[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 map_size;
	unsigned int bits_per_word;
	struct blk_align_bitmap *map;
};

struct blk_mq_hw_ctx {
	struct {
		spinlock_t		lock;
		struct list_head	dispatch;
	} ____cacheline_aligned_in_smp;

	unsigned long		state;		/* BLK_MQ_S_* flags */
	struct delayed_work	run_work;
	struct delayed_work	delay_work;
	cpumask_var_t		cpumask;
	int			next_cpu;
	int			next_cpu_batch;

	unsigned long		flags;		/* BLK_MQ_F_* flags */

	struct request_queue	*queue;
	unsigned int		queue_num;
	struct blk_flush_queue	*fq;

	void			*driver_data;

	struct blk_mq_ctxmap	ctx_map;

	unsigned int		nr_ctx;
	struct blk_mq_ctx	**ctxs;

	atomic_t		wait_index;

	struct blk_mq_tags	*tags;

	unsigned long		queued;
	unsigned long		run;
#define BLK_MQ_MAX_DISPATCH_ORDER	10
	unsigned long		dispatched[BLK_MQ_MAX_DISPATCH_ORDER];

	unsigned int		numa_node;
	unsigned int		cmd_size;	/* per-request extra data */

	atomic_t		nr_active;

	struct blk_mq_cpu_notifier	cpu_notifier;
	struct kobject		kobj;
};

struct blk_mq_tag_set {
	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 struct blk_mq_hw_ctx *(map_queue_fn)(struct request_queue *, const int);
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 void (busy_iter_fn)(struct blk_mq_hw_ctx *, struct request *, void *,
		bool);

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

	/*
	 * Map to specific hardware queue
	 */
	map_queue_fn		*map_queue;

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

	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;
};

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_SYSFS_UP	= 1 << 3,

	BLK_MQ_S_STOPPED	= 0,
	BLK_MQ_S_TAG_ACTIVE	= 1,

	BLK_MQ_MAX_DEPTH	= 10240,

	BLK_MQ_CPU_WORK_BATCH	= 8,
};

struct request_queue *blk_mq_init_queue(struct blk_mq_tag_set *);
void blk_mq_finish_init(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_run_queues(struct request_queue *q, bool async);
void blk_mq_free_request(struct request *rq);
bool blk_mq_can_queue(struct blk_mq_hw_ctx *);
struct request *blk_mq_alloc_request(struct request_queue *q, int rw,
		gfp_t gfp, bool reserved);
struct request *blk_mq_tag_to_rq(struct blk_mq_tags *tags, unsigned int tag);

struct blk_mq_hw_ctx *blk_mq_map_queue(struct request_queue *, const int ctx_index);
struct blk_mq_hw_ctx *blk_mq_alloc_single_hw_queue(struct blk_mq_tag_set *, unsigned int, int);

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_kick_requeue_list(struct request_queue *q);
void blk_mq_complete_request(struct request *rq);

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_delay_queue(struct blk_mq_hw_ctx *hctx, unsigned long msecs);
void blk_mq_tag_busy_iter(struct blk_mq_hw_ctx *hctx, busy_iter_fn *fn,
		void *priv);

/*
 * Driver command data is immediately after the request. So subtract request
 * size to get back to the original request.
 */
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 (void *) rq + sizeof(*rq);
}

#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 queue_for_each_ctx(q, ctx, i)					\
	for ((i) = 0; (i) < (q)->nr_queues &&				\
	     ({ ctx = per_cpu_ptr((q)->queue_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)++)

#define blk_ctx_sum(q, sum)						\
({									\
	struct blk_mq_ctx *__x;						\
	unsigned int __ret = 0, __i;					\
									\
	queue_for_each_ctx((q), __x, __i)				\
		__ret += sum;						\
	__ret;								\
})

