[mirror_ubuntu-jammy-kernel.git] / block / blk-mq.h

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

#include "blk-stat.h"
#include "blk-mq-tag.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];

	/* incremented at dispatch time */
	unsigned long		rq_dispatched[2];
	unsigned long		rq_merged;

	/* incremented at completion time */
	unsigned long		____cacheline_aligned_in_smp rq_completed[2];

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

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 request_queue *, struct list_head *, bool);
void blk_mq_add_to_requeue_list(struct request *rq, bool at_head,
				bool kick_requeue_list);
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);

/*
 * 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_rq_map(struct blk_mq_tag_set *set,
					unsigned int hctx_idx,
					unsigned int nr_tags,
					unsigned int reserved_tags);
int blk_mq_alloc_rqs(struct blk_mq_tag_set *set, struct blk_mq_tags *tags,
		     unsigned int hctx_idx, unsigned int depth);

/*
 * Internal helpers for request insertion into sw queues
 */
void __blk_mq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq,
				bool at_head);
void blk_mq_request_bypass_insert(struct request *rq, bool at_head,
				  bool run_queue);
void blk_mq_insert_requests(struct blk_mq_hw_ctx *hctx, struct blk_mq_ctx *ctx,
				struct list_head *list);

/* Used by blk_insert_cloned_request() to issue request directly */
blk_status_t blk_mq_request_issue_directly(struct request *rq, bool last);
void blk_mq_try_issue_list_directly(struct blk_mq_hw_ctx *hctx,
				    struct list_head *list);

/*
 * 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 q->queue_hw_ctx[q->tag_set->map[type].mq_map[cpu]];
}

/*
 * blk_mq_map_queue() - map (cmd_flags,type) to hardware queue
 * @q: request queue
 * @flags: request command flags
 * @cpu: cpu ctx
 */
static inline struct blk_mq_hw_ctx *blk_mq_map_queue(struct request_queue *q,
						     unsigned int flags,
						     struct blk_mq_ctx *ctx)
{
	enum hctx_type type = HCTX_TYPE_DEFAULT;

	/*
	 * The caller ensure that if REQ_HIPRI, poll must be enabled.
	 */
	if (flags & REQ_HIPRI)
		type = HCTX_TYPE_POLL;
	else if ((flags & REQ_OP_MASK) == REQ_OP_READ)
		type = HCTX_TYPE_READ;
	
	return ctx->hctxs[type];
}

/*
 * sysfs helpers
 */
extern void blk_mq_sysfs_init(struct request_queue *q);
extern void blk_mq_sysfs_deinit(struct request_queue *q);
extern int __blk_mq_register_dev(struct device *dev, struct request_queue *q);
extern int blk_mq_sysfs_register(struct request_queue *q);
extern void blk_mq_sysfs_unregister(struct request_queue *q);
extern void blk_mq_hctx_kobj_init(struct blk_mq_hw_ctx *hctx);

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;
	unsigned int cmd_flags;

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

static inline struct blk_mq_tags *blk_mq_tags_from_data(struct blk_mq_alloc_data *data)
{
	if (data->flags & BLK_MQ_REQ_INTERNAL)
		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 hd_struct *part);
void blk_mq_in_flight_rw(struct request_queue *q, struct hd_struct *part,
			 unsigned int inflight[2]);

static inline void blk_mq_put_dispatch_budget(struct blk_mq_hw_ctx *hctx)
{
	struct request_queue *q = hctx->queue;

	if (q->mq_ops->put_budget)
		q->mq_ops->put_budget(hctx);
}

static inline bool blk_mq_get_dispatch_budget(struct blk_mq_hw_ctx *hctx)
{
	struct request_queue *q = hctx->queue;

	if (q->mq_ops->get_budget)
		return q->mq_ops->get_budget(hctx);
	return true;
}

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;
		atomic_dec(&hctx->nr_active);
	}
}

