[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,
};

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_add_to_requeue_list(struct request *rq, bool at_head);
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);
/*
 * Internal helpers for request insertion into sw queues
 */
void blk_mq_request_bypass_insert(struct request *rq, bool at_head);

/*
 * 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_ELV))
		return data->hctx->tags;
	return data->hctx->sched_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 = atomic_read(&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) {		\
		int srcu_idx;					\
								\
		might_sleep_if(check_sleep);			\
		srcu_idx = srcu_read_lock((q)->tag_set->srcu);	\
		(dispatch_ops);					\
		srcu_read_unlock((q)->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
3e08773c	39	void blk_mq_submit_bio(struct bio *bio);
5a72e899 JA	40	int blk_mq_poll(struct request_queue q, blk_qc_t cookie, struct io_comp_batch iob,
5a72e899 JA	41	unsigned int flags);
c7e2d94b	42	void blk_mq_exit_queue(struct request_queue *q);
e3a2b3f9	43	int blk_mq_update_nr_requests(struct request_queue *q, unsigned int nr);
aed3ea94	44	void blk_mq_wake_waiters(struct request_queue *q);
1fd40b5e ML	45	bool blk_mq_dispatch_rq_list(struct blk_mq_hw_ctx hctx, struct list_head ,
1fd40b5e ML	46	unsigned int);
214a4418	47	void blk_mq_add_to_requeue_list(struct request *rq, bool at_head);
2c3ad667	48	void blk_mq_flush_busy_ctxs(struct blk_mq_hw_ctx hctx, struct list_head list);
b347689f ML	49	struct request blk_mq_dequeue_from_ctx(struct blk_mq_hw_ctx hctx,
b347689f ML	50	struct blk_mq_ctx *start);
2e315dc0	51	void blk_mq_put_rq_ref(struct request *rq);
2c3ad667 JA	52
	53	/*
	54	* Internal helpers for allocating/freeing the request map
	55	*/
cc71a6f4 JA	56	void blk_mq_free_rqs(struct blk_mq_tag_set set, struct blk_mq_tags tags,
cc71a6f4 JA	57	unsigned int hctx_idx);
e155b0c2	58	void blk_mq_free_rq_map(struct blk_mq_tags *tags);
63064be1 JG	59	struct blk_mq_tags blk_mq_alloc_map_and_rqs(struct blk_mq_tag_set set,
63064be1 JG	60	unsigned int hctx_idx, unsigned int depth);
645db34e JG	61	void blk_mq_free_map_and_rqs(struct blk_mq_tag_set *set,
	62	struct blk_mq_tags *tags,
	63	unsigned int hctx_idx);
2c3ad667 JA	64	/*
	65	* Internal helpers for request insertion into sw queues
	66	*/
2394395c	67	void blk_mq_request_bypass_insert(struct request *rq, bool at_head);
396eaf21	68
320ae51f JA	69	/*
	70	* CPU -> queue mappings
	71	*/
ed76e329	72	extern int blk_mq_hw_queue_to_node(struct blk_mq_queue_map *qmap, unsigned int);
320ae51f	73
b3c661b1 JA	74	/*
	75	* blk_mq_map_queue_type() - map (hctx_type,cpu) to hardware queue
	76	* @q: request queue
e20ba6e1	77	* @type: the hctx type index
b3c661b1 JA	78	* @cpu: CPU
	79	*/
	80	static inline struct blk_mq_hw_ctx blk_mq_map_queue_type(struct request_queue q,
e20ba6e1	81	enum hctx_type type,
b3c661b1	82	unsigned int cpu)
7d7e0f90	83	{
4e5cc99e	84	return xa_load(&q->hctx_table, q->tag_set->map[type].mq_map[cpu]);
7d7e0f90 CH	85	}
7d7e0f90 CH	86
16458cf3	87	static inline enum hctx_type blk_mq_get_hctx_type(blk_opf_t opf)
ff2c5660	88	{
e20ba6e1 CH	89	enum hctx_type type = HCTX_TYPE_DEFAULT;
e20ba6e1 CH	90
bb94aea1	91	/*
6ce913fe	92	* The caller ensure that if REQ_POLLED, poll must be enabled.
