[linux-2.6-block.git] / include / linux / blk-mq.h

#ifndef BLK_MQ_H
#define BLK_MQ_H

#include <linux/blkdev.h>

struct blk_mq_tags;
struct blk_flush_queue;

struct blk_mq_cpu_notifier {
	struct list_head list;
	void *data;
	int (*notify)(void *data, unsigned long action, unsigned int cpu);
};

struct blk_mq_ctxmap {
	unsigned int size;
	unsigned int bits_per_word;
	struct blk_align_bitmap *map;
};

struct blk_mq_hw_ctx {
	struct {
		spinlock_t		lock;
		struct list_head	dispatch;
		unsigned long		state;		/* BLK_MQ_S_* flags */
	} ____cacheline_aligned_in_smp;

	struct work_struct	run_work;
	cpumask_var_t		cpumask;
	int			next_cpu;
	int			next_cpu_batch;

	unsigned long		flags;		/* BLK_MQ_F_* flags */

	struct request_queue	*queue;
	struct blk_flush_queue	*fq;

	void			*driver_data;

	struct blk_mq_ctxmap	ctx_map;

	struct blk_mq_ctx	**ctxs;
	unsigned int		nr_ctx;

	atomic_t		wait_index;

	struct blk_mq_tags	*tags;

	unsigned long		queued;
	unsigned long		run;
#define BLK_MQ_MAX_DISPATCH_ORDER	7
	unsigned long		dispatched[BLK_MQ_MAX_DISPATCH_ORDER];

	unsigned int		numa_node;
	unsigned int		queue_num;

	atomic_t		nr_active;

	struct delayed_work	delay_work;

	struct blk_mq_cpu_notifier	cpu_notifier;
	struct kobject		kobj;

	unsigned long		poll_considered;
	unsigned long		poll_invoked;
	unsigned long		poll_success;
};

struct blk_mq_tag_set {
	unsigned int		*mq_map;
	struct blk_mq_ops	*ops;
	unsigned int		nr_hw_queues;
	unsigned int		queue_depth;	/* max hw supported */
	unsigned int		reserved_tags;
	unsigned int		cmd_size;	/* per-request extra data */
	int			numa_node;
	unsigned int		timeout;
	unsigned int		flags;		/* BLK_MQ_F_* */
	void			*driver_data;

	struct blk_mq_tags	**tags;

	struct mutex		tag_list_lock;
	struct list_head	tag_list;
};

struct blk_mq_queue_data {
	struct request *rq;
	struct list_head *list;
	bool last;
};

typedef int (queue_rq_fn)(struct blk_mq_hw_ctx *, const struct blk_mq_queue_data *);
typedef enum blk_eh_timer_return (timeout_fn)(struct request *, bool);
typedef int (init_hctx_fn)(struct blk_mq_hw_ctx *, void *, unsigned int);
typedef void (exit_hctx_fn)(struct blk_mq_hw_ctx *, unsigned int);
typedef int (init_request_fn)(void *, struct request *, unsigned int,
		unsigned int, unsigned int);
typedef void (exit_request_fn)(void *, struct request *, unsigned int,
		unsigned int);
typedef int (reinit_request_fn)(void *, struct request *);

typedef void (busy_iter_fn)(struct blk_mq_hw_ctx *, struct request *, void *,
		bool);
typedef void (busy_tag_iter_fn)(struct request *, void *, bool);
typedef int (poll_fn)(struct blk_mq_hw_ctx *, unsigned int);


struct blk_mq_ops {
	/*
	 * Queue request
	 */
	queue_rq_fn		*queue_rq;

	/*
	 * Called on request timeout
	 */
	timeout_fn		*timeout;

	/*
	 * Called to poll for completion of a specific tag.
	 */
	poll_fn			*poll;

	softirq_done_fn		*complete;

	/*
	 * Called when the block layer side of a hardware queue has been
	 * set up, allowing the driver to allocate/init matching structures.
	 * Ditto for exit/teardown.
	 */
	init_hctx_fn		*init_hctx;
	exit_hctx_fn		*exit_hctx;

