[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 {
	struct gendisk			*disk;
	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->disk->queue == q)
		return blkg;

	blkg = radix_tree_lookup(&blkcg->blkg_tree, q->id);
	if (blkg && blkg->disk->queue != 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)->disk->queue)))

/**
 * 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)->disk->queue)))

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 {
84d7d462	56	struct gendisk *disk;
1231039d	57	struct list_head q_node;
bbb1ebe7 CH	58	struct hlist_node blkcg_node;
	59	struct blkcg *blkcg;
	60
	61	/* all non-root blkcg_gq's are guaranteed to have access to parent */
	62	struct blkcg_gq *parent;
	63
	64	/* reference count */
	65	struct percpu_ref refcnt;
	66
	67	/* is this blkg online? protected by both blkcg and q locks */
	68	bool online;
	69
	70	struct blkg_iostat_set __percpu *iostat_cpu;
	71	struct blkg_iostat_set iostat;
	72
	73	struct blkg_policy_data *pd[BLKCG_MAX_POLS];
	74
	75	spinlock_t async_bio_lock;
	76	struct bio_list async_bios;
	77	union {
	78	struct work_struct async_bio_work;
	79	struct work_struct free_work;
	80	};
	81
	82	atomic_t use_delay;
	83	atomic64_t delay_nsec;
	84	atomic64_t delay_start;
	85	u64 last_delay;
	86	int last_use;
	87
	88	struct rcu_head rcu_head;
	89	};
	90
dec223c9 CH	91	struct blkcg {
	92	struct cgroup_subsys_state css;
	93	spinlock_t lock;
	94	refcount_t online_pin;
	95
	96	struct radix_tree_root blkg_tree;
	97	struct blkcg_gq __rcu *blkg_hint;
	98	struct hlist_head blkg_list;
	99
	100	struct blkcg_policy_data *cpd[BLKCG_MAX_POLS];
	101
	102	struct list_head all_blkcgs_node;
3b8cc629 WL	103
	104	/*
	105	* List of updated percpu blkg_iostat_set's since the last flush.
	106	*/
	107	struct llist_head __percpu *lhead;
	108
dec223c9 CH	109	#ifdef CONFIG_BLK_CGROUP_FC_APPID
	110	char fc_app_id[FC_APPID_LEN];
	111	#endif
	112	#ifdef CONFIG_CGROUP_WRITEBACK
	113	struct list_head cgwb_list;
	114	#endif
	115	};
	116
	117	static inline struct blkcg css_to_blkcg(struct cgroup_subsys_state css)
	118	{
	119	return css ? container_of(css, struct blkcg, css) : NULL;
	120	}
672fdcf0 ML	121
	122	/*
	123	* A blkcg_gq (blkg) is association between a block cgroup (blkcg) and a
	124	* request_queue (q). This is used by blkcg policies which need to track
	125	* information per blkcg - q pair.
	126	*
	127	* There can be multiple active blkcg policies and each blkg:policy pair is
	128	* represented by a blkg_policy_data which is allocated and freed by each
	129	* policy's pd_alloc/free_fn() methods. A policy can allocate private data
	130	* area by allocating larger data structure which embeds blkg_policy_data
	131	* at the beginning.
	132	*/
	133	struct blkg_policy_data {
	134	/* the blkg and policy id this per-policy data belongs to */
	135	struct blkcg_gq *blkg;
	136	int plid;
dfd6200a	137	bool online;
672fdcf0 ML	138	};
	139
	140	/*
	141	* Policies that need to keep per-blkcg data which is independent from any
	142	* request_queue associated to it should implement cpd_alloc/free_fn()
	143	* methods. A policy can allocate private data area by allocating larger
	144	* data structure which embeds blkcg_policy_data at the beginning.
	145	* cpd_init() is invoked to let each policy handle per-blkcg data.
