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

/* SPDX-License-Identifier: GPL-2.0 */
#ifndef INT_BLK_MQ_H
#define INT_BLK_MQ_H

#include <linux/blk-mq.h>
#include "blk-stat.h"

struct blk_mq_tag_set;

struct blk_mq_ctxs {
	struct kobject kobj;
	struct blk_mq_ctx __percpu	*queue_ctx;
};

/**
 * struct blk_mq_ctx - State for a software queue facing the submitting CPUs
 */
struct blk_mq_ctx {
	struct {
		spinlock_t		lock;
		struct list_head	rq_lists[HCTX_MAX_TYPES];
	} ____cacheline_aligned_in_smp;

	unsigned int		cpu;
	unsigned short		index_hw[HCTX_MAX_TYPES];
	struct blk_mq_hw_ctx 	*hctxs[HCTX_MAX_TYPES];

	struct request_queue	*queue;
	struct blk_mq_ctxs      *ctxs;
	struct kobject		kobj;
} ____cacheline_aligned_in_smp;

enum {
	BLK_MQ_NO_TAG		= -1U,
	BLK_MQ_TAG_MIN		= 1,
	BLK_MQ_TAG_MAX		= BLK_MQ_NO_TAG - 1,
};

typedef unsigned int __bitwise blk_insert_t;
#define BLK_MQ_INSERT_AT_HEAD		((__force blk_insert_t)0x01)

void blk_mq_submit_bio(struct bio *bio);
int blk_mq_poll(struct request_queue *q, blk_qc_t cookie, struct io_comp_batch *iob,
		unsigned int flags);
void blk_mq_exit_queue(struct request_queue *q);
int blk_mq_update_nr_requests(struct request_queue *q, unsigned int nr);
void blk_mq_wake_waiters(struct request_queue *q);
bool blk_mq_dispatch_rq_list(struct blk_mq_hw_ctx *hctx, struct list_head *,
			     unsigned int);
void blk_mq_flush_busy_ctxs(struct blk_mq_hw_ctx *hctx, struct list_head *list);
struct request *blk_mq_dequeue_from_ctx(struct blk_mq_hw_ctx *hctx,
					struct blk_mq_ctx *start);
void blk_mq_put_rq_ref(struct request *rq);

/*
 * Internal helpers for allocating/freeing the request map
 */
void blk_mq_free_rqs(struct blk_mq_tag_set *set, struct blk_mq_tags *tags,
		     unsigned int hctx_idx);
void blk_mq_free_rq_map(struct blk_mq_tags *tags);
struct blk_mq_tags *blk_mq_alloc_map_and_rqs(struct blk_mq_tag_set *set,
				unsigned int hctx_idx, unsigned int depth);
void blk_mq_free_map_and_rqs(struct blk_mq_tag_set *set,
			     struct blk_mq_tags *tags,
			     unsigned int hctx_idx);

/*
 * CPU -> queue mappings
 */
extern int blk_mq_hw_queue_to_node(struct blk_mq_queue_map *qmap, unsigned int);

/*
 * blk_mq_map_queue_type() - map (hctx_type,cpu) to hardware queue
 * @q: request queue
 * @type: the hctx type index
 * @cpu: CPU
 */
static inline struct blk_mq_hw_ctx *blk_mq_map_queue_type(struct request_queue *q,
							  enum hctx_type type,
							  unsigned int cpu)
{
	return xa_load(&q->hctx_table, q->tag_set->map[type].mq_map[cpu]);
}

static inline enum hctx_type blk_mq_get_hctx_type(blk_opf_t opf)
{
	enum hctx_type type = HCTX_TYPE_DEFAULT;

	/*
	 * The caller ensure that if REQ_POLLED, poll must be enabled.
	 */
	if (opf & REQ_POLLED)
		type = HCTX_TYPE_POLL;
	else if ((opf & REQ_OP_MASK) == REQ_OP_READ)
		type = HCTX_TYPE_READ;
	return type;
}

/*
 * blk_mq_map_queue() - map (cmd_flags,type) to hardware queue
 * @q: request queue
 * @opf: operation type (REQ_OP_*) and flags (e.g. REQ_POLLED).
 * @ctx: software queue cpu ctx
 */
static inline struct blk_mq_hw_ctx *blk_mq_map_queue(struct request_queue *q,
						     blk_opf_t opf,
						     struct blk_mq_ctx *ctx)
{
	return ctx->hctxs[blk_mq_get_hctx_type(opf)];
}

/*
 * sysfs helpers
 */
extern void blk_mq_sysfs_init(struct request_queue *q);
extern void blk_mq_sysfs_deinit(struct request_queue *q);
int blk_mq_sysfs_register(struct gendisk *disk);
void blk_mq_sysfs_unregister(struct gendisk *disk);
int blk_mq_sysfs_register_hctxs(struct request_queue *q);
void blk_mq_sysfs_unregister_hctxs(struct request_queue *q);
extern void blk_mq_hctx_kobj_init(struct blk_mq_hw_ctx *hctx);
void blk_mq_free_plug_rqs(struct blk_plug *plug);
void blk_mq_flush_plug_list(struct blk_plug *plug, bool from_schedule);

void blk_mq_cancel_work_sync(struct request_queue *q);

void blk_mq_release(struct request_queue *q);

static inline struct blk_mq_ctx *__blk_mq_get_ctx(struct request_queue *q,
					   unsigned int cpu)
{
	return per_cpu_ptr(q->queue_ctx, cpu);
}

