iteration and the first node to be visited.
357	*/
358	#define blkg_for_each_descendant_pre(d_blkg, pos_css, p_blkg) \
359	css_for_each_descendant_pre((pos_css), &(p_blkg)->blkcg->css) \
79fcc5be	360	if (((d_blkg) = blkg_lookup(css_to_blkcg(pos_css), \
a06377c5	361	(p_blkg)->q)))
672fdcf0 ML	362
	363	/**
	364	* blkg_for_each_descendant_post - post-order walk of a blkg's descendants
	365	* @d_blkg: loop cursor pointing to the current descendant
	366	* @pos_css: used for iteration
	367	* @p_blkg: target blkg to walk descendants of
	368	*
	369	* Similar to blkg_for_each_descendant_pre() but performs post-order
	370	* traversal instead. Synchronization rules are the same. @p_blkg is
	371	* included in the iteration and the last node to be visited.
	372	*/
	373	#define blkg_for_each_descendant_post(d_blkg, pos_css, p_blkg) \
	374	css_for_each_descendant_post((pos_css), &(p_blkg)->blkcg->css) \
79fcc5be	375	if (((d_blkg) = blkg_lookup(css_to_blkcg(pos_css), \
a06377c5	376	(p_blkg)->q)))
672fdcf0 ML	377
	378	bool __blkcg_punt_bio_submit(struct bio *bio);
	379
	380	static inline bool blkcg_punt_bio_submit(struct bio *bio)
	381	{
	382	if (bio->bi_opf & REQ_CGROUP_PUNT)
	383	return __blkcg_punt_bio_submit(bio);
	384	else
	385	return false;
	386	}
	387
	388	static inline void blkcg_bio_issue_init(struct bio *bio)
	389	{
	390	bio_issue_init(&bio->bi_issue, bio_sectors(bio));
	391	}
	392
	393	static inline void blkcg_use_delay(struct blkcg_gq *blkg)
	394	{
	395	if (WARN_ON_ONCE(atomic_read(&blkg->use_delay) < 0))
	396	return;
	397	if (atomic_add_return(1, &blkg->use_delay) == 1)
	398	atomic_inc(&blkg->blkcg->css.cgroup->congestion_count);
	399	}
	400
	401	static inline int blkcg_unuse_delay(struct blkcg_gq *blkg)
	402	{
	403	int old = atomic_read(&blkg->use_delay);
	404
	405	if (WARN_ON_ONCE(old < 0))
	406	return 0;
	407	if (old == 0)
	408	return 0;
	409
	410	/*
	411	* We do this song and dance because we can race with somebody else
	412	* adding or removing delay. If we just did an atomic_dec we'd end up
	413	* negative and we'd already be in trouble. We need to subtract 1 and
	414	* then check to see if we were the last delay so we can drop the
	415	* congestion count on the cgroup.
	416	*/
96388f57 UB	417	while (old && !atomic_try_cmpxchg(&blkg->use_delay, &old, old - 1))
96388f57 UB	418	;
672fdcf0 ML	419
	420	if (old == 0)
	421	return 0;
	422	if (old == 1)
	423	atomic_dec(&blkg->blkcg->css.cgroup->congestion_count);
	424	return 1;
	425	}
	426
	427	/**
	428	* blkcg_set_delay - Enable allocator delay mechanism with the specified delay amount
	429	* @blkg: target blkg
	430	* @delay: delay duration in nsecs
	431	*
	432	* When enabled with this function, the delay is not decayed and must be
	433	* explicitly cleared with blkcg_clear_delay(). Must not be mixed with
	434	* blkcg_[un]use_delay() and blkcg_add_delay() usages.
	435	*/
	436	static inline void blkcg_set_delay(struct blkcg_gq *blkg, u64 delay)
	437	{
	438	int old = atomic_read(&blkg->use_delay);
	439
	440	/* We only want 1 person setting the congestion count for this blkg. */
96388f57	441	if (!old && atomic_try_cmpxchg(&blkg->use_delay, &old, -1))
672fdcf0 ML	442	atomic_inc(&blkg->blkcg->css.cgroup->congestion_count);
	443
	444	atomic64_set(&blkg->delay_nsec, delay);
	445	}
	446
	447	/**
	448	* blkcg_clear_delay - Disable allocator delay mechanism
	449	* @blkg: target blkg
	450	*
	451	* Disable use_delay mechanism. See blkcg_set_delay().