	/*
	 * Called for every command allocated by the block layer to allow
	 * the driver to set up driver specific data.
	 *
	 * Tag greater than or equal to queue_depth is for setting up
	 * flush request.
	 *
	 * Ditto for exit/teardown.
	 */
	init_request_fn		*init_request;
	exit_request_fn		*exit_request;
	reinit_request_fn	*reinit_request;
};

enum {
	BLK_MQ_RQ_QUEUE_OK	= 0,	/* queued fine */
	BLK_MQ_RQ_QUEUE_BUSY	= 1,	/* requeue IO for later */
	BLK_MQ_RQ_QUEUE_ERROR	= 2,	/* end IO with error */

	BLK_MQ_F_SHOULD_MERGE	= 1 << 0,
	BLK_MQ_F_TAG_SHARED	= 1 << 1,
	BLK_MQ_F_SG_MERGE	= 1 << 2,
	BLK_MQ_F_DEFER_ISSUE	= 1 << 4,
	BLK_MQ_F_ALLOC_POLICY_START_BIT = 8,
	BLK_MQ_F_ALLOC_POLICY_BITS = 1,

	BLK_MQ_S_STOPPED	= 0,
	BLK_MQ_S_TAG_ACTIVE	= 1,

	BLK_MQ_MAX_DEPTH	= 10240,

	BLK_MQ_CPU_WORK_BATCH	= 8,
};
#define BLK_MQ_FLAG_TO_ALLOC_POLICY(flags) \
	((flags >> BLK_MQ_F_ALLOC_POLICY_START_BIT) & \
		((1 << BLK_MQ_F_ALLOC_POLICY_BITS) - 1))
#define BLK_ALLOC_POLICY_TO_MQ_FLAG(policy) \
	((policy & ((1 << BLK_MQ_F_ALLOC_POLICY_BITS) - 1)) \
		<< BLK_MQ_F_ALLOC_POLICY_START_BIT)

struct request_queue *blk_mq_init_queue(struct blk_mq_tag_set *);
struct request_queue *blk_mq_init_allocated_queue(struct blk_mq_tag_set *set,
						  struct request_queue *q);
int blk_mq_register_disk(struct gendisk *);
void blk_mq_unregister_disk(struct gendisk *);

int blk_mq_alloc_tag_set(struct blk_mq_tag_set *set);
void blk_mq_free_tag_set(struct blk_mq_tag_set *set);

void blk_mq_flush_plug_list(struct blk_plug *plug, bool from_schedule);

void blk_mq_insert_request(struct request *, bool, bool, bool);
void blk_mq_free_request(struct request *rq);
void blk_mq_free_hctx_request(struct blk_mq_hw_ctx *, struct request *rq);
bool blk_mq_can_queue(struct blk_mq_hw_ctx *);

enum {
	BLK_MQ_REQ_NOWAIT	= (1 << 0), /* return when out of requests */
	BLK_MQ_REQ_RESERVED	= (1 << 1), /* allocate from reserved pool */
};

struct request *blk_mq_alloc_request(struct request_queue *q, int rw,
		unsigned int flags);
struct request *blk_mq_alloc_request_hctx(struct request_queue *q, int op,
		unsigned int flags, unsigned int hctx_idx);
struct request *blk_mq_tag_to_rq(struct blk_mq_tags *tags, unsigned int tag);
struct cpumask *blk_mq_tags_cpumask(struct blk_mq_tags *tags);

enum {
	BLK_MQ_UNIQUE_TAG_BITS = 16,
	BLK_MQ_UNIQUE_TAG_MASK = (1 << BLK_MQ_UNIQUE_TAG_BITS) - 1,
};

u32 blk_mq_unique_tag(struct request *rq);

static inline u16 blk_mq_unique_tag_to_hwq(u32 unique_tag)
{
	return unique_tag >> BLK_MQ_UNIQUE_TAG_BITS;
}

static inline u16 blk_mq_unique_tag_to_tag(u32 unique_tag)
{
	return unique_tag & BLK_MQ_UNIQUE_TAG_MASK;
}


int blk_mq_request_started(struct request *rq);
void blk_mq_start_request(struct request *rq);
void blk_mq_end_request(struct request *rq, int error);
void __blk_mq_end_request(struct request *rq, int error);