/*
 * This assumes per-cpu software queueing queues. They could be per-node
 * as well, for instance. For now this is hardcoded as-is. Note that we don't
 * care about preemption, since we know the ctx's are persistent. This does
 * mean that we can't rely on ctx always matching the currently running CPU.
 */
static inline struct blk_mq_ctx *blk_mq_get_ctx(struct request_queue *q)
{
	return __blk_mq_get_ctx(q, raw_smp_processor_id());
}

struct blk_mq_alloc_data {
	/* input parameter */
	struct request_queue *q;
	blk_mq_req_flags_t flags;
	unsigned int shallow_depth;
	blk_opf_t cmd_flags;
	req_flags_t rq_flags;

	/* allocate multiple requests/tags in one go */
	unsigned int nr_tags;
	struct request **cached_rq;

	/* input & output parameter */
	struct blk_mq_ctx *ctx;
	struct blk_mq_hw_ctx *hctx;
};

struct blk_mq_tags *blk_mq_init_tags(unsigned int nr_tags,
		unsigned int reserved_tags, int node, int alloc_policy);
void blk_mq_free_tags(struct blk_mq_tags *tags);
int blk_mq_init_bitmaps(struct sbitmap_queue *bitmap_tags,
		struct sbitmap_queue *breserved_tags, unsigned int queue_depth,
		unsigned int reserved, int node, int alloc_policy);

unsigned int blk_mq_get_tag(struct blk_mq_alloc_data *data);
unsigned long blk_mq_get_tags(struct blk_mq_alloc_data *data, int nr_tags,
		unsigned int *offset);
void blk_mq_put_tag(struct blk_mq_tags *tags, struct blk_mq_ctx *ctx,
		unsigned int tag);
void blk_mq_put_tags(struct blk_mq_tags *tags, int *tag_array, int nr_tags);
int blk_mq_tag_update_depth(struct blk_mq_hw_ctx *hctx,
		struct blk_mq_tags **tags, unsigned int depth, bool can_grow);
void blk_mq_tag_resize_shared_tags(struct blk_mq_tag_set *set,
		unsigned int size);
void blk_mq_tag_update_sched_shared_tags(struct request_queue *q);

void blk_mq_tag_wakeup_all(struct blk_mq_tags *tags, bool);
void blk_mq_queue_tag_busy_iter(struct request_queue *q, busy_tag_iter_fn *fn,
		void *priv);
void blk_mq_all_tag_iter(struct blk_mq_tags *tags, busy_tag_iter_fn *fn,
		void *priv);

static inline struct sbq_wait_state *bt_wait_ptr(struct sbitmap_queue *bt,
						 struct blk_mq_hw_ctx *hctx)
{
	if (!hctx)
		return &bt->ws[0];
	return sbq_wait_ptr(bt, &hctx->wait_index);
}

void __blk_mq_tag_busy(struct blk_mq_hw_ctx *);
void __blk_mq_tag_idle(struct blk_mq_hw_ctx *);

static inline void blk_mq_tag_busy(struct blk_mq_hw_ctx *hctx)
{
	if (hctx->flags & BLK_MQ_F_TAG_QUEUE_SHARED)
		__blk_mq_tag_busy(hctx);
}

static inline void blk_mq_tag_idle(struct blk_mq_hw_ctx *hctx)
{
	if (hctx->flags & BLK_MQ_F_TAG_QUEUE_SHARED)
		__blk_mq_tag_idle(hctx);
}

static inline bool blk_mq_tag_is_reserved(struct blk_mq_tags *tags,
					  unsigned int tag)
{
	return tag < tags->nr_reserved_tags;
}

static inline bool blk_mq_is_shared_tags(unsigned int flags)
{
	return flags & BLK_MQ_F_TAG_HCTX_SHARED;
}

static inline struct blk_mq_tags *blk_mq_tags_from_data(struct blk_mq_alloc_data *data)
{
	if (data->rq_flags & RQF_SCHED_TAGS)
		return data->hctx->sched_tags;
	return data->hctx->tags;
}

static inline bool blk_mq_hctx_stopped(struct blk_mq_hw_ctx *hctx)
{
	return test_bit(BLK_MQ_S_STOPPED, &hctx->state);
}

static inline bool blk_mq_hw_queue_mapped(struct blk_mq_hw_ctx *hctx)
{
	return hctx->nr_ctx && hctx->tags;
}

unsigned int blk_mq_in_flight(struct request_queue *q,
		struct block_device *part);
void blk_mq_in_flight_rw(struct request_queue *q, struct block_device *part,
		unsigned int inflight[2]);