	146	*/
	147	struct blkcg_policy_data {
	148	/* the blkcg and policy id this per-policy data belongs to */
	149	struct blkcg *blkcg;
	150	int plid;
	151	};
	152
	153	typedef struct blkcg_policy_data *(blkcg_pol_alloc_cpd_fn)(gfp_t gfp);
	154	typedef void (blkcg_pol_init_cpd_fn)(struct blkcg_policy_data *cpd);
	155	typedef void (blkcg_pol_free_cpd_fn)(struct blkcg_policy_data *cpd);
	156	typedef void (blkcg_pol_bind_cpd_fn)(struct blkcg_policy_data *cpd);
0a0b4f79 CH	157	typedef struct blkg_policy_data (blkcg_pol_alloc_pd_fn)(struct gendisk disk,
0a0b4f79 CH	158	struct blkcg *blkcg, gfp_t gfp);
672fdcf0 ML	159	typedef void (blkcg_pol_init_pd_fn)(struct blkg_policy_data *pd);
	160	typedef void (blkcg_pol_online_pd_fn)(struct blkg_policy_data *pd);
	161	typedef void (blkcg_pol_offline_pd_fn)(struct blkg_policy_data *pd);
	162	typedef void (blkcg_pol_free_pd_fn)(struct blkg_policy_data *pd);
	163	typedef void (blkcg_pol_reset_pd_stats_fn)(struct blkg_policy_data *pd);
3607849d	164	typedef void (blkcg_pol_stat_pd_fn)(struct blkg_policy_data *pd,
672fdcf0 ML	165	struct seq_file *s);
	166
	167	struct blkcg_policy {
	168	int plid;
	169	/* cgroup files for the policy */
	170	struct cftype *dfl_cftypes;
	171	struct cftype *legacy_cftypes;
	172
	173	/* operations */
	174	blkcg_pol_alloc_cpd_fn *cpd_alloc_fn;
	175	blkcg_pol_init_cpd_fn *cpd_init_fn;
	176	blkcg_pol_free_cpd_fn *cpd_free_fn;
	177	blkcg_pol_bind_cpd_fn *cpd_bind_fn;
	178
	179	blkcg_pol_alloc_pd_fn *pd_alloc_fn;
	180	blkcg_pol_init_pd_fn *pd_init_fn;
	181	blkcg_pol_online_pd_fn *pd_online_fn;
	182	blkcg_pol_offline_pd_fn *pd_offline_fn;
	183	blkcg_pol_free_pd_fn *pd_free_fn;
	184	blkcg_pol_reset_pd_stats_fn *pd_reset_stats_fn;
	185	blkcg_pol_stat_pd_fn *pd_stat_fn;
	186	};
	187
	188	extern struct blkcg blkcg_root;
	189	extern bool blkcg_debug_stats;
	190
9823538f CH	191	int blkcg_init_disk(struct gendisk *disk);
9823538f CH	192	void blkcg_exit_disk(struct gendisk *disk);
672fdcf0 ML	193
	194	/* Blkio controller policy registration */
	195	int blkcg_policy_register(struct blkcg_policy *pol);
	196	void blkcg_policy_unregister(struct blkcg_policy *pol);
40e4996e CH	197	int blkcg_activate_policy(struct gendisk disk, const struct blkcg_policy pol);
40e4996e CH	198	void blkcg_deactivate_policy(struct gendisk *disk,
672fdcf0 ML	199	const struct blkcg_policy *pol);
	200
	201	const char blkg_dev_name(struct blkcg_gq blkg);
	202	void blkcg_print_blkgs(struct seq_file sf, struct blkcg blkcg,
	203	u64 (prfill)(struct seq_file ,
	204	struct blkg_policy_data *, int),
	205	const struct blkcg_policy *pol, int data,
	206	bool show_total);
	207	u64 __blkg_prfill_u64(struct seq_file sf, struct blkg_policy_data pd, u64 v);
	208
	209	struct blkg_conf_ctx {
	210	struct block_device *bdev;
	211	struct blkcg_gq *blkg;
	212	char *body;
	213	};
	214
	215	struct block_device blkcg_conf_open_bdev(char *inputp);
	216	int blkg_conf_prep(struct blkcg blkcg, const struct blkcg_policy pol,
	217	char input, struct blkg_conf_ctx ctx);
	218	void blkg_conf_finish(struct blkg_conf_ctx *ctx);
	219
672fdcf0 ML	220	/**
	221	* bio_issue_as_root_blkg - see if this bio needs to be issued as root blkg
	222	* @return: true if this bio needs to be submitted with the root blkg context.
	223	*
	224	* In order to avoid priority inversions we sometimes need to issue a bio as if
	225	* it were attached to the root blkg, and then backcharge to the actual owning
bbb1ebe7 CH	226	* blkg. The idea is we do bio_blkcg_css() to look up the actual context for
	227	* the bio and attach the appropriate blkg to the bio. Then we call this helper
	228	* and if it is true run with the root blkg for that queue and then do any
672fdcf0 ML	229	* backcharging to the originating cgroup once the io is complete.
