[linux-block.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_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 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;

	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 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);

/**
 * 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];
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
780db207	40	void blk_mq_freeze_queue(struct request_queue *q);
3edcc0ce	41	void blk_mq_free_queue(struct request_queue *q);
e3a2b3f9	42	int blk_mq_update_nr_requests(struct request_queue *q, unsigned int nr);
aed3ea94	43	void blk_mq_wake_waiters(struct request_queue *q);
de148297	44	bool blk_mq_dispatch_rq_list(struct request_queue , struct list_head , bool);
2c3ad667	45	void blk_mq_flush_busy_ctxs(struct blk_mq_hw_ctx hctx, struct list_head list);
8ab6bb9e	46	bool blk_mq_get_driver_tag(struct request *rq);
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);
b0850297	68	void blk_mq_request_bypass_insert(struct request *rq, bool run_queue);
bd166ef1 JA	69	void blk_mq_insert_requests(struct blk_mq_hw_ctx hctx, struct blk_mq_ctx ctx,
bd166ef1 JA	70	struct list_head *list);
320ae51f	71
d6a51a97 JW	72	blk_status_t blk_mq_try_issue_directly(struct blk_mq_hw_ctx *hctx,
	73	struct request *rq,
	74	blk_qc_t *cookie,
	75	bool bypass, bool last);
6ce3dd6e ML	76	void blk_mq_try_issue_list_directly(struct blk_mq_hw_ctx *hctx,
6ce3dd6e ML	77	struct list_head *list);
396eaf21	78
320ae51f JA	79	/*
	80	* CPU -> queue mappings
	81	*/
ed76e329	82	extern int blk_mq_hw_queue_to_node(struct blk_mq_queue_map *qmap, unsigned int);
320ae51f	83
b3c661b1 JA	84	/*
	85	* blk_mq_map_queue_type() - map (hctx_type,cpu) to hardware queue
	86	* @q: request queue
e20ba6e1	87	* @type: the hctx type index
b3c661b1 JA	88	* @cpu: CPU
	89	*/
	90	static inline struct blk_mq_hw_ctx blk_mq_map_queue_type(struct request_queue q,
e20ba6e1	91	enum hctx_type type,
b3c661b1	92	unsigned int cpu)
7d7e0f90	93	{
e20ba6e1	94	return q->queue_hw_ctx[q->tag_set->map[type].mq_map[cpu]];
7d7e0f90 CH	95	}
7d7e0f90 CH	96
b3c661b1 JA	97	/*
	98	* blk_mq_map_queue() - map (cmd_flags,type) to hardware queue
	99	* @q: request queue
	100	* @flags: request command flags
8ccdf4a3	101	* @cpu: cpu ctx
b3c661b1 JA	102	*/
	103	static inline struct blk_mq_hw_ctx blk_mq_map_queue(struct request_queue q,
	104	unsigned int flags,
8ccdf4a3	105	struct blk_mq_ctx *ctx)
ff2c5660	106	{
e20ba6e1 CH	107	enum hctx_type type = HCTX_TYPE_DEFAULT;
e20ba6e1 CH	108
5aceaeb2 CH	109	if ((flags & REQ_HIPRI) &&
	110	q->tag_set->nr_maps > HCTX_TYPE_POLL &&
	111	q->tag_set->map[HCTX_TYPE_POLL].nr_queues &&
	112	test_bit(QUEUE_FLAG_POLL, &q->queue_flags))
e20ba6e1	113	type = HCTX_TYPE_POLL;
b3c661b1	114
5aceaeb2 CH	115	else if (((flags & REQ_OP_MASK) == REQ_OP_READ) &&
	116	q->tag_set->nr_maps > HCTX_TYPE_READ &&
	117	q->tag_set->map[HCTX_TYPE_READ].nr_queues)
e20ba6e1	118	type = HCTX_TYPE_READ;
5aceaeb2	119
8ccdf4a3	120	return ctx->hctxs[type];
ff2c5660 JA	121	}
ff2c5660 JA	122
67aec14c JA	123	/*
	124	* sysfs helpers
	125	*/
737f98cf	126	extern void blk_mq_sysfs_init(struct request_queue *q);
7ea5fe31	127	extern void blk_mq_sysfs_deinit(struct request_queue *q);
2d0364c8	128	extern int __blk_mq_register_dev(struct device dev, struct request_queue q);
67aec14c JA	129	extern int blk_mq_sysfs_register(struct request_queue *q);
67aec14c JA	130	extern void blk_mq_sysfs_unregister(struct request_queue *q);
868f2f0b	131	extern void blk_mq_hctx_kobj_init(struct blk_mq_hw_ctx *hctx);
67aec14c	132
e09aae7e ML	133	void blk_mq_release(struct request_queue *q);
e09aae7e ML	134
1d9bd516 TH	135	/**
	136	* blk_mq_rq_state() - read the current MQ_RQ_* state of a request
	137	* @rq: target request.