static inline void blk_mq_put_dispatch_budget(struct request_queue *q,
					      int budget_token)
{
	if (q->mq_ops->put_budget)
		q->mq_ops->put_budget(q, budget_token);
}

static inline int blk_mq_get_dispatch_budget(struct request_queue *q)
{
	if (q->mq_ops->get_budget)
		return q->mq_ops->get_budget(q);
	return 0;
}

static inline void blk_mq_set_rq_budget_token(struct request *rq, int token)
{
	if (token < 0)
		return;

	if (rq->q->mq_ops->set_rq_budget_token)
		rq->q->mq_ops->set_rq_budget_token(rq, token);
}

static inline int blk_mq_get_rq_budget_token(struct request *rq)
{
	if (rq->q->mq_ops->get_rq_budget_token)
		return rq->q->mq_ops->get_rq_budget_token(rq);
	return -1;
}

static inline void __blk_mq_inc_active_requests(struct blk_mq_hw_ctx *hctx)
{
	if (blk_mq_is_shared_tags(hctx->flags))
		atomic_inc(&hctx->queue->nr_active_requests_shared_tags);
	else
		atomic_inc(&hctx->nr_active);
}

static inline void __blk_mq_sub_active_requests(struct blk_mq_hw_ctx *hctx,
		int val)
{
	if (blk_mq_is_shared_tags(hctx->flags))
		atomic_sub(val, &hctx->queue->nr_active_requests_shared_tags);
	else
		atomic_sub(val, &hctx->nr_active);
}

static inline void __blk_mq_dec_active_requests(struct blk_mq_hw_ctx *hctx)
{
	__blk_mq_sub_active_requests(hctx, 1);
}

static inline int __blk_mq_active_requests(struct blk_mq_hw_ctx *hctx)
{
	if (blk_mq_is_shared_tags(hctx->flags))
		return atomic_read(&hctx->queue->nr_active_requests_shared_tags);
	return atomic_read(&hctx->nr_active);
}
static inline void __blk_mq_put_driver_tag(struct blk_mq_hw_ctx *hctx,
					   struct request *rq)
{
	blk_mq_put_tag(hctx->tags, rq->mq_ctx, rq->tag);
	rq->tag = BLK_MQ_NO_TAG;

	if (rq->rq_flags & RQF_MQ_INFLIGHT) {
		rq->rq_flags &= ~RQF_MQ_INFLIGHT;
		__blk_mq_dec_active_requests(hctx);
	}
}

static inline void blk_mq_put_driver_tag(struct request *rq)
{
	if (rq->tag == BLK_MQ_NO_TAG || rq->internal_tag == BLK_MQ_NO_TAG)
		return;

	__blk_mq_put_driver_tag(rq->mq_hctx, rq);
}

bool __blk_mq_get_driver_tag(struct blk_mq_hw_ctx *hctx, struct request *rq);

static inline bool blk_mq_get_driver_tag(struct request *rq)
{
	struct blk_mq_hw_ctx *hctx = rq->mq_hctx;

	if (rq->tag != BLK_MQ_NO_TAG &&
	    !(hctx->flags & BLK_MQ_F_TAG_QUEUE_SHARED)) {
		hctx->tags->rqs[rq->tag] = rq;
		return true;
	}

	return __blk_mq_get_driver_tag(hctx, rq);
}

static inline void blk_mq_clear_mq_map(struct blk_mq_queue_map *qmap)
{
	int cpu;

	for_each_possible_cpu(cpu)
		qmap->mq_map[cpu] = 0;
}

/*
 * blk_mq_plug() - Get caller context plug
 * @bio : the bio being submitted by the caller context
 *
 * Plugging, by design, may delay the insertion of BIOs into the elevator in
 * order to increase BIO merging opportunities. This however can cause BIO
 * insertion order to change from the order in which submit_bio() is being
 * executed in the case of multiple contexts concurrently issuing BIOs to a
 * device, even if these context are synchronized to tightly control BIO issuing
 * order. While this is not a problem with regular block devices, this ordering
 * change can cause write BIO failures with zoned block devices as these
 * require sequential write patterns to zones. Prevent this from happening by
 * ignoring the plug state of a BIO issuing context if it is for a zoned block
 * device and the BIO to plug is a write operation.
 *
 * Return current->plug if the bio can be plugged and NULL otherwise
 */
static inline struct blk_plug *blk_mq_plug( struct bio *bio)
{
	/* Zoned block device write operation case: do not plug the BIO */
	if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) &&
	    bdev_op_is_zoned_write(bio->bi_bdev, bio_op(bio)))
		return NULL;

	/*
	 * For regular block devices or read operations, use the context plug
	 * which may be NULL if blk_start_plug() was not executed.
	 */
	return current->plug;
}

/* Free all requests on the list */
static inline void blk_mq_free_requests(struct list_head *list)
{
	while (!list_empty(list)) {
		struct request *rq = list_entry_rq(list->next);

		list_del_init(&rq->queuelist);
		blk_mq_free_request(rq);
	}
}

/*
 * For shared tag users, we track the number of currently active users
 * and attempt to provide a fair share of the tag depth for each of them.
 */
static inline bool hctx_may_queue(struct blk_mq_hw_ctx *hctx,
				  struct sbitmap_queue *bt)
{
	unsigned int depth, users;

	if (!hctx || !(hctx->flags & BLK_MQ_F_TAG_QUEUE_SHARED))
		return true;