	230	*/
	231	static inline bool bio_issue_as_root_blkg(struct bio *bio)
	232	{
	233	return (bio->bi_opf & (REQ_META \| REQ_SWAP)) != 0;
	234	}
	235
	236	/**
9a9c261e	237	* blkg_lookup - lookup blkg for the specified blkcg - q pair
672fdcf0	238	* @blkcg: blkcg of interest
9a9c261e	239	* @q: request_queue of interest
672fdcf0	240	*
9a9c261e	241	* Lookup blkg for the @blkcg - @q pair.
4a69f325 CH	242
4a69f325 CH	243	* Must be called in a RCU critical section.
672fdcf0	244	*/
4a69f325	245	static inline struct blkcg_gq blkg_lookup(struct blkcg blkcg,
9a9c261e	246	struct request_queue *q)
672fdcf0 ML	247	{
	248	struct blkcg_gq *blkg;
	249
4a69f325 CH	250	WARN_ON_ONCE(!rcu_read_lock_held());
4a69f325 CH	251
672fdcf0	252	if (blkcg == &blkcg_root)
9a9c261e	253	return q->root_blkg;
672fdcf0 ML	254
672fdcf0 ML	255	blkg = rcu_dereference(blkcg->blkg_hint);
9a9c261e	256	if (blkg && blkg->disk->queue == q)
672fdcf0 ML	257	return blkg;
672fdcf0 ML	258
9a9c261e CH	259	blkg = radix_tree_lookup(&blkcg->blkg_tree, q->id);
9a9c261e CH	260	if (blkg && blkg->disk->queue != q)
4a69f325 CH	261	blkg = NULL;
4a69f325 CH	262	return blkg;
672fdcf0 ML	263	}
672fdcf0 ML	264
672fdcf0 ML	265	/**
	266	* blkg_to_pdata - get policy private data
	267	* @blkg: blkg of interest
	268	* @pol: policy of interest
	269	*
	270	* Return pointer to private data associated with the @blkg-@pol pair.
	271	*/
	272	static inline struct blkg_policy_data blkg_to_pd(struct blkcg_gq blkg,
	273	struct blkcg_policy *pol)
	274	{
	275	return blkg ? blkg->pd[pol->plid] : NULL;
	276	}
	277
	278	static inline struct blkcg_policy_data blkcg_to_cpd(struct blkcg blkcg,
	279	struct blkcg_policy *pol)
	280	{
	281	return blkcg ? blkcg->cpd[pol->plid] : NULL;
	282	}
	283
	284	/**
	285	* pdata_to_blkg - get blkg associated with policy private data
	286	* @pd: policy private data of interest
	287	*
	288	* @pd is policy private data. Determine the blkg it's associated with.
	289	*/
	290	static inline struct blkcg_gq pd_to_blkg(struct blkg_policy_data pd)
	291	{
	292	return pd ? pd->blkg : NULL;
	293	}
	294
	295	static inline struct blkcg cpd_to_blkcg(struct blkcg_policy_data cpd)
	296	{
	297	return cpd ? cpd->blkcg : NULL;
	298	}
	299
	300	/**
	301	* blkg_path - format cgroup path of blkg
	302	* @blkg: blkg of interest
	303	* @buf: target buffer
	304	* @buflen: target buffer length
	305	*
	306	* Format the path of the cgroup of @blkg into @buf.
	307	*/
	308	static inline int blkg_path(struct blkcg_gq blkg, char buf, int buflen)
	309	{
	310	return cgroup_path(blkg->blkcg->css.cgroup, buf, buflen);
	311	}
	312
	313	/**
	314	* blkg_get - get a blkg reference
	315	* @blkg: blkg to get
	316	*
	317	* The caller should be holding an existing reference.
	318	*/
	319	static inline void blkg_get(struct blkcg_gq *blkg)
	320	{
	321	percpu_ref_get(&blkg->refcnt);
	322	}
	323
	324	/**
	325	* blkg_tryget - try and get a blkg reference
	326	* @blkg: blkg to get
	327	*
	328	* This is for use when doing an RCU lookup of the blkg. We may be in the midst
329	* of freeing this blkg, so we can only use it if the refcnt is not zero.