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);
}

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
 * @q: request queue
 * @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 the target request queue
 * 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 request_queue *q,
					   struct bio *bio)
{
	/*
	 * For regular block devices or read operations, use the context plug
	 * which may be NULL if blk_start_plug() was not executed.
	 */
	if (!blk_queue_is_zoned(q) || !op_is_write(bio_op(bio)))
		return current->plug;

	/* Zoned block device write operation case: do not plug the BIO */
	return NULL;
}

#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
cf43e6be	5	#include "blk-stat.h"
244c65a3	6	#include "blk-mq-tag.h"
cf43e6be	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 JA	27
	28	/* incremented at dispatch time */
	29	unsigned long rq_dispatched[2];
	30	unsigned long rq_merged;
	31
	32	/* incremented at completion time */
	33	unsigned long ____cacheline_aligned_in_smp rq_completed[2];
	34
	35	struct request_queue *queue;
1db4909e	36	struct blk_mq_ctxs *ctxs;
320ae51f	37	struct kobject kobj;
4bb659b1	38	} ____cacheline_aligned_in_smp;
320ae51f	39
c7e2d94b	40	void blk_mq_exit_queue(struct request_queue *q);
e3a2b3f9	41	int blk_mq_update_nr_requests(struct request_queue *q, unsigned int nr);
aed3ea94	42	void blk_mq_wake_waiters(struct request_queue *q);
de148297	43	bool blk_mq_dispatch_rq_list(struct request_queue , struct list_head , bool);
e6c98712 BVA	44	void blk_mq_add_to_requeue_list(struct request *rq, bool at_head,
e6c98712 BVA	45	bool kick_requeue_list);
2c3ad667	46	void blk_mq_flush_busy_ctxs(struct blk_mq_hw_ctx hctx, struct list_head list);
b347689f ML	47	struct request blk_mq_dequeue_from_ctx(struct blk_mq_hw_ctx hctx,
b347689f ML	48	struct blk_mq_ctx *start);
2c3ad667 JA	49
	50	/*
	51	* Internal helpers for allocating/freeing the request map
	52	*/
cc71a6f4 JA	53	void blk_mq_free_rqs(struct blk_mq_tag_set set, struct blk_mq_tags tags,
	54	unsigned int hctx_idx);
	55	void blk_mq_free_rq_map(struct blk_mq_tags *tags);
	56	struct blk_mq_tags blk_mq_alloc_rq_map(struct blk_mq_tag_set set,
	57	unsigned int hctx_idx,
	58	unsigned int nr_tags,
	59	unsigned int reserved_tags);
	60	int blk_mq_alloc_rqs(struct blk_mq_tag_set set, struct blk_mq_tags tags,
	61	unsigned int hctx_idx, unsigned int depth);
2c3ad667 JA	62
	63	/*
	64	* Internal helpers for request insertion into sw queues
	65	*/
	66	void __blk_mq_insert_request(struct blk_mq_hw_ctx hctx, struct request rq,
	67	bool at_head);
01e99aec ML	68	void blk_mq_request_bypass_insert(struct request *rq, bool at_head,
01e99aec ML	69	bool run_queue);
bd166ef1 JA	70	void blk_mq_insert_requests(struct blk_mq_hw_ctx hctx, struct blk_mq_ctx ctx,
bd166ef1 JA	71	struct list_head *list);
320ae51f	72
fd9c40f6 BVA	73	/* Used by blk_insert_cloned_request() to issue request directly */
fd9c40f6 BVA	74	blk_status_t blk_mq_request_issue_directly(struct request *rq, bool last);