	/*
	 * Don't try dividing an ant
	 */
	if (bt->sb.depth == 1)
		return true;

	if (blk_mq_is_shared_tags(hctx->flags)) {
		struct request_queue *q = hctx->queue;

		if (!test_bit(QUEUE_FLAG_HCTX_ACTIVE, &q->queue_flags))
			return true;
	} else {
		if (!test_bit(BLK_MQ_S_TAG_ACTIVE, &hctx->state))
			return true;
	}

	users = READ_ONCE(hctx->tags->active_queues);
	if (!users)
		return true;

	/*
	 * Allow at least some tags
	 */
	depth = max((bt->sb.depth + users - 1) / users, 4U);
	return __blk_mq_active_requests(hctx) < depth;
}

/* run the code block in @dispatch_ops with rcu/srcu read lock held */
#define __blk_mq_run_dispatch_ops(q, check_sleep, dispatch_ops)	\
do {								\
	if ((q)->tag_set->flags & BLK_MQ_F_BLOCKING) {		\
		struct blk_mq_tag_set *__tag_set = (q)->tag_set; \
		int srcu_idx;					\
								\
		might_sleep_if(check_sleep);			\
		srcu_idx = srcu_read_lock(__tag_set->srcu);	\
		(dispatch_ops);					\
		srcu_read_unlock(__tag_set->srcu, srcu_idx);	\
	} else {						\
		rcu_read_lock();				\
		(dispatch_ops);					\
		rcu_read_unlock();				\
	}							\
} while (0)

#define blk_mq_run_dispatch_ops(q, dispatch_ops)		\
	__blk_mq_run_dispatch_ops(q, true, dispatch_ops)	\