	138	*/
12f5b931	139	static inline enum mq_rq_state blk_mq_rq_state(struct request *rq)
1d9bd516	140	{
12f5b931	141	return READ_ONCE(rq->state);
1d9bd516 TH	142	}
1d9bd516 TH	143
1aecfe48 ML	144	static inline struct blk_mq_ctx __blk_mq_get_ctx(struct request_queue q,
	145	unsigned int cpu)
	146	{
	147	return per_cpu_ptr(q->queue_ctx, cpu);
	148	}
	149
	150	/*
	151	* This assumes per-cpu software queueing queues. They could be per-node
	152	* as well, for instance. For now this is hardcoded as-is. Note that we don't
	153	* care about preemption, since we know the ctx's are persistent. This does
	154	* mean that we can't rely on ctx always matching the currently running CPU.
	155	*/
	156	static inline struct blk_mq_ctx blk_mq_get_ctx(struct request_queue q)
	157	{
	158	return __blk_mq_get_ctx(q, get_cpu());
	159	}
	160
	161	static inline void blk_mq_put_ctx(struct blk_mq_ctx *ctx)
	162	{
	163	put_cpu();
	164	}
	165
cb96a42c ML	166	struct blk_mq_alloc_data {
	167	/* input parameter */
	168	struct request_queue *q;
9a95e4ef	169	blk_mq_req_flags_t flags;
229a9287	170	unsigned int shallow_depth;
f9afca4d	171	unsigned int cmd_flags;
cb96a42c ML	172
	173	/* input & output parameter */
	174	struct blk_mq_ctx *ctx;
	175	struct blk_mq_hw_ctx *hctx;
	176	};
	177
4941115b JA	178	static inline struct blk_mq_tags blk_mq_tags_from_data(struct blk_mq_alloc_data data)
4941115b JA	179	{
bd166ef1 JA	180	if (data->flags & BLK_MQ_REQ_INTERNAL)
	181	return data->hctx->sched_tags;
	182
4941115b JA	183	return data->hctx->tags;
	184	}
	185
5d1b25c1 BVA	186	static inline bool blk_mq_hctx_stopped(struct blk_mq_hw_ctx *hctx)
	187	{
	188	return test_bit(BLK_MQ_S_STOPPED, &hctx->state);
	189	}
	190
19c66e59 ML	191	static inline bool blk_mq_hw_queue_mapped(struct blk_mq_hw_ctx *hctx)
	192	{
	193	return hctx->nr_ctx && hctx->tags;
	194	}
	195
e016b782	196	unsigned int blk_mq_in_flight(struct request_queue q, struct hd_struct part);
bf0ddaba OS	197	void blk_mq_in_flight_rw(struct request_queue q, struct hd_struct part,
bf0ddaba OS	198	unsigned int inflight[2]);
f299b7c7	199
de148297 ML	200	static inline void blk_mq_put_dispatch_budget(struct blk_mq_hw_ctx *hctx)
	201	{
	202	struct request_queue *q = hctx->queue;
	203
	204	if (q->mq_ops->put_budget)
	205	q->mq_ops->put_budget(hctx);
	206	}
	207
88022d72	208	static inline bool blk_mq_get_dispatch_budget(struct blk_mq_hw_ctx *hctx)
de148297 ML	209	{
	210	struct request_queue *q = hctx->queue;
	211
	212	if (q->mq_ops->get_budget)
	213	return q->mq_ops->get_budget(hctx);
88022d72	214	return true;
de148297 ML	215	}
de148297 ML	216
244c65a3 ML	217	static inline void __blk_mq_put_driver_tag(struct blk_mq_hw_ctx *hctx,
	218	struct request *rq)
	219	{
	220	blk_mq_put_tag(hctx, hctx->tags, rq->mq_ctx, rq->tag);
	221	rq->tag = -1;
	222
	223	if (rq->rq_flags & RQF_MQ_INFLIGHT) {
	224	rq->rq_flags &= ~RQF_MQ_INFLIGHT;
	225	atomic_dec(&hctx->nr_active);
	226	}
	227	}
	228
	229	static inline void blk_mq_put_driver_tag_hctx(struct blk_mq_hw_ctx *hctx,
	230	struct request *rq)
	231	{
	232	if (rq->tag == -1 \|\| rq->internal_tag == -1)
	233	return;
	234
	235	__blk_mq_put_driver_tag(hctx, rq);
	236	}
	237
	238	static inline void blk_mq_put_driver_tag(struct request *rq)
	239	{
244c65a3 ML	240	if (rq->tag == -1 \|\| rq->internal_tag == -1)
	241	return;
	242
ea4f995e	243	__blk_mq_put_driver_tag(rq->mq_hctx, rq);
244c65a3 ML	244	}
244c65a3 ML	245
ed76e329	246	static inline void blk_mq_clear_mq_map(struct blk_mq_queue_map *qmap)
0da73d00 MI	247	{
	248	int cpu;
	249
	250	for_each_possible_cpu(cpu)
ed76e329	251	qmap->mq_map[cpu] = 0;
0da73d00 MI	252	}
0da73d00 MI	253
320ae51f	254	#endif