6ce3dd6e ML	75	void blk_mq_try_issue_list_directly(struct blk_mq_hw_ctx *hctx,
6ce3dd6e ML	76	struct list_head *list);
396eaf21	77
320ae51f JA	78	/*
	79	* CPU -> queue mappings
	80	*/
ed76e329	81	extern int blk_mq_hw_queue_to_node(struct blk_mq_queue_map *qmap, unsigned int);
320ae51f	82
b3c661b1 JA	83	/*
	84	* blk_mq_map_queue_type() - map (hctx_type,cpu) to hardware queue
	85	* @q: request queue
e20ba6e1	86	* @type: the hctx type index
b3c661b1 JA	87	* @cpu: CPU
	88	*/
	89	static inline struct blk_mq_hw_ctx blk_mq_map_queue_type(struct request_queue q,
e20ba6e1	90	enum hctx_type type,
b3c661b1	91	unsigned int cpu)
7d7e0f90	92	{
e20ba6e1	93	return q->queue_hw_ctx[q->tag_set->map[type].mq_map[cpu]];
7d7e0f90 CH	94	}
7d7e0f90 CH	95
b3c661b1 JA	96	/*
	97	* blk_mq_map_queue() - map (cmd_flags,type) to hardware queue
	98	* @q: request queue
	99	* @flags: request command flags
8ccdf4a3	100	* @cpu: cpu ctx
b3c661b1 JA	101	*/
	102	static inline struct blk_mq_hw_ctx blk_mq_map_queue(struct request_queue q,
	103	unsigned int flags,
8ccdf4a3	104	struct blk_mq_ctx *ctx)
ff2c5660	105	{
e20ba6e1 CH	106	enum hctx_type type = HCTX_TYPE_DEFAULT;
e20ba6e1 CH	107
bb94aea1 JW	108	/*
	109	* The caller ensure that if REQ_HIPRI, poll must be enabled.
	110	*/
	111	if (flags & REQ_HIPRI)
e20ba6e1	112	type = HCTX_TYPE_POLL;
bb94aea1	113	else if ((flags & REQ_OP_MASK) == REQ_OP_READ)
e20ba6e1	114	type = HCTX_TYPE_READ;
5aceaeb2	115
8ccdf4a3	116	return ctx->hctxs[type];
ff2c5660 JA	117	}
ff2c5660 JA	118
67aec14c JA	119	/*
	120	* sysfs helpers
	121	*/
737f98cf	122	extern void blk_mq_sysfs_init(struct request_queue *q);
7ea5fe31	123	extern void blk_mq_sysfs_deinit(struct request_queue *q);
2d0364c8	124	extern int __blk_mq_register_dev(struct device dev, struct request_queue q);
67aec14c JA	125	extern int blk_mq_sysfs_register(struct request_queue *q);
67aec14c JA	126	extern void blk_mq_sysfs_unregister(struct request_queue *q);
868f2f0b	127	extern void blk_mq_hctx_kobj_init(struct blk_mq_hw_ctx *hctx);
67aec14c	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;
f9afca4d	153	unsigned int cmd_flags;
cb96a42c ML	154
	155	/* input & output parameter */
	156	struct blk_mq_ctx *ctx;
	157	struct blk_mq_hw_ctx *hctx;
	158	};
	159
4941115b JA	160	static inline struct blk_mq_tags blk_mq_tags_from_data(struct blk_mq_alloc_data data)
4941115b JA	161	{
bd166ef1 JA	162	if (data->flags & BLK_MQ_REQ_INTERNAL)
	163	return data->hctx->sched_tags;
	164
4941115b JA	165	return data->hctx->tags;
	166	}
	167
5d1b25c1 BVA	168	static inline bool blk_mq_hctx_stopped(struct blk_mq_hw_ctx *hctx)
	169	{
	170	return test_bit(BLK_MQ_S_STOPPED, &hctx->state);
	171	}
	172
19c66e59 ML	173	static inline bool blk_mq_hw_queue_mapped(struct blk_mq_hw_ctx *hctx)
	174	{
	175	return hctx->nr_ctx && hctx->tags;
	176	}
	177
e016b782	178	unsigned int blk_mq_in_flight(struct request_queue q, struct hd_struct part);
bf0ddaba OS	179	void blk_mq_in_flight_rw(struct request_queue q, struct hd_struct part,