void blk_mq_requeue_request(struct request *rq);
void blk_mq_add_to_requeue_list(struct request *rq, bool at_head);
void blk_mq_cancel_requeue_work(struct request_queue *q);
void blk_mq_kick_requeue_list(struct request_queue *q);
void blk_mq_delay_kick_requeue_list(struct request_queue *q, unsigned long msecs);
void blk_mq_abort_requeue_list(struct request_queue *q);
void blk_mq_complete_request(struct request *rq, int error);

void blk_mq_stop_hw_queue(struct blk_mq_hw_ctx *hctx);
void blk_mq_start_hw_queue(struct blk_mq_hw_ctx *hctx);
void blk_mq_stop_hw_queues(struct request_queue *q);
void blk_mq_start_hw_queues(struct request_queue *q);
void blk_mq_start_stopped_hw_queues(struct request_queue *q, bool async);
void blk_mq_run_hw_queues(struct request_queue *q, bool async);
void blk_mq_delay_queue(struct blk_mq_hw_ctx *hctx, unsigned long msecs);
void blk_mq_tagset_busy_iter(struct blk_mq_tag_set *tagset,
		busy_tag_iter_fn *fn, void *priv);
void blk_mq_freeze_queue(struct request_queue *q);
void blk_mq_unfreeze_queue(struct request_queue *q);
void blk_mq_freeze_queue_start(struct request_queue *q);
int blk_mq_reinit_tagset(struct blk_mq_tag_set *set);

void blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, int nr_hw_queues);

/*
 * Driver command data is immediately after the request. So subtract request
 * size to get back to the original request, add request size to get the PDU.
 */
static inline struct request *blk_mq_rq_from_pdu(void *pdu)
{
	return pdu - sizeof(struct request);
}
static inline void *blk_mq_rq_to_pdu(struct request *rq)
{
	return rq + 1;
}

#define queue_for_each_hw_ctx(q, hctx, i)				\
	for ((i) = 0; (i) < (q)->nr_hw_queues &&			\
	     ({ hctx = (q)->queue_hw_ctx[i]; 1; }); (i)++)

#define hctx_for_each_ctx(hctx, ctx, i)					\
	for ((i) = 0; (i) < (hctx)->nr_ctx &&				\
	     ({ ctx = (hctx)->ctxs[(i)]; 1; }); (i)++)