	452	*/
	453	static inline void blkcg_clear_delay(struct blkcg_gq *blkg)
	454	{
	455	int old = atomic_read(&blkg->use_delay);
	456
	457	/* We only want 1 person clearing the congestion count for this blkg. */
96388f57	458	if (old && atomic_try_cmpxchg(&blkg->use_delay, &old, 0))
672fdcf0 ML	459	atomic_dec(&blkg->blkcg->css.cgroup->congestion_count);
	460	}
	461
6b2b0459 TH	462	/**
	463	* blk_cgroup_mergeable - Determine whether to allow or disallow merges
	464	* @rq: request to merge into
	465	* @bio: bio to merge
	466	*
	467	* @bio and @rq should belong to the same cgroup and their issue_as_root should
	468	* match. The latter is necessary as we don't want to throttle e.g. a metadata
	469	* update because it happens to be next to a regular IO.
	470	*/
	471	static inline bool blk_cgroup_mergeable(struct request rq, struct bio bio)
	472	{
	473	return rq->bio->bi_blkg == bio->bi_blkg &&
	474	bio_issue_as_root_blkg(rq->bio) == bio_issue_as_root_blkg(bio);
	475	}
	476
672fdcf0 ML	477	void blk_cgroup_bio_start(struct bio *bio);
	478	void blkcg_add_delay(struct blkcg_gq *blkg, u64 now, u64 delta);
	479	#else /* CONFIG_BLK_CGROUP */
	480
	481	struct blkg_policy_data {
	482	};
	483
	484	struct blkcg_policy_data {
	485	};
	486
	487	struct blkcg_policy {
	488	};
	489
bbb1ebe7 CH	490	struct blkcg {
	491	};
	492
672fdcf0	493	static inline struct blkcg_gq blkg_lookup(struct blkcg blkcg, void *key) { return NULL; }
9823538f CH	494	static inline int blkcg_init_disk(struct gendisk *disk) { return 0; }
9823538f CH	495	static inline void blkcg_exit_disk(struct gendisk *disk) { }
672fdcf0 ML	496	static inline int blkcg_policy_register(struct blkcg_policy *pol) { return 0; }
672fdcf0 ML	497	static inline void blkcg_policy_unregister(struct blkcg_policy *pol) { }
40e4996e	498	static inline int blkcg_activate_policy(struct gendisk *disk,
672fdcf0	499	const struct blkcg_policy *pol) { return 0; }
40e4996e	500	static inline void blkcg_deactivate_policy(struct gendisk *disk,
672fdcf0 ML	501	const struct blkcg_policy *pol) { }
672fdcf0 ML	502
672fdcf0 ML	503	static inline struct blkg_policy_data blkg_to_pd(struct blkcg_gq blkg,
	504	struct blkcg_policy *pol) { return NULL; }
	505	static inline struct blkcg_gq pd_to_blkg(struct blkg_policy_data pd) { return NULL; }
	506	static inline char blkg_path(struct blkcg_gq blkg) { return NULL; }
	507	static inline void blkg_get(struct blkcg_gq *blkg) { }
	508	static inline void blkg_put(struct blkcg_gq *blkg) { }
	509
	510	static inline bool blkcg_punt_bio_submit(struct bio *bio) { return false; }
	511	static inline void blkcg_bio_issue_init(struct bio *bio) { }
	512	static inline void blk_cgroup_bio_start(struct bio *bio) { }
6b2b0459	513	static inline bool blk_cgroup_mergeable(struct request rq, struct bio bio) { return true; }
672fdcf0 ML	514
	515	#define blk_queue_for_each_rl(rl, q) \
	516	for ((rl) = &(q)->root_rl; (rl); (rl) = NULL)
	517
672fdcf0 ML	518	#endif /* CONFIG_BLK_CGROUP */
	519
	520	#endif /* _BLK_CGROUP_PRIVATE_H */