bb94aea1	93	*/
7e923f40	94	if (opf & REQ_POLLED)
e20ba6e1	95	type = HCTX_TYPE_POLL;
7e923f40	96	else if ((opf & REQ_OP_MASK) == REQ_OP_READ)
e20ba6e1	97	type = HCTX_TYPE_READ;
b637108a ML	98	return type;
	99	}
	100
	101	/*
	102	* blk_mq_map_queue() - map (cmd_flags,type) to hardware queue
	103	* @q: request queue
7e923f40	104	* @opf: operation type (REQ_OP_*) and flags (e.g. REQ_POLLED).
b637108a ML	105	* @ctx: software queue cpu ctx
	106	*/
	107	static inline struct blk_mq_hw_ctx blk_mq_map_queue(struct request_queue q,
16458cf3	108	blk_opf_t opf,
b637108a ML	109	struct blk_mq_ctx *ctx)
b637108a ML	110	{
7e923f40	111	return ctx->hctxs[blk_mq_get_hctx_type(opf)];
ff2c5660 JA	112	}
ff2c5660 JA	113
67aec14c JA	114	/*
	115	* sysfs helpers
	116	*/
737f98cf	117	extern void blk_mq_sysfs_init(struct request_queue *q);
7ea5fe31	118	extern void blk_mq_sysfs_deinit(struct request_queue *q);
8682b92e CH	119	int blk_mq_sysfs_register(struct gendisk *disk);
8682b92e CH	120	void blk_mq_sysfs_unregister(struct gendisk *disk);
eaa870f9 CH	121	int blk_mq_sysfs_register_hctxs(struct request_queue *q);
eaa870f9 CH	122	void blk_mq_sysfs_unregister_hctxs(struct request_queue *q);
868f2f0b	123	extern void blk_mq_hctx_kobj_init(struct blk_mq_hw_ctx *hctx);
47c122e3	124	void blk_mq_free_plug_rqs(struct blk_plug *plug);
dbb6f764	125	void blk_mq_flush_plug_list(struct blk_plug *plug, bool from_schedule);
67aec14c	126
2a19b28f ML	127	void blk_mq_cancel_work_sync(struct request_queue *q);
2a19b28f ML	128
e09aae7e ML	129	void blk_mq_release(struct request_queue *q);
e09aae7e ML	130
1aecfe48 ML	131	static inline struct blk_mq_ctx __blk_mq_get_ctx(struct request_queue q,
	132	unsigned int cpu)
	133	{
	134	return per_cpu_ptr(q->queue_ctx, cpu);
	135	}
	136
	137	/*
	138	* This assumes per-cpu software queueing queues. They could be per-node
	139	* as well, for instance. For now this is hardcoded as-is. Note that we don't
	140	* care about preemption, since we know the ctx's are persistent. This does
	141	* mean that we can't rely on ctx always matching the currently running CPU.
	142	*/
	143	static inline struct blk_mq_ctx blk_mq_get_ctx(struct request_queue q)
	144	{
c05f4220	145	return __blk_mq_get_ctx(q, raw_smp_processor_id());
1aecfe48 ML	146	}
1aecfe48 ML	147
cb96a42c ML	148	struct blk_mq_alloc_data {
	149	/* input parameter */
	150	struct request_queue *q;
9a95e4ef	151	blk_mq_req_flags_t flags;
229a9287	152	unsigned int shallow_depth;
16458cf3	153	blk_opf_t cmd_flags;
ecaf97f4	154	req_flags_t rq_flags;
cb96a42c	155
47c122e3 JA	156	/* allocate multiple requests/tags in one go */
	157	unsigned int nr_tags;
	158	struct request **cached_rq;
	159
cb96a42c ML	160	/* input & output parameter */
	161	struct blk_mq_ctx *ctx;
	162	struct blk_mq_hw_ctx *hctx;
	163	};
	164
bebe84eb CH	165	struct blk_mq_tags *blk_mq_init_tags(unsigned int nr_tags,
	166	unsigned int reserved_tags, int node, int alloc_policy);
	167	void blk_mq_free_tags(struct blk_mq_tags *tags);
	168	int blk_mq_init_bitmaps(struct sbitmap_queue *bitmap_tags,
	169	struct sbitmap_queue *breserved_tags, unsigned int queue_depth,
	170	unsigned int reserved, int node, int alloc_policy);
	171
	172	unsigned int blk_mq_get_tag(struct blk_mq_alloc_data *data);
	173	unsigned long blk_mq_get_tags(struct blk_mq_alloc_data *data, int nr_tags,
	174	unsigned int *offset);