#endif
Commit	Line	Data
b2441318	1	/* SPDX-License-Identifier: GPL-2.0 */
320ae51f JA	2	#ifndef INT_BLK_MQ_H
	3	#define INT_BLK_MQ_H
	4
90110e04	5	#include <linux/blk-mq.h>
cf43e6be JA	6	#include "blk-stat.h"
cf43e6be JA	7
24d2f903 CH	8	struct blk_mq_tag_set;
24d2f903 CH	9
1db4909e ML	10	struct blk_mq_ctxs {
	11	struct kobject kobj;
	12	struct blk_mq_ctx __percpu *queue_ctx;
	13	};
	14
fe644072 LW	15	/**
	16	* struct blk_mq_ctx - State for a software queue facing the submitting CPUs
	17	*/
320ae51f JA	18	struct blk_mq_ctx {
	19	struct {
	20	spinlock_t lock;
c16d6b5a ML	21	struct list_head rq_lists[HCTX_MAX_TYPES];
c16d6b5a ML	22	} ____cacheline_aligned_in_smp;
320ae51f JA	23
320ae51f JA	24	unsigned int cpu;
f31967f0	25	unsigned short index_hw[HCTX_MAX_TYPES];
8ccdf4a3	26	struct blk_mq_hw_ctx *hctxs[HCTX_MAX_TYPES];
320ae51f	27
320ae51f	28	struct request_queue *queue;
1db4909e	29	struct blk_mq_ctxs *ctxs;
320ae51f	30	struct kobject kobj;
4bb659b1	31	} ____cacheline_aligned_in_smp;
320ae51f	32
bebe84eb CH	33	enum {
	34	BLK_MQ_NO_TAG = -1U,
	35	BLK_MQ_TAG_MIN = 1,
	36	BLK_MQ_TAG_MAX = BLK_MQ_NO_TAG - 1,
	37	};
	38
710fa378 CH	39	typedef unsigned int __bitwise blk_insert_t;
	40	#define BLK_MQ_INSERT_AT_HEAD ((__force blk_insert_t)0x01)
	41
3e08773c	42	void blk_mq_submit_bio(struct bio *bio);
5a72e899 JA	43	int blk_mq_poll(struct request_queue q, blk_qc_t cookie, struct io_comp_batch iob,
5a72e899 JA	44	unsigned int flags);
c7e2d94b	45	void blk_mq_exit_queue(struct request_queue *q);
e3a2b3f9	46	int blk_mq_update_nr_requests(struct request_queue *q, unsigned int nr);
aed3ea94	47	void blk_mq_wake_waiters(struct request_queue *q);
1fd40b5e ML	48	bool blk_mq_dispatch_rq_list(struct blk_mq_hw_ctx hctx, struct list_head ,
1fd40b5e ML	49	unsigned int);
2c3ad667	50	void blk_mq_flush_busy_ctxs(struct blk_mq_hw_ctx hctx, struct list_head list);
b347689f ML	51	struct request blk_mq_dequeue_from_ctx(struct blk_mq_hw_ctx hctx,
b347689f ML	52	struct blk_mq_ctx *start);
2e315dc0	53	void blk_mq_put_rq_ref(struct request *rq);
2c3ad667 JA	54
	55	/*
	56	* Internal helpers for allocating/freeing the request map
	57	*/
cc71a6f4 JA	58	void blk_mq_free_rqs(struct blk_mq_tag_set set, struct blk_mq_tags tags,
cc71a6f4 JA	59	unsigned int hctx_idx);
e155b0c2	60	void blk_mq_free_rq_map(struct blk_mq_tags *tags);
63064be1 JG	61	struct blk_mq_tags blk_mq_alloc_map_and_rqs(struct blk_mq_tag_set set,
63064be1 JG	62	unsigned int hctx_idx, unsigned int depth);
645db34e JG	63	void blk_mq_free_map_and_rqs(struct blk_mq_tag_set *set,
	64	struct blk_mq_tags *tags,
	65	unsigned int hctx_idx);
396eaf21	66
320ae51f JA	67	/*
	68	* CPU -> queue mappings
	69	*/
ed76e329	70	extern int blk_mq_hw_queue_to_node(struct blk_mq_queue_map *qmap, unsigned int);
320ae51f	71
b3c661b1 JA	72	/*
	73	* blk_mq_map_queue_type() - map (hctx_type,cpu) to hardware queue
	74	* @q: request queue
e20ba6e1	75	* @type: the hctx type index
b3c661b1 JA	76	* @cpu: CPU
	77	*/
	78	static inline struct blk_mq_hw_ctx blk_mq_map_queue_type(struct request_queue q,
e20ba6e1	79	enum hctx_type type,
b3c661b1	80	unsigned int cpu)
7d7e0f90	81	{
4e5cc99e	82	return xa_load(&q->hctx_table, q->tag_set->map[type].mq_map[cpu]);
7d7e0f90 CH	83	}
7d7e0f90 CH	84
16458cf3	85	static inline enum hctx_type blk_mq_get_hctx_type(blk_opf_t opf)
ff2c5660	86	{
e20ba6e1 CH	87	enum hctx_type type = HCTX_TYPE_DEFAULT;
e20ba6e1 CH	88
bb94aea1	89	/*
6ce913fe	90	* The caller ensure that if REQ_POLLED, poll must be enabled.
bb94aea1	91	*/
7e923f40	92	if (opf & REQ_POLLED)
e20ba6e1	93	type = HCTX_TYPE_POLL;
7e923f40	94	else if ((opf & REQ_OP_MASK) == REQ_OP_READ)
e20ba6e1	95	type = HCTX_TYPE_READ;
b637108a ML	96	return type;
	97	}
	98
	99	/*
	100	* blk_mq_map_queue() - map (cmd_flags,type) to hardware queue
	101	* @q: request queue
7e923f40	102	* @opf: operation type (REQ_OP_*) and flags (e.g. REQ_POLLED).
b637108a ML	103	* @ctx: software queue cpu ctx
	104	*/
	105	static inline struct blk_mq_hw_ctx blk_mq_map_queue(struct request_queue q,
16458cf3	106	blk_opf_t opf,
b637108a ML	107	struct blk_mq_ctx *ctx)
b637108a ML	108	{
7e923f40	109	return ctx->hctxs[blk_mq_get_hctx_type(opf)];
ff2c5660 JA	110	}