330	*/
331	static inline bool blkg_tryget(struct blkcg_gq *blkg)
332	{
333	return blkg && percpu_ref_tryget(&blkg->refcnt);
334	}
335
336	/**
337	* blkg_put - put a blkg reference
338	* @blkg: blkg to put
339	*/
340	static inline void blkg_put(struct blkcg_gq *blkg)
341	{
342	percpu_ref_put(&blkg->refcnt);
343	}
344
345	/**
346	* blkg_for_each_descendant_pre - pre-order walk of a blkg's descendants
347	* @d_blkg: loop cursor pointing to the current descendant
348	* @pos_css: used for iteration
349	* @p_blkg: target blkg to walk descendants of
350	*
351	* Walk @c_blkg through the descendants of @p_blkg. Must be used with RCU
352	* read locked. If called under either blkcg or queue lock, the iteration
353	* is guaranteed to include all and only online blkgs. The caller may
354	* update @pos_css by calling css_rightmost_descendant() to skip subtree.
355	* @p_blkg is included in the iteration and the first node to be visited.
356	*/
357	#define blkg_for_each_descendant_pre(d_blkg, pos_css, p_blkg) \
358	css_for_each_descendant_pre((pos_css), &(p_blkg)->blkcg->css) \
79fcc5be	359	if (((d_blkg) = blkg_lookup(css_to_blkcg(pos_css), \
9a9c261e	360	(p_blkg)->disk->queue)))
672fdcf0 ML	361
	362	/**
	363	* blkg_for_each_descendant_post - post-order walk of a blkg's descendants
	364	* @d_blkg: loop cursor pointing to the current descendant
	365	* @pos_css: used for iteration
	366	* @p_blkg: target blkg to walk descendants of
	367	*
	368	* Similar to blkg_for_each_descendant_pre() but performs post-order
	369	* traversal instead. Synchronization rules are the same. @p_blkg is
	370	* included in the iteration and the last node to be visited.
	371	*/
	372	#define blkg_for_each_descendant_post(d_blkg, pos_css, p_blkg) \
	373	css_for_each_descendant_post((pos_css), &(p_blkg)->blkcg->css) \
79fcc5be	374	if (((d_blkg) = blkg_lookup(css_to_blkcg(pos_css), \
9a9c261e	375	(p_blkg)->disk->queue)))
672fdcf0 ML	376
	377	bool __blkcg_punt_bio_submit(struct bio *bio);
	378
	379	static inline bool blkcg_punt_bio_submit(struct bio *bio)
	380	{
	381	if (bio->bi_opf & REQ_CGROUP_PUNT)
	382	return __blkcg_punt_bio_submit(bio);
	383	else
	384	return false;
	385	}
	386
	387	static inline void blkcg_bio_issue_init(struct bio *bio)
	388	{
	389	bio_issue_init(&bio->bi_issue, bio_sectors(bio));
	390	}
	391
	392	static inline void blkcg_use_delay(struct blkcg_gq *blkg)
	393	{
	394	if (WARN_ON_ONCE(atomic_read(&blkg->use_delay) < 0))
	395	return;
	396	if (atomic_add_return(1, &blkg->use_delay) == 1)
	397	atomic_inc(&blkg->blkcg->css.cgroup->congestion_count);
	398	}
	399
	400	static inline int blkcg_unuse_delay(struct blkcg_gq *blkg)
	401	{
	402	int old = atomic_read(&blkg->use_delay);
	403
	404	if (WARN_ON_ONCE(old < 0))
	405	return 0;
	406	if (old == 0)
	407	return 0;
	408
	409	/*
	410	* We do this song and dance because we can race with somebody else
	411	* adding or removing delay. If we just did an atomic_dec we'd end up
	412	* negative and we'd already be in trouble. We need to subtract 1 and
	413	* then check to see if we were the last delay so we can drop the
	414	* congestion count on the cgroup.
	415	*/
96388f57 UB	416	while (old && !atomic_try_cmpxchg(&blkg->use_delay, &old, old - 1))
96388f57 UB	417	;
672fdcf0 ML	418
	419	if (old == 0)
	420	return 0;
	421	if (old == 1)
	422	atomic_dec(&blkg->blkcg->css.cgroup->congestion_count);
	423	return 1;
	424	}
	425
	426	/**
	427	* blkcg_set_delay - Enable allocator delay mechanism with the specified delay amount
	428	* @blkg: target blkg
	429	* @delay: delay duration in nsecs
	430	*
	431	* When enabled with this function, the delay is not decayed and must be
	432	* explicitly cleared with blkcg_clear_delay(). Must not be mixed with
	433	* blkcg_[un]use_delay() and blkcg_add_delay() usages.