	175	void blk_mq_put_tag(struct blk_mq_tags tags, struct blk_mq_ctx ctx,
	176	unsigned int tag);
	177	void blk_mq_put_tags(struct blk_mq_tags tags, int tag_array, int nr_tags);
	178	int blk_mq_tag_update_depth(struct blk_mq_hw_ctx *hctx,
	179	struct blk_mq_tags **tags, unsigned int depth, bool can_grow);
	180	void blk_mq_tag_resize_shared_tags(struct blk_mq_tag_set *set,
	181	unsigned int size);
	182	void blk_mq_tag_update_sched_shared_tags(struct request_queue *q);
	183
	184	void blk_mq_tag_wakeup_all(struct blk_mq_tags *tags, bool);
	185	void blk_mq_queue_tag_busy_iter(struct request_queue q, busy_tag_iter_fn fn,
	186	void *priv);
	187	void blk_mq_all_tag_iter(struct blk_mq_tags tags, busy_tag_iter_fn fn,
	188	void *priv);
	189
	190	static inline struct sbq_wait_state bt_wait_ptr(struct sbitmap_queue bt,
	191	struct blk_mq_hw_ctx *hctx)
	192	{
	193	if (!hctx)
	194	return &bt->ws[0];
	195	return sbq_wait_ptr(bt, &hctx->wait_index);
	196	}
	197
	198	void __blk_mq_tag_busy(struct blk_mq_hw_ctx *);
	199	void __blk_mq_tag_idle(struct blk_mq_hw_ctx *);
	200
	201	static inline void blk_mq_tag_busy(struct blk_mq_hw_ctx *hctx)
	202	{
	203	if (hctx->flags & BLK_MQ_F_TAG_QUEUE_SHARED)
	204	__blk_mq_tag_busy(hctx);
	205	}
	206
	207	static inline void blk_mq_tag_idle(struct blk_mq_hw_ctx *hctx)
	208	{
	209	if (hctx->flags & BLK_MQ_F_TAG_QUEUE_SHARED)
	210	__blk_mq_tag_idle(hctx);
	211	}
	212
	213	static inline bool blk_mq_tag_is_reserved(struct blk_mq_tags *tags,
	214	unsigned int tag)
	215	{
	216	return tag < tags->nr_reserved_tags;
	217	}
	218
079a2e3e	219	static inline bool blk_mq_is_shared_tags(unsigned int flags)
32bc15af JG	220	{
	221	return flags & BLK_MQ_F_TAG_HCTX_SHARED;
	222	}
	223
4941115b JA	224	static inline struct blk_mq_tags blk_mq_tags_from_data(struct blk_mq_alloc_data data)
4941115b JA	225	{
56f8da64 JA	226	if (!(data->rq_flags & RQF_ELV))
	227	return data->hctx->tags;
	228	return data->hctx->sched_tags;
4941115b JA	229	}
4941115b JA	230
5d1b25c1 BVA	231	static inline bool blk_mq_hctx_stopped(struct blk_mq_hw_ctx *hctx)
	232	{
	233	return test_bit(BLK_MQ_S_STOPPED, &hctx->state);
	234	}
	235
19c66e59 ML	236	static inline bool blk_mq_hw_queue_mapped(struct blk_mq_hw_ctx *hctx)
	237	{
	238	return hctx->nr_ctx && hctx->tags;
	239	}
	240
8446fe92 CH	241	unsigned int blk_mq_in_flight(struct request_queue *q,
	242	struct block_device *part);
	243	void blk_mq_in_flight_rw(struct request_queue q, struct block_device part,
	244	unsigned int inflight[2]);
f299b7c7	245
2a5a24aa ML	246	static inline void blk_mq_put_dispatch_budget(struct request_queue *q,
2a5a24aa ML	247	int budget_token)
de148297	248	{
de148297	249	if (q->mq_ops->put_budget)
2a5a24aa	250	q->mq_ops->put_budget(q, budget_token);
de148297 ML	251	}
de148297 ML	252
2a5a24aa	253	static inline int blk_mq_get_dispatch_budget(struct request_queue *q)
de148297	254	{
de148297	255	if (q->mq_ops->get_budget)
65c76369	256	return q->mq_ops->get_budget(q);
2a5a24aa ML	257	return 0;
	258	}
	259
	260	static inline void blk_mq_set_rq_budget_token(struct request *rq, int token)
	261	{
	262	if (token < 0)
	263	return;
	264
	265	if (rq->q->mq_ops->set_rq_budget_token)
	266	rq->q->mq_ops->set_rq_budget_token(rq, token);
	267	}
	268
	269	static inline int blk_mq_get_rq_budget_token(struct request *rq)
	270	{
	271	if (rq->q->mq_ops->get_rq_budget_token)
	272	return rq->q->mq_ops->get_rq_budget_token(rq);