ff2c5660 JA	111
67aec14c JA	112	/*
	113	* sysfs helpers
	114	*/
737f98cf	115	extern void blk_mq_sysfs_init(struct request_queue *q);
7ea5fe31	116	extern void blk_mq_sysfs_deinit(struct request_queue *q);
8682b92e CH	117	int blk_mq_sysfs_register(struct gendisk *disk);
8682b92e CH	118	void blk_mq_sysfs_unregister(struct gendisk *disk);
eaa870f9 CH	119	int blk_mq_sysfs_register_hctxs(struct request_queue *q);
eaa870f9 CH	120	void blk_mq_sysfs_unregister_hctxs(struct request_queue *q);
868f2f0b	121	extern void blk_mq_hctx_kobj_init(struct blk_mq_hw_ctx *hctx);
47c122e3	122	void blk_mq_free_plug_rqs(struct blk_plug *plug);
dbb6f764	123	void blk_mq_flush_plug_list(struct blk_plug *plug, bool from_schedule);
67aec14c	124
2a19b28f ML	125	void blk_mq_cancel_work_sync(struct request_queue *q);
2a19b28f ML	126
e09aae7e ML	127	void blk_mq_release(struct request_queue *q);
e09aae7e ML	128
1aecfe48 ML	129	static inline struct blk_mq_ctx __blk_mq_get_ctx(struct request_queue q,
	130	unsigned int cpu)
	131	{
	132	return per_cpu_ptr(q->queue_ctx, cpu);
	133	}
	134
	135	/*
	136	* This assumes per-cpu software queueing queues. They could be per-node
	137	* as well, for instance. For now this is hardcoded as-is. Note that we don't
	138	* care about preemption, since we know the ctx's are persistent. This does
	139	* mean that we can't rely on ctx always matching the currently running CPU.
	140	*/
	141	static inline struct blk_mq_ctx blk_mq_get_ctx(struct request_queue q)
	142	{
c05f4220	143	return __blk_mq_get_ctx(q, raw_smp_processor_id());
1aecfe48 ML	144	}
1aecfe48 ML	145
cb96a42c ML	146	struct blk_mq_alloc_data {
	147	/* input parameter */
	148	struct request_queue *q;
9a95e4ef	149	blk_mq_req_flags_t flags;
229a9287	150	unsigned int shallow_depth;
16458cf3	151	blk_opf_t cmd_flags;
ecaf97f4	152	req_flags_t rq_flags;
cb96a42c	153
47c122e3 JA	154	/* allocate multiple requests/tags in one go */
	155	unsigned int nr_tags;
	156	struct request **cached_rq;
	157
cb96a42c ML	158	/* input & output parameter */
	159	struct blk_mq_ctx *ctx;
	160	struct blk_mq_hw_ctx *hctx;
	161	};
	162
bebe84eb CH	163	struct blk_mq_tags *blk_mq_init_tags(unsigned int nr_tags,
	164	unsigned int reserved_tags, int node, int alloc_policy);
	165	void blk_mq_free_tags(struct blk_mq_tags *tags);
	166	int blk_mq_init_bitmaps(struct sbitmap_queue *bitmap_tags,
	167	struct sbitmap_queue *breserved_tags, unsigned int queue_depth,
	168	unsigned int reserved, int node, int alloc_policy);
	169
	170	unsigned int blk_mq_get_tag(struct blk_mq_alloc_data *data);
	171	unsigned long blk_mq_get_tags(struct blk_mq_alloc_data *data, int nr_tags,
	172	unsigned int *offset);
	173	void blk_mq_put_tag(struct blk_mq_tags tags, struct blk_mq_ctx ctx,
	174	unsigned int tag);
	175	void blk_mq_put_tags(struct blk_mq_tags tags, int tag_array, int nr_tags);
	176	int blk_mq_tag_update_depth(struct blk_mq_hw_ctx *hctx,
	177	struct blk_mq_tags **tags, unsigned int depth, bool can_grow);
	178	void blk_mq_tag_resize_shared_tags(struct blk_mq_tag_set *set,
	179	unsigned int size);
	180	void blk_mq_tag_update_sched_shared_tags(struct request_queue *q);
	181
	182	void blk_mq_tag_wakeup_all(struct blk_mq_tags *tags, bool);
	183	void blk_mq_queue_tag_busy_iter(struct request_queue q, busy_tag_iter_fn fn,
	184	void *priv);
	185	void blk_mq_all_tag_iter(struct blk_mq_tags tags, busy_tag_iter_fn fn,
	186	void *priv);
	187
	188	static inline struct sbq_wait_state bt_wait_ptr(struct sbitmap_queue bt,
	189	struct blk_mq_hw_ctx *hctx)
	190	{
	191	if (!hctx)
	192	return &bt->ws[0];
	193	return sbq_wait_ptr(bt, &hctx->wait_index);
	194	}
	195
	196	void __blk_mq_tag_busy(struct blk_mq_hw_ctx *);
	197	void __blk_mq_tag_idle(struct blk_mq_hw_ctx *);
	198
	199	static inline void blk_mq_tag_busy(struct blk_mq_hw_ctx *hctx)
	200	{
	201	if (hctx->flags & BLK_MQ_F_TAG_QUEUE_SHARED)
	202	__blk_mq_tag_busy(hctx);
	203	}
	204
	205	static inline void blk_mq_tag_idle(struct blk_mq_hw_ctx *hctx)
	206	{
	207	if (hctx->flags & BLK_MQ_F_TAG_QUEUE_SHARED)
	208	__blk_mq_tag_idle(hctx);
	209	}
	210
	211	static inline bool blk_mq_tag_is_reserved(struct blk_mq_tags *tags,
	212	unsigned int tag)
