[linux.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];

	/* 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_freeze_queue(struct request_queue *q);
void blk_mq_free_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_flush_busy_ctxs(struct blk_mq_hw_ctx *hctx, struct list_head *list);
bool blk_mq_get_driver_tag(struct request *rq);
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 run_queue);
void blk_mq_insert_requests(struct blk_mq_hw_ctx *hctx, struct blk_mq_ctx *ctx,
				struct list_head *list);

blk_status_t blk_mq_try_issue_directly(struct blk_mq_hw_ctx *hctx,
						struct request *rq,
						blk_qc_t *cookie,
						bool bypass, 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
 */
static inline struct blk_mq_hw_ctx *blk_mq_map_queue(struct request_queue *q,
						     unsigned int flags,
						     unsigned int cpu)
{
	enum hctx_type type = HCTX_TYPE_DEFAULT;

	if ((flags & REQ_HIPRI) &&
	    q->tag_set->nr_maps > HCTX_TYPE_POLL && 
	    q->tag_set->map[HCTX_TYPE_POLL].nr_queues &&
	    test_bit(QUEUE_FLAG_POLL, &q->queue_flags))
		type = HCTX_TYPE_POLL;

	else if (((flags & REQ_OP_MASK) == REQ_OP_READ) &&
	         q->tag_set->nr_maps > HCTX_TYPE_READ &&
		 q->tag_set->map[HCTX_TYPE_READ].nr_queues)
		type = HCTX_TYPE_READ;
	
	return blk_mq_map_queue_type(q, type, cpu);
}

/*
 * 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);

/**
 * blk_mq_rq_state() - read the current MQ_RQ_* state of a request
 * @rq: target request.
 */
static inline enum mq_rq_state blk_mq_rq_state(struct request *rq)
{
	return READ_ONCE(rq->state);
}

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, get_cpu());
}

static inline void blk_mq_put_ctx(struct blk_mq_ctx *ctx)
{
	put_cpu();
}

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, hctx->tags, rq->mq_ctx, rq->tag);
	rq->tag = -1;

	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_hctx(struct blk_mq_hw_ctx *hctx,
				       struct request *rq)
{
	if (rq->tag == -1 || rq->internal_tag == -1)
		return;

	__blk_mq_put_driver_tag(hctx, rq);
}

static inline void blk_mq_put_driver_tag(struct request *rq)
{
	if (rq->tag == -1 || rq->internal_tag == -1)
		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;
}