	273	return -1;
de148297 ML	274	}
de148297 ML	275
bccf5e26 JG	276	static inline void __blk_mq_inc_active_requests(struct blk_mq_hw_ctx *hctx)
bccf5e26 JG	277	{
079a2e3e JG	278	if (blk_mq_is_shared_tags(hctx->flags))
079a2e3e JG	279	atomic_inc(&hctx->queue->nr_active_requests_shared_tags);
bccf5e26 JG	280	else
	281	atomic_inc(&hctx->nr_active);
	282	}
	283
3b87c6ea ML	284	static inline void __blk_mq_sub_active_requests(struct blk_mq_hw_ctx *hctx,
3b87c6ea ML	285	int val)
bccf5e26	286	{
079a2e3e	287	if (blk_mq_is_shared_tags(hctx->flags))
3b87c6ea	288	atomic_sub(val, &hctx->queue->nr_active_requests_shared_tags);
bccf5e26	289	else
3b87c6ea ML	290	atomic_sub(val, &hctx->nr_active);
	291	}
	292
	293	static inline void __blk_mq_dec_active_requests(struct blk_mq_hw_ctx *hctx)
	294	{
	295	__blk_mq_sub_active_requests(hctx, 1);
bccf5e26 JG	296	}
	297
	298	static inline int __blk_mq_active_requests(struct blk_mq_hw_ctx *hctx)
	299	{
079a2e3e JG	300	if (blk_mq_is_shared_tags(hctx->flags))
079a2e3e JG	301	return atomic_read(&hctx->queue->nr_active_requests_shared_tags);
bccf5e26 JG	302	return atomic_read(&hctx->nr_active);
bccf5e26 JG	303	}
4e2f62e5 JA	304	static inline void __blk_mq_put_driver_tag(struct blk_mq_hw_ctx *hctx,
	305	struct request *rq)
	306	{
	307	blk_mq_put_tag(hctx->tags, rq->mq_ctx, rq->tag);
	308	rq->tag = BLK_MQ_NO_TAG;
	309
	310	if (rq->rq_flags & RQF_MQ_INFLIGHT) {
	311	rq->rq_flags &= ~RQF_MQ_INFLIGHT;
bccf5e26	312	__blk_mq_dec_active_requests(hctx);
4e2f62e5 JA	313	}
	314	}
	315
	316	static inline void blk_mq_put_driver_tag(struct request *rq)
	317	{
	318	if (rq->tag == BLK_MQ_NO_TAG \|\| rq->internal_tag == BLK_MQ_NO_TAG)
	319	return;
	320
	321	__blk_mq_put_driver_tag(rq->mq_hctx, rq);
	322	}
	323
a808a9d5 JA	324	bool __blk_mq_get_driver_tag(struct blk_mq_hw_ctx hctx, struct request rq);
	325
	326	static inline bool blk_mq_get_driver_tag(struct request *rq)
	327	{
	328	struct blk_mq_hw_ctx *hctx = rq->mq_hctx;
	329
	330	if (rq->tag != BLK_MQ_NO_TAG &&
	331	!(hctx->flags & BLK_MQ_F_TAG_QUEUE_SHARED)) {
	332	hctx->tags->rqs[rq->tag] = rq;
	333	return true;
	334	}
	335
	336	return __blk_mq_get_driver_tag(hctx, rq);
	337	}
61347154	338
ed76e329	339	static inline void blk_mq_clear_mq_map(struct blk_mq_queue_map *qmap)
0da73d00 MI	340	{
	341	int cpu;
	342
	343	for_each_possible_cpu(cpu)
ed76e329	344	qmap->mq_map[cpu] = 0;
0da73d00 MI	345	}
0da73d00 MI	346
b49773e7 DLM	347	/*
b49773e7 DLM	348	* blk_mq_plug() - Get caller context plug
b49773e7 DLM	349	* @bio : the bio being submitted by the caller context
	350	*
	351	* Plugging, by design, may delay the insertion of BIOs into the elevator in
	352	* order to increase BIO merging opportunities. This however can cause BIO
	353	* insertion order to change from the order in which submit_bio() is being
	354	* executed in the case of multiple contexts concurrently issuing BIOs to a
	355	* device, even if these context are synchronized to tightly control BIO issuing
	356	* order. While this is not a problem with regular block devices, this ordering
	357	* change can cause write BIO failures with zoned block devices as these
	358	* require sequential write patterns to zones. Prevent this from happening by
6deacb3b CH	359	* ignoring the plug state of a BIO issuing context if it is for a zoned block
6deacb3b CH	360	* device and the BIO to plug is a write operation.