	213	{
	214	return tag < tags->nr_reserved_tags;
	215	}
	216
079a2e3e	217	static inline bool blk_mq_is_shared_tags(unsigned int flags)
32bc15af JG	218	{
	219	return flags & BLK_MQ_F_TAG_HCTX_SHARED;
	220	}
	221
4941115b JA	222	static inline struct blk_mq_tags blk_mq_tags_from_data(struct blk_mq_alloc_data data)
4941115b JA	223	{
dd6216bb CH	224	if (data->rq_flags & RQF_SCHED_TAGS)
	225	return data->hctx->sched_tags;
	226	return data->hctx->tags;
4941115b JA	227	}
4941115b JA	228
5d1b25c1 BVA	229	static inline bool blk_mq_hctx_stopped(struct blk_mq_hw_ctx *hctx)
	230	{
	231	return test_bit(BLK_MQ_S_STOPPED, &hctx->state);
	232	}
	233
19c66e59 ML	234	static inline bool blk_mq_hw_queue_mapped(struct blk_mq_hw_ctx *hctx)
	235	{
	236	return hctx->nr_ctx && hctx->tags;
	237	}
	238
8446fe92 CH	239	unsigned int blk_mq_in_flight(struct request_queue *q,
	240	struct block_device *part);
	241	void blk_mq_in_flight_rw(struct request_queue q, struct block_device part,
	242	unsigned int inflight[2]);
f299b7c7	243
2a5a24aa ML	244	static inline void blk_mq_put_dispatch_budget(struct request_queue *q,
2a5a24aa ML	245	int budget_token)
de148297	246	{
de148297	247	if (q->mq_ops->put_budget)
2a5a24aa	248	q->mq_ops->put_budget(q, budget_token);
de148297 ML	249	}
de148297 ML	250
2a5a24aa	251	static inline int blk_mq_get_dispatch_budget(struct request_queue *q)
de148297	252	{
de148297	253	if (q->mq_ops->get_budget)
65c76369	254	return q->mq_ops->get_budget(q);
2a5a24aa ML	255	return 0;
	256	}
	257
	258	static inline void blk_mq_set_rq_budget_token(struct request *rq, int token)
	259	{
	260	if (token < 0)
	261	return;
	262
	263	if (rq->q->mq_ops->set_rq_budget_token)
	264	rq->q->mq_ops->set_rq_budget_token(rq, token);
	265	}
	266
	267	static inline int blk_mq_get_rq_budget_token(struct request *rq)
	268	{
	269	if (rq->q->mq_ops->get_rq_budget_token)
	270	return rq->q->mq_ops->get_rq_budget_token(rq);
	271	return -1;
de148297 ML	272	}
de148297 ML	273
bccf5e26 JG	274	static inline void __blk_mq_inc_active_requests(struct blk_mq_hw_ctx *hctx)
bccf5e26 JG	275	{
079a2e3e JG	276	if (blk_mq_is_shared_tags(hctx->flags))
079a2e3e JG	277	atomic_inc(&hctx->queue->nr_active_requests_shared_tags);
bccf5e26 JG	278	else
	279	atomic_inc(&hctx->nr_active);
	280	}
	281
3b87c6ea ML	282	static inline void __blk_mq_sub_active_requests(struct blk_mq_hw_ctx *hctx,
3b87c6ea ML	283	int val)
bccf5e26	284	{
079a2e3e	285	if (blk_mq_is_shared_tags(hctx->flags))
3b87c6ea	286	atomic_sub(val, &hctx->queue->nr_active_requests_shared_tags);
bccf5e26	287	else
3b87c6ea ML	288	atomic_sub(val, &hctx->nr_active);
	289	}
	290
	291	static inline void __blk_mq_dec_active_requests(struct blk_mq_hw_ctx *hctx)
	292	{
	293	__blk_mq_sub_active_requests(hctx, 1);
bccf5e26 JG	294	}
	295
	296	static inline int __blk_mq_active_requests(struct blk_mq_hw_ctx *hctx)
	297	{
079a2e3e JG	298	if (blk_mq_is_shared_tags(hctx->flags))
079a2e3e JG	299	return atomic_read(&hctx->queue->nr_active_requests_shared_tags);
bccf5e26 JG	300	return atomic_read(&hctx->nr_active);
bccf5e26 JG	301	}
4e2f62e5 JA	302	static inline void __blk_mq_put_driver_tag(struct blk_mq_hw_ctx *hctx,
	303	struct request *rq)
	304	{
	305	blk_mq_put_tag(hctx->tags, rq->mq_ctx, rq->tag);
	306	rq->tag = BLK_MQ_NO_TAG;
	307
	308	if (rq->rq_flags & RQF_MQ_INFLIGHT) {
	309	rq->rq_flags &= ~RQF_MQ_INFLIGHT;
bccf5e26	310	__blk_mq_dec_active_requests(hctx);
4e2f62e5 JA	311	}
	312	}
	313
	314	static inline void blk_mq_put_driver_tag(struct request *rq)
	315	{
	316	if (rq->tag == BLK_MQ_NO_TAG \|\| rq->internal_tag == BLK_MQ_NO_TAG)
	317	return;
	318
	319	__blk_mq_put_driver_tag(rq->mq_hctx, rq);
	320	}
	321
a808a9d5 JA	322	bool __blk_mq_get_driver_tag(struct blk_mq_hw_ctx hctx, struct request rq);
	323
	324	static inline bool blk_mq_get_driver_tag(struct request *rq)
	325	{
	326	struct blk_mq_hw_ctx *hctx = rq->mq_hctx;
	327
	328	if (rq->tag != BLK_MQ_NO_TAG &&
	329	!(hctx->flags & BLK_MQ_F_TAG_QUEUE_SHARED)) {
	330	hctx->tags->rqs[rq->tag] = rq;
	331	return true;
	332	}
	333
	334	return __blk_mq_get_driver_tag(hctx, rq);
	335	}
61347154	336
ed76e329	337	static inline void blk_mq_clear_mq_map(struct blk_mq_queue_map *qmap)