bf0ddaba OS	180	unsigned int inflight[2]);
f299b7c7	181
de148297 ML	182	static inline void blk_mq_put_dispatch_budget(struct blk_mq_hw_ctx *hctx)
	183	{
	184	struct request_queue *q = hctx->queue;
	185
	186	if (q->mq_ops->put_budget)
	187	q->mq_ops->put_budget(hctx);
	188	}
	189
88022d72	190	static inline bool blk_mq_get_dispatch_budget(struct blk_mq_hw_ctx *hctx)
de148297 ML	191	{
	192	struct request_queue *q = hctx->queue;
	193
	194	if (q->mq_ops->get_budget)
	195	return q->mq_ops->get_budget(hctx);
88022d72	196	return true;
de148297 ML	197	}
de148297 ML	198
244c65a3 ML	199	static inline void __blk_mq_put_driver_tag(struct blk_mq_hw_ctx *hctx,
	200	struct request *rq)
	201	{
cae740a0	202	blk_mq_put_tag(hctx->tags, rq->mq_ctx, rq->tag);
76647368	203	rq->tag = BLK_MQ_NO_TAG;
244c65a3 ML	204
	205	if (rq->rq_flags & RQF_MQ_INFLIGHT) {
	206	rq->rq_flags &= ~RQF_MQ_INFLIGHT;
	207	atomic_dec(&hctx->nr_active);
	208	}
	209	}
	210
244c65a3 ML	211	static inline void blk_mq_put_driver_tag(struct request *rq)
244c65a3 ML	212	{
76647368	213	if (rq->tag == BLK_MQ_NO_TAG \|\| rq->internal_tag == BLK_MQ_NO_TAG)
244c65a3 ML	214	return;
244c65a3 ML	215
ea4f995e	216	__blk_mq_put_driver_tag(rq->mq_hctx, rq);
244c65a3 ML	217	}
244c65a3 ML	218
ed76e329	219	static inline void blk_mq_clear_mq_map(struct blk_mq_queue_map *qmap)
0da73d00 MI	220	{
	221	int cpu;
	222
	223	for_each_possible_cpu(cpu)
ed76e329	224	qmap->mq_map[cpu] = 0;
0da73d00 MI	225	}
0da73d00 MI	226
b49773e7 DLM	227	/*
	228	* blk_mq_plug() - Get caller context plug
	229	* @q: request queue
	230	* @bio : the bio being submitted by the caller context
	231	*
	232	* Plugging, by design, may delay the insertion of BIOs into the elevator in
	233	* order to increase BIO merging opportunities. This however can cause BIO
	234	* insertion order to change from the order in which submit_bio() is being
	235	* executed in the case of multiple contexts concurrently issuing BIOs to a
	236	* device, even if these context are synchronized to tightly control BIO issuing
	237	* order. While this is not a problem with regular block devices, this ordering
	238	* change can cause write BIO failures with zoned block devices as these
	239	* require sequential write patterns to zones. Prevent this from happening by
	240	* ignoring the plug state of a BIO issuing context if the target request queue
	241	* is for a zoned block device and the BIO to plug is a write operation.
	242	*
	243	* Return current->plug if the bio can be plugged and NULL otherwise
	244	*/
	245	static inline struct blk_plug blk_mq_plug(struct request_queue q,
	246	struct bio *bio)
	247	{
	248	/*
	249	* For regular block devices or read operations, use the context plug
	250	* which may be NULL if blk_start_plug() was not executed.
	251	*/
	252	if (!blk_queue_is_zoned(q) \|\| !op_is_write(bio_op(bio)))
	253	return current->plug;
	254
	255	/* Zoned block device write operation case: do not plug the BIO */
	256	return NULL;
	257	}
	258
320ae51f	259	#endif