b49773e7 DLM	361	*
	362	* Return current->plug if the bio can be plugged and NULL otherwise
	363	*/
6deacb3b	364	static inline struct blk_plug blk_mq_plug( struct bio bio)
b49773e7	365	{
6deacb3b	366	/* Zoned block device write operation case: do not plug the BIO */
8cafdb5a PR	367	if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) &&
8cafdb5a PR	368	bdev_op_is_zoned_write(bio->bi_bdev, bio_op(bio)))
6deacb3b CH	369	return NULL;
6deacb3b CH	370
b49773e7 DLM	371	/*
	372	* For regular block devices or read operations, use the context plug
	373	* which may be NULL if blk_start_plug() was not executed.
	374	*/
6deacb3b	375	return current->plug;
b49773e7 DLM	376	}
b49773e7 DLM	377
fd2ef39c JK	378	/* Free all requests on the list */
	379	static inline void blk_mq_free_requests(struct list_head *list)
	380	{
	381	while (!list_empty(list)) {
	382	struct request *rq = list_entry_rq(list->next);
	383
	384	list_del_init(&rq->queuelist);
	385	blk_mq_free_request(rq);
	386	}
	387	}
	388
a0235d23 JG	389	/*
	390	* For shared tag users, we track the number of currently active users
	391	* and attempt to provide a fair share of the tag depth for each of them.
	392	*/
	393	static inline bool hctx_may_queue(struct blk_mq_hw_ctx *hctx,
	394	struct sbitmap_queue *bt)
	395	{
	396	unsigned int depth, users;
	397
	398	if (!hctx \|\| !(hctx->flags & BLK_MQ_F_TAG_QUEUE_SHARED))
	399	return true;
a0235d23 JG	400
	401	/*
	402	* Don't try dividing an ant
	403	*/
	404	if (bt->sb.depth == 1)
	405	return true;
	406
079a2e3e	407	if (blk_mq_is_shared_tags(hctx->flags)) {
f1b49fdc	408	struct request_queue *q = hctx->queue;
f1b49fdc	409
2569063c	410	if (!test_bit(QUEUE_FLAG_HCTX_ACTIVE, &q->queue_flags))
f1b49fdc	411	return true;
f1b49fdc JG	412	} else {
	413	if (!test_bit(BLK_MQ_S_TAG_ACTIVE, &hctx->state))
	414	return true;
f1b49fdc JG	415	}
f1b49fdc JG	416
e155b0c2 JG	417	users = atomic_read(&hctx->tags->active_queues);
e155b0c2 JG	418
a0235d23 JG	419	if (!users)
	420	return true;
	421
	422	/*
	423	* Allow at least some tags
	424	*/
	425	depth = max((bt->sb.depth + users - 1) / users, 4U);
bccf5e26	426	return __blk_mq_active_requests(hctx) < depth;
a0235d23 JG	427	}
a0235d23 JG	428
2a904d00	429	/* run the code block in @dispatch_ops with rcu/srcu read lock held */
41adf531	430	#define __blk_mq_run_dispatch_ops(q, check_sleep, dispatch_ops) \
2a904d00	431	do { \
80bd4a7a	432	if ((q)->tag_set->flags & BLK_MQ_F_BLOCKING) { \
2a904d00 ML	433	int srcu_idx; \
2a904d00 ML	434	\
41adf531	435	might_sleep_if(check_sleep); \
80bd4a7a	436	srcu_idx = srcu_read_lock((q)->tag_set->srcu); \
2a904d00	437	(dispatch_ops); \
80bd4a7a CH	438	srcu_read_unlock((q)->tag_set->srcu, srcu_idx); \
	439	} else { \
	440	rcu_read_lock(); \
	441	(dispatch_ops); \
	442	rcu_read_unlock(); \
2a904d00 ML	443	} \
2a904d00 ML	444	} while (0)
a0235d23	445
41adf531 ML	446	#define blk_mq_run_dispatch_ops(q, dispatch_ops) \
	447	__blk_mq_run_dispatch_ops(q, true, dispatch_ops) \
	448
320ae51f	449	#endif