0da73d00 MI	338	{
	339	int cpu;
	340
	341	for_each_possible_cpu(cpu)
ed76e329	342	qmap->mq_map[cpu] = 0;
0da73d00 MI	343	}
0da73d00 MI	344
b49773e7 DLM	345	/*
b49773e7 DLM	346	* blk_mq_plug() - Get caller context plug
b49773e7 DLM	347	* @bio : the bio being submitted by the caller context
	348	*
	349	* Plugging, by design, may delay the insertion of BIOs into the elevator in
	350	* order to increase BIO merging opportunities. This however can cause BIO
	351	* insertion order to change from the order in which submit_bio() is being
	352	* executed in the case of multiple contexts concurrently issuing BIOs to a
	353	* device, even if these context are synchronized to tightly control BIO issuing
	354	* order. While this is not a problem with regular block devices, this ordering
	355	* change can cause write BIO failures with zoned block devices as these
	356	* require sequential write patterns to zones. Prevent this from happening by
6deacb3b CH	357	* ignoring the plug state of a BIO issuing context if it is for a zoned block
6deacb3b CH	358	* device and the BIO to plug is a write operation.
b49773e7 DLM	359	*
	360	* Return current->plug if the bio can be plugged and NULL otherwise
	361	*/
6deacb3b	362	static inline struct blk_plug blk_mq_plug( struct bio bio)
b49773e7	363	{
6deacb3b	364	/* Zoned block device write operation case: do not plug the BIO */
8cafdb5a PR	365	if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) &&
8cafdb5a PR	366	bdev_op_is_zoned_write(bio->bi_bdev, bio_op(bio)))
6deacb3b CH	367	return NULL;
6deacb3b CH	368
b49773e7 DLM	369	/*
	370	* For regular block devices or read operations, use the context plug
	371	* which may be NULL if blk_start_plug() was not executed.
	372	*/
6deacb3b	373	return current->plug;
b49773e7 DLM	374	}
b49773e7 DLM	375
fd2ef39c JK	376	/* Free all requests on the list */
	377	static inline void blk_mq_free_requests(struct list_head *list)
	378	{
	379	while (!list_empty(list)) {
	380	struct request *rq = list_entry_rq(list->next);
	381
	382	list_del_init(&rq->queuelist);
	383	blk_mq_free_request(rq);
	384	}
	385	}
	386
a0235d23 JG	387	/*
	388	* For shared tag users, we track the number of currently active users
	389	* and attempt to provide a fair share of the tag depth for each of them.
	390	*/
	391	static inline bool hctx_may_queue(struct blk_mq_hw_ctx *hctx,
	392	struct sbitmap_queue *bt)
	393	{
	394	unsigned int depth, users;
	395
	396	if (!hctx \|\| !(hctx->flags & BLK_MQ_F_TAG_QUEUE_SHARED))
	397	return true;
a0235d23 JG	398
	399	/*
	400	* Don't try dividing an ant
	401	*/
	402	if (bt->sb.depth == 1)
	403	return true;
	404
079a2e3e	405	if (blk_mq_is_shared_tags(hctx->flags)) {
f1b49fdc	406	struct request_queue *q = hctx->queue;
f1b49fdc	407
2569063c	408	if (!test_bit(QUEUE_FLAG_HCTX_ACTIVE, &q->queue_flags))
f1b49fdc	409	return true;
f1b49fdc JG	410	} else {
	411	if (!test_bit(BLK_MQ_S_TAG_ACTIVE, &hctx->state))
	412	return true;
f1b49fdc JG	413	}
f1b49fdc JG	414
4f1731df	415	users = READ_ONCE(hctx->tags->active_queues);
a0235d23 JG	416	if (!users)
	417	return true;
	418
	419	/*
	420	* Allow at least some tags
	421	*/
	422	depth = max((bt->sb.depth + users - 1) / users, 4U);
bccf5e26	423	return __blk_mq_active_requests(hctx) < depth;
a0235d23 JG	424	}
a0235d23 JG	425
2a904d00	426	/* run the code block in @dispatch_ops with rcu/srcu read lock held */
41adf531	427	#define __blk_mq_run_dispatch_ops(q, check_sleep, dispatch_ops) \
2a904d00	428	do { \
80bd4a7a	429	if ((q)->tag_set->flags & BLK_MQ_F_BLOCKING) { \
00e885ef	430	struct blk_mq_tag_set *__tag_set = (q)->tag_set; \
2a904d00 ML	431	int srcu_idx; \
2a904d00 ML	432	\
41adf531	433	might_sleep_if(check_sleep); \
00e885ef	434	srcu_idx = srcu_read_lock(__tag_set->srcu); \
2a904d00	435	(dispatch_ops); \
00e885ef	436	srcu_read_unlock(__tag_set->srcu, srcu_idx); \
80bd4a7a CH	437	} else { \
	438	rcu_read_lock(); \
	439	(dispatch_ops); \
	440	rcu_read_unlock(); \
2a904d00 ML	441	} \
2a904d00 ML	442	} while (0)
a0235d23	443
41adf531 ML	444	#define blk_mq_run_dispatch_ops(q, dispatch_ops) \
	445	__blk_mq_run_dispatch_ops(q, true, dispatch_ops) \
	446
320ae51f	447	#endif