#endif
Commit	Line	Data
320ae51f JA	1	#ifndef BLK_MQ_H
	2	#define BLK_MQ_H
	3
	4	#include <linux/blkdev.h>
	5
	6	struct blk_mq_tags;
f70ced09	7	struct blk_flush_queue;
320ae51f JA	8
	9	struct blk_mq_cpu_notifier {
	10	struct list_head list;
	11	void *data;
e814e71b	12	int (notify)(void data, unsigned long action, unsigned int cpu);
320ae51f JA	13	};
320ae51f JA	14
1429d7c9	15	struct blk_mq_ctxmap {
569fd0ce	16	unsigned int size;
1429d7c9 JA	17	unsigned int bits_per_word;
	18	struct blk_align_bitmap *map;
	19	};
	20
320ae51f JA	21	struct blk_mq_hw_ctx {
	22	struct {
	23	spinlock_t lock;
	24	struct list_head dispatch;
8d354f13	25	unsigned long state; /* BLK_MQ_S_* flags */
320ae51f JA	26	} ____cacheline_aligned_in_smp;
320ae51f JA	27
27489a3c	28	struct work_struct run_work;
e4043dcf	29	cpumask_var_t cpumask;
506e931f JA	30	int next_cpu;
506e931f JA	31	int next_cpu_batch;
320ae51f JA	32
	33	unsigned long flags; /* BLK_MQ_F_* flags */
	34
	35	struct request_queue *queue;
f70ced09	36	struct blk_flush_queue *fq;
320ae51f JA	37
	38	void *driver_data;
	39
1429d7c9 JA	40	struct blk_mq_ctxmap ctx_map;
1429d7c9 JA	41
4bb659b1	42	struct blk_mq_ctx **ctxs;
8d354f13	43	unsigned int nr_ctx;
4bb659b1	44
8537b120	45	atomic_t wait_index;
320ae51f	46
320ae51f JA	47	struct blk_mq_tags *tags;
	48
	49	unsigned long queued;
	50	unsigned long run;
8d354f13	51	#define BLK_MQ_MAX_DISPATCH_ORDER 7
320ae51f JA	52	unsigned long dispatched[BLK_MQ_MAX_DISPATCH_ORDER];
320ae51f JA	53
320ae51f	54	unsigned int numa_node;
17ded320	55	unsigned int queue_num;
320ae51f	56
0d2602ca JA	57	atomic_t nr_active;
0d2602ca JA	58
8d354f13 JA	59	struct delayed_work delay_work;
8d354f13 JA	60
320ae51f JA	61	struct blk_mq_cpu_notifier cpu_notifier;
320ae51f JA	62	struct kobject kobj;
05229bee	63
6e219353	64	unsigned long poll_considered;
05229bee JA	65	unsigned long poll_invoked;
05229bee JA	66	unsigned long poll_success;
320ae51f JA	67	};
320ae51f JA	68
24d2f903	69	struct blk_mq_tag_set {
bdd17e75	70	unsigned int *mq_map;
320ae51f JA	71	struct blk_mq_ops *ops;
320ae51f JA	72	unsigned int nr_hw_queues;
e3a2b3f9	73	unsigned int queue_depth; /* max hw supported */
320ae51f JA	74	unsigned int reserved_tags;
	75	unsigned int cmd_size; /* per-request extra data */
	76	int numa_node;
	77	unsigned int timeout;
	78	unsigned int flags; /* BLK_MQ_F_* */
24d2f903 CH	79	void *driver_data;
	80
	81	struct blk_mq_tags **tags;
0d2602ca JA	82
	83	struct mutex tag_list_lock;
	84	struct list_head tag_list;
320ae51f JA	85	};
320ae51f JA	86
74c45052 JA	87	struct blk_mq_queue_data {
	88	struct request *rq;
	89	struct list_head *list;
	90	bool last;
	91	};
	92
	93	typedef int (queue_rq_fn)(struct blk_mq_hw_ctx , const struct blk_mq_queue_data );
0152fb6b	94	typedef enum blk_eh_timer_return (timeout_fn)(struct request *, bool);
320ae51f JA	95	typedef int (init_hctx_fn)(struct blk_mq_hw_ctx , void , unsigned int);
320ae51f JA	96	typedef void (exit_hctx_fn)(struct blk_mq_hw_ctx *, unsigned int);
24d2f903 CH	97	typedef int (init_request_fn)(void , struct request , unsigned int,
	98	unsigned int, unsigned int);
	99	typedef void (exit_request_fn)(void , struct request , unsigned int,
	100	unsigned int);
486cf989	101	typedef int (reinit_request_fn)(void , struct request );
320ae51f	102
81481eb4 CH	103	typedef void (busy_iter_fn)(struct blk_mq_hw_ctx , struct request , void *,
81481eb4 CH	104	bool);
f26cdc85	105	typedef void (busy_tag_iter_fn)(struct request , void , bool);