#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];
320ae51f JA	26
	27	/* incremented at dispatch time */
	28	unsigned long rq_dispatched[2];
	29	unsigned long rq_merged;
	30
	31	/* incremented at completion time */
	32	unsigned long ____cacheline_aligned_in_smp rq_completed[2];
	33
	34	struct request_queue *queue;
1db4909e	35	struct blk_mq_ctxs *ctxs;
320ae51f	36	struct kobject kobj;
4bb659b1	37	} ____cacheline_aligned_in_smp;
320ae51f	38
780db207	39	void blk_mq_freeze_queue(struct request_queue *q);
3edcc0ce	40	void blk_mq_free_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);
2c3ad667	44	void blk_mq_flush_busy_ctxs(struct blk_mq_hw_ctx hctx, struct list_head list);
8ab6bb9e	45	bool blk_mq_get_driver_tag(struct request *rq);
b347689f ML	46	struct request blk_mq_dequeue_from_ctx(struct blk_mq_hw_ctx hctx,
b347689f ML	47	struct blk_mq_ctx *start);
2c3ad667 JA	48
	49	/*
	50	* Internal helpers for allocating/freeing the request map
	51	*/
cc71a6f4 JA	52	void blk_mq_free_rqs(struct blk_mq_tag_set set, struct blk_mq_tags tags,
	53	unsigned int hctx_idx);
	54	void blk_mq_free_rq_map(struct blk_mq_tags *tags);
	55	struct blk_mq_tags blk_mq_alloc_rq_map(struct blk_mq_tag_set set,
	56	unsigned int hctx_idx,
	57	unsigned int nr_tags,
	58	unsigned int reserved_tags);
	59	int blk_mq_alloc_rqs(struct blk_mq_tag_set set, struct blk_mq_tags tags,
	60	unsigned int hctx_idx, unsigned int depth);
2c3ad667 JA	61
	62	/*
	63	* Internal helpers for request insertion into sw queues
	64	*/
	65	void __blk_mq_insert_request(struct blk_mq_hw_ctx hctx, struct request rq,
	66	bool at_head);
b0850297	67	void blk_mq_request_bypass_insert(struct request *rq, bool run_queue);
bd166ef1 JA	68	void blk_mq_insert_requests(struct blk_mq_hw_ctx hctx, struct blk_mq_ctx ctx,
bd166ef1 JA	69	struct list_head *list);
320ae51f	70
d6a51a97 JW	71	blk_status_t blk_mq_try_issue_directly(struct blk_mq_hw_ctx *hctx,
	72	struct request *rq,
	73	blk_qc_t *cookie,
	74	bool bypass, 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
	100	* @cpu: CPU
	101	*/
	102	static inline struct blk_mq_hw_ctx blk_mq_map_queue(struct request_queue q,
	103	unsigned int flags,
	104	unsigned int cpu)
ff2c5660	105	{
e20ba6e1 CH	106	enum hctx_type type = HCTX_TYPE_DEFAULT;
e20ba6e1 CH	107
5aceaeb2 CH	108	if ((flags & REQ_HIPRI) &&
	109	q->tag_set->nr_maps > HCTX_TYPE_POLL &&
	110	q->tag_set->map[HCTX_TYPE_POLL].nr_queues &&
	111	test_bit(QUEUE_FLAG_POLL, &q->queue_flags))
e20ba6e1	112	type = HCTX_TYPE_POLL;
b3c661b1	113
5aceaeb2 CH	114	else if (((flags & REQ_OP_MASK) == REQ_OP_READ) &&
	115	q->tag_set->nr_maps > HCTX_TYPE_READ &&
	116	q->tag_set->map[HCTX_TYPE_READ].nr_queues)
e20ba6e1	117	type = HCTX_TYPE_READ;
5aceaeb2	118
e20ba6e1	119	return blk_mq_map_queue_type(q, type, cpu);
ff2c5660 JA	120	}
ff2c5660 JA	121
67aec14c JA	122	/*
	123	* sysfs helpers
	124	*/
737f98cf	125	extern void blk_mq_sysfs_init(struct request_queue *q);
7ea5fe31	126	extern void blk_mq_sysfs_deinit(struct request_queue *q);
2d0364c8	127	extern int __blk_mq_register_dev(struct device dev, struct request_queue q);
67aec14c JA	128	extern int blk_mq_sysfs_register(struct request_queue *q);
67aec14c JA	129	extern void blk_mq_sysfs_unregister(struct request_queue *q);
868f2f0b	130	extern void blk_mq_hctx_kobj_init(struct blk_mq_hw_ctx *hctx);
67aec14c	131
e09aae7e ML	132	void blk_mq_release(struct request_queue *q);
e09aae7e ML	133
1d9bd516 TH	134	/**
	135	* blk_mq_rq_state() - read the current MQ_RQ_* state of a request
	136	* @rq: target request.
	137	*/
12f5b931	138	static inline enum mq_rq_state blk_mq_rq_state(struct request *rq)