	434	*/
	435	static inline void blkcg_set_delay(struct blkcg_gq *blkg, u64 delay)
	436	{
	437	int old = atomic_read(&blkg->use_delay);
	438
	439	/* We only want 1 person setting the congestion count for this blkg. */
96388f57	440	if (!old && atomic_try_cmpxchg(&blkg->use_delay, &old, -1))
672fdcf0 ML	441	atomic_inc(&blkg->blkcg->css.cgroup->congestion_count);
	442
	443	atomic64_set(&blkg->delay_nsec, delay);
	444	}
	445
	446	/**
	447	* blkcg_clear_delay - Disable allocator delay mechanism
	448	* @blkg: target blkg
	449	*
	450	* Disable use_delay mechanism. See blkcg_set_delay().
	451	*/
	452	static inline void blkcg_clear_delay(struct blkcg_gq *blkg)
	453	{
	454	int old = atomic_read(&blkg->use_delay);
	455
	456	/* We only want 1 person clearing the congestion count for this blkg. */
96388f57	457	if (old && atomic_try_cmpxchg(&blkg->use_delay, &old, 0))
672fdcf0 ML	458	atomic_dec(&blkg->blkcg->css.cgroup->congestion_count);
	459	}
	460
6b2b0459 TH	461	/**
	462	* blk_cgroup_mergeable - Determine whether to allow or disallow merges
	463	* @rq: request to merge into
	464	* @bio: bio to merge
	465	*
	466	* @bio and @rq should belong to the same cgroup and their issue_as_root should
	467	* match. The latter is necessary as we don't want to throttle e.g. a metadata
	468	* update because it happens to be next to a regular IO.
	469	*/
	470	static inline bool blk_cgroup_mergeable(struct request rq, struct bio bio)
	471	{
	472	return rq->bio->bi_blkg == bio->bi_blkg &&
	473	bio_issue_as_root_blkg(rq->bio) == bio_issue_as_root_blkg(bio);
	474	}
	475
672fdcf0 ML	476	void blk_cgroup_bio_start(struct bio *bio);
	477	void blkcg_add_delay(struct blkcg_gq *blkg, u64 now, u64 delta);
	478	#else /* CONFIG_BLK_CGROUP */
	479
	480	struct blkg_policy_data {
	481	};
	482
	483	struct blkcg_policy_data {
	484	};
	485
	486	struct blkcg_policy {
	487	};
	488
bbb1ebe7 CH	489	struct blkcg {
	490	};
	491
672fdcf0	492	static inline struct blkcg_gq blkg_lookup(struct blkcg blkcg, void *key) { return NULL; }
9823538f CH	493	static inline int blkcg_init_disk(struct gendisk *disk) { return 0; }
9823538f CH	494	static inline void blkcg_exit_disk(struct gendisk *disk) { }
672fdcf0 ML	495	static inline int blkcg_policy_register(struct blkcg_policy *pol) { return 0; }
672fdcf0 ML	496	static inline void blkcg_policy_unregister(struct blkcg_policy *pol) { }
40e4996e	497	static inline int blkcg_activate_policy(struct gendisk *disk,
672fdcf0	498	const struct blkcg_policy *pol) { return 0; }
40e4996e	499	static inline void blkcg_deactivate_policy(struct gendisk *disk,
672fdcf0 ML	500	const struct blkcg_policy *pol) { }
672fdcf0 ML	501
672fdcf0 ML	502	static inline struct blkg_policy_data blkg_to_pd(struct blkcg_gq blkg,
	503	struct blkcg_policy *pol) { return NULL; }
	504	static inline struct blkcg_gq pd_to_blkg(struct blkg_policy_data pd) { return NULL; }
	505	static inline char blkg_path(struct blkcg_gq blkg) { return NULL; }
	506	static inline void blkg_get(struct blkcg_gq *blkg) { }
	507	static inline void blkg_put(struct blkcg_gq *blkg) { }
	508
	509	static inline bool blkcg_punt_bio_submit(struct bio *bio) { return false; }
	510	static inline void blkcg_bio_issue_init(struct bio *bio) { }
	511	static inline void blk_cgroup_bio_start(struct bio *bio) { }
6b2b0459	512	static inline bool blk_cgroup_mergeable(struct request rq, struct bio bio) { return true; }
672fdcf0 ML	513
	514	#define blk_queue_for_each_rl(rl, q) \
	515	for ((rl) = &(q)->root_rl; (rl); (rl) = NULL)
	516
672fdcf0 ML	517	#endif /* CONFIG_BLK_CGROUP */
	518
	519	#endif /* _BLK_CGROUP_PRIVATE_H */