05229bee JA	106	typedef int (poll_fn)(struct blk_mq_hw_ctx *, unsigned int);
05229bee JA	107
81481eb4	108
320ae51f JA	109	struct blk_mq_ops {
	110	/*
	111	* Queue request
	112	*/
	113	queue_rq_fn *queue_rq;
	114
320ae51f JA	115	/*
	116	* Called on request timeout
	117	*/
0152fb6b	118	timeout_fn *timeout;
320ae51f	119
05229bee JA	120	/*
	121	* Called to poll for completion of a specific tag.
	122	*/
	123	poll_fn *poll;
	124
30a91cb4 CH	125	softirq_done_fn *complete;
30a91cb4 CH	126
320ae51f JA	127	/*
	128	* Called when the block layer side of a hardware queue has been
	129	* set up, allowing the driver to allocate/init matching structures.
	130	* Ditto for exit/teardown.
	131	*/
	132	init_hctx_fn *init_hctx;
	133	exit_hctx_fn *exit_hctx;
e9b267d9 CH	134
	135	/*
	136	* Called for every command allocated by the block layer to allow
	137	* the driver to set up driver specific data.
f70ced09 ML	138	*
	139	* Tag greater than or equal to queue_depth is for setting up
	140	* flush request.
	141	*
e9b267d9 CH	142	* Ditto for exit/teardown.
	143	*/
	144	init_request_fn *init_request;
	145	exit_request_fn *exit_request;
486cf989	146	reinit_request_fn *reinit_request;
320ae51f JA	147	};
	148
	149	enum {
	150	BLK_MQ_RQ_QUEUE_OK = 0, /* queued fine */
	151	BLK_MQ_RQ_QUEUE_BUSY = 1, /* requeue IO for later */
	152	BLK_MQ_RQ_QUEUE_ERROR = 2, /* end IO with error */
	153
	154	BLK_MQ_F_SHOULD_MERGE = 1 << 0,
8a58d1f1 JA	155	BLK_MQ_F_TAG_SHARED = 1 << 1,
8a58d1f1 JA	156	BLK_MQ_F_SG_MERGE = 1 << 2,
e167dfb5	157	BLK_MQ_F_DEFER_ISSUE = 1 << 4,
24391c0d SL	158	BLK_MQ_F_ALLOC_POLICY_START_BIT = 8,
24391c0d SL	159	BLK_MQ_F_ALLOC_POLICY_BITS = 1,
320ae51f	160
5d12f905	161	BLK_MQ_S_STOPPED = 0,
0d2602ca	162	BLK_MQ_S_TAG_ACTIVE = 1,
320ae51f	163
a4391c64	164	BLK_MQ_MAX_DEPTH = 10240,
506e931f JA	165
506e931f JA	166	BLK_MQ_CPU_WORK_BATCH = 8,
320ae51f	167	};
24391c0d SL	168	#define BLK_MQ_FLAG_TO_ALLOC_POLICY(flags) \
	169	((flags >> BLK_MQ_F_ALLOC_POLICY_START_BIT) & \
	170	((1 << BLK_MQ_F_ALLOC_POLICY_BITS) - 1))
	171	#define BLK_ALLOC_POLICY_TO_MQ_FLAG(policy) \
	172	((policy & ((1 << BLK_MQ_F_ALLOC_POLICY_BITS) - 1)) \
	173	<< BLK_MQ_F_ALLOC_POLICY_START_BIT)
320ae51f	174
24d2f903	175	struct request_queue blk_mq_init_queue(struct blk_mq_tag_set );
b62c21b7 MS	176	struct request_queue blk_mq_init_allocated_queue(struct blk_mq_tag_set set,
b62c21b7 MS	177	struct request_queue *q);
320ae51f JA	178	int blk_mq_register_disk(struct gendisk *);
320ae51f JA	179	void blk_mq_unregister_disk(struct gendisk *);
320ae51f	180
24d2f903 CH	181	int blk_mq_alloc_tag_set(struct blk_mq_tag_set *set);
	182	void blk_mq_free_tag_set(struct blk_mq_tag_set *set);
	183
320ae51f JA	184	void blk_mq_flush_plug_list(struct blk_plug *plug, bool from_schedule);
320ae51f JA	185
feb71dae	186	void blk_mq_insert_request(struct request *, bool, bool, bool);
320ae51f	187	void blk_mq_free_request(struct request *rq);
7c7f2f2b	188	void blk_mq_free_hctx_request(struct blk_mq_hw_ctx , struct request rq);
320ae51f	189	bool blk_mq_can_queue(struct blk_mq_hw_ctx *);
6f3b0e8b CH	190
	191	enum {
	192	BLK_MQ_REQ_NOWAIT = (1 << 0), /* return when out of requests */
	193	BLK_MQ_REQ_RESERVED = (1 << 1), /* allocate from reserved pool */
	194	};
	195
4ce01dd1	196	struct request blk_mq_alloc_request(struct request_queue q, int rw,
6f3b0e8b	197	unsigned int flags);
1f5bd336 ML	198	struct request blk_mq_alloc_request_hctx(struct request_queue q, int op,