1d9bd516	139	{
12f5b931	140	return READ_ONCE(rq->state);
1d9bd516 TH	141	}
1d9bd516 TH	142
1aecfe48 ML	143	static inline struct blk_mq_ctx __blk_mq_get_ctx(struct request_queue q,
	144	unsigned int cpu)
	145	{
	146	return per_cpu_ptr(q->queue_ctx, cpu);
	147	}
	148
	149	/*
	150	* This assumes per-cpu software queueing queues. They could be per-node
	151	* as well, for instance. For now this is hardcoded as-is. Note that we don't
	152	* care about preemption, since we know the ctx's are persistent. This does
	153	* mean that we can't rely on ctx always matching the currently running CPU.
	154	*/
	155	static inline struct blk_mq_ctx blk_mq_get_ctx(struct request_queue q)
	156	{
	157	return __blk_mq_get_ctx(q, get_cpu());
	158	}
	159
	160	static inline void blk_mq_put_ctx(struct blk_mq_ctx *ctx)
	161	{
	162	put_cpu();
	163	}
	164
cb96a42c ML	165	struct blk_mq_alloc_data {
	166	/* input parameter */
	167	struct request_queue *q;
9a95e4ef	168	blk_mq_req_flags_t flags;
229a9287	169	unsigned int shallow_depth;
f9afca4d	170	unsigned int cmd_flags;
cb96a42c ML	171
	172	/* input & output parameter */
	173	struct blk_mq_ctx *ctx;
	174	struct blk_mq_hw_ctx *hctx;
	175	};
	176
4941115b JA	177	static inline struct blk_mq_tags blk_mq_tags_from_data(struct blk_mq_alloc_data data)
4941115b JA	178	{
bd166ef1 JA	179	if (data->flags & BLK_MQ_REQ_INTERNAL)
	180	return data->hctx->sched_tags;
	181
4941115b JA	182	return data->hctx->tags;
	183	}
	184
5d1b25c1 BVA	185	static inline bool blk_mq_hctx_stopped(struct blk_mq_hw_ctx *hctx)
	186	{
	187	return test_bit(BLK_MQ_S_STOPPED, &hctx->state);
	188	}
	189
19c66e59 ML	190	static inline bool blk_mq_hw_queue_mapped(struct blk_mq_hw_ctx *hctx)
	191	{
	192	return hctx->nr_ctx && hctx->tags;
	193	}
	194
e016b782	195	unsigned int blk_mq_in_flight(struct request_queue q, struct hd_struct part);
bf0ddaba OS	196	void blk_mq_in_flight_rw(struct request_queue q, struct hd_struct part,
bf0ddaba OS	197	unsigned int inflight[2]);
f299b7c7	198
de148297 ML	199	static inline void blk_mq_put_dispatch_budget(struct blk_mq_hw_ctx *hctx)
	200	{
	201	struct request_queue *q = hctx->queue;
	202
	203	if (q->mq_ops->put_budget)
	204	q->mq_ops->put_budget(hctx);
	205	}
	206
88022d72	207	static inline bool blk_mq_get_dispatch_budget(struct blk_mq_hw_ctx *hctx)
de148297 ML	208	{
	209	struct request_queue *q = hctx->queue;
	210
	211	if (q->mq_ops->get_budget)
	212	return q->mq_ops->get_budget(hctx);
88022d72	213	return true;
de148297 ML	214	}
de148297 ML	215
244c65a3 ML	216	static inline void __blk_mq_put_driver_tag(struct blk_mq_hw_ctx *hctx,
	217	struct request *rq)
	218	{
	219	blk_mq_put_tag(hctx, hctx->tags, rq->mq_ctx, rq->tag);
	220	rq->tag = -1;
	221
	222	if (rq->rq_flags & RQF_MQ_INFLIGHT) {
	223	rq->rq_flags &= ~RQF_MQ_INFLIGHT;
	224	atomic_dec(&hctx->nr_active);
	225	}
	226	}
	227
	228	static inline void blk_mq_put_driver_tag_hctx(struct blk_mq_hw_ctx *hctx,
	229	struct request *rq)
	230	{
	231	if (rq->tag == -1 \|\| rq->internal_tag == -1)
	232	return;
	233
	234	__blk_mq_put_driver_tag(hctx, rq);
	235	}
	236
	237	static inline void blk_mq_put_driver_tag(struct request *rq)
	238	{
244c65a3 ML	239	if (rq->tag == -1 \|\| rq->internal_tag == -1)
	240	return;
	241
ea4f995e	242	__blk_mq_put_driver_tag(rq->mq_hctx, rq);
244c65a3 ML	243	}
244c65a3 ML	244
ed76e329	245	static inline void blk_mq_clear_mq_map(struct blk_mq_queue_map *qmap)
0da73d00 MI	246	{
	247	int cpu;
	248
	249	for_each_possible_cpu(cpu)
ed76e329	250	qmap->mq_map[cpu] = 0;
0da73d00 MI	251	}
0da73d00 MI	252
320ae51f	253	#endif