1f5bd336 ML	199	unsigned int flags, unsigned int hctx_idx);
0e62f51f	200	struct request blk_mq_tag_to_rq(struct blk_mq_tags tags, unsigned int tag);
f26cdc85	201	struct cpumask blk_mq_tags_cpumask(struct blk_mq_tags tags);
320ae51f	202
205fb5f5 BVA	203	enum {
	204	BLK_MQ_UNIQUE_TAG_BITS = 16,
	205	BLK_MQ_UNIQUE_TAG_MASK = (1 << BLK_MQ_UNIQUE_TAG_BITS) - 1,
	206	};
	207
	208	u32 blk_mq_unique_tag(struct request *rq);
	209
	210	static inline u16 blk_mq_unique_tag_to_hwq(u32 unique_tag)
	211	{
	212	return unique_tag >> BLK_MQ_UNIQUE_TAG_BITS;
	213	}
	214
	215	static inline u16 blk_mq_unique_tag_to_tag(u32 unique_tag)
	216	{
	217	return unique_tag & BLK_MQ_UNIQUE_TAG_MASK;
	218	}
	219
320ae51f	220
973c0191	221	int blk_mq_request_started(struct request *rq);
e2490073	222	void blk_mq_start_request(struct request *rq);
c8a446ad CH	223	void blk_mq_end_request(struct request *rq, int error);
c8a446ad CH	224	void __blk_mq_end_request(struct request *rq, int error);
320ae51f	225
ed0791b2	226	void blk_mq_requeue_request(struct request *rq);
6fca6a61	227	void blk_mq_add_to_requeue_list(struct request *rq, bool at_head);
c68ed59f	228	void blk_mq_cancel_requeue_work(struct request_queue *q);
6fca6a61	229	void blk_mq_kick_requeue_list(struct request_queue *q);
2849450a	230	void blk_mq_delay_kick_requeue_list(struct request_queue *q, unsigned long msecs);
1885b24d	231	void blk_mq_abort_requeue_list(struct request_queue *q);
f4829a9b	232	void blk_mq_complete_request(struct request *rq, int error);
30a91cb4	233
320ae51f JA	234	void blk_mq_stop_hw_queue(struct blk_mq_hw_ctx *hctx);
320ae51f JA	235	void blk_mq_start_hw_queue(struct blk_mq_hw_ctx *hctx);
280d45f6	236	void blk_mq_stop_hw_queues(struct request_queue *q);
2f268556	237	void blk_mq_start_hw_queues(struct request_queue *q);
1b4a3258	238	void blk_mq_start_stopped_hw_queues(struct request_queue *q, bool async);
b94ec296	239	void blk_mq_run_hw_queues(struct request_queue *q, bool async);
70f4db63	240	void blk_mq_delay_queue(struct blk_mq_hw_ctx *hctx, unsigned long msecs);
e0489487 SG	241	void blk_mq_tagset_busy_iter(struct blk_mq_tag_set *tagset,
e0489487 SG	242	busy_tag_iter_fn fn, void priv);
c761d96b	243	void blk_mq_freeze_queue(struct request_queue *q);
b4c6a028 KB	244	void blk_mq_unfreeze_queue(struct request_queue *q);
b4c6a028 KB	245	void blk_mq_freeze_queue_start(struct request_queue *q);
486cf989	246	int blk_mq_reinit_tagset(struct blk_mq_tag_set *set);
320ae51f	247
868f2f0b KB	248	void blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, int nr_hw_queues);
868f2f0b KB	249
320ae51f JA	250	/*
320ae51f JA	251	* Driver command data is immediately after the request. So subtract request
2963e3f7	252	* size to get back to the original request, add request size to get the PDU.
320ae51f JA	253	*/
	254	static inline struct request blk_mq_rq_from_pdu(void pdu)
	255	{
	256	return pdu - sizeof(struct request);
	257	}
	258	static inline void blk_mq_rq_to_pdu(struct request rq)
	259	{
2963e3f7	260	return rq + 1;
320ae51f JA	261	}
320ae51f JA	262
320ae51f	263	#define queue_for_each_hw_ctx(q, hctx, i) \
0d0b7d42 JA	264	for ((i) = 0; (i) < (q)->nr_hw_queues && \
0d0b7d42 JA	265	({ hctx = (q)->queue_hw_ctx[i]; 1; }); (i)++)
320ae51f	266
320ae51f	267	#define hctx_for_each_ctx(hctx, ctx, i) \
0d0b7d42 JA	268	for ((i) = 0; (i) < (hctx)->nr_ctx && \
0d0b7d42 JA	269	({ ctx = (hctx)->ctxs[(i)]; 1; }); (i)++)
320ae51f	270
320ae51f	271	#endif