#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 JA	15	struct blk_mq_ctxmap {
	16	unsigned int map_size;
	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;
	25	} ____cacheline_aligned_in_smp;
	26
	27	unsigned long state; /* BLK_MQ_S_* flags */
70f4db63 CH	28	struct delayed_work run_work;
70f4db63 CH	29	struct delayed_work delay_work;
e4043dcf	30	cpumask_var_t cpumask;
506e931f JA	31	int next_cpu;
506e931f JA	32	int next_cpu_batch;
320ae51f JA	33
	34	unsigned long flags; /* BLK_MQ_F_* flags */
	35
	36	struct request_queue *queue;
	37	unsigned int queue_num;
f70ced09	38	struct blk_flush_queue *fq;
320ae51f JA	39
	40	void *driver_data;
	41
1429d7c9 JA	42	struct blk_mq_ctxmap ctx_map;
1429d7c9 JA	43
4bb659b1 JA	44	unsigned int nr_ctx;
	45	struct blk_mq_ctx **ctxs;
	46
8537b120	47	atomic_t wait_index;
320ae51f	48
320ae51f JA	49	struct blk_mq_tags *tags;
	50
	51	unsigned long queued;
	52	unsigned long run;
	53	#define BLK_MQ_MAX_DISPATCH_ORDER 10
	54	unsigned long dispatched[BLK_MQ_MAX_DISPATCH_ORDER];
	55
320ae51f JA	56	unsigned int numa_node;
	57	unsigned int cmd_size; /* per-request extra data */
	58
0d2602ca JA	59	atomic_t nr_active;
0d2602ca JA	60
320ae51f JA	61	struct blk_mq_cpu_notifier cpu_notifier;
	62	struct kobject kobj;
	63	};
	64
24d2f903	65	struct blk_mq_tag_set {
320ae51f JA	66	struct blk_mq_ops *ops;
320ae51f JA	67	unsigned int nr_hw_queues;
e3a2b3f9	68	unsigned int queue_depth; /* max hw supported */
320ae51f JA	69	unsigned int reserved_tags;
	70	unsigned int cmd_size; /* per-request extra data */
	71	int numa_node;
	72	unsigned int timeout;
	73	unsigned int flags; /* BLK_MQ_F_* */
24d2f903 CH	74	void *driver_data;
	75
	76	struct blk_mq_tags **tags;
0d2602ca JA	77
	78	struct mutex tag_list_lock;
	79	struct list_head tag_list;
320ae51f JA	80	};
320ae51f JA	81
74c45052 JA	82	struct blk_mq_queue_data {
	83	struct request *rq;
	84	struct list_head *list;
	85	bool last;
	86	};
	87
	88	typedef int (queue_rq_fn)(struct blk_mq_hw_ctx , const struct blk_mq_queue_data );
320ae51f	89	typedef struct blk_mq_hw_ctx (map_queue_fn)(struct request_queue , const int);
0152fb6b	90	typedef enum blk_eh_timer_return (timeout_fn)(struct request *, bool);
320ae51f JA	91	typedef int (init_hctx_fn)(struct blk_mq_hw_ctx , void , unsigned int);
320ae51f JA	92	typedef void (exit_hctx_fn)(struct blk_mq_hw_ctx *, unsigned int);
24d2f903 CH	93	typedef int (init_request_fn)(void , struct request , unsigned int,
	94	unsigned int, unsigned int);
	95	typedef void (exit_request_fn)(void , struct request , unsigned int,
	96	unsigned int);
320ae51f	97
81481eb4 CH	98	typedef void (busy_iter_fn)(struct blk_mq_hw_ctx , struct request , void *,
	99	bool);
	100
320ae51f JA	101	struct blk_mq_ops {
	102	/*
	103	* Queue request
	104	*/
	105	queue_rq_fn *queue_rq;
	106
	107	/*
	108	* Map to specific hardware queue
	109	*/
	110	map_queue_fn *map_queue;
	111
	112	/*
	113	* Called on request timeout
	114	*/
0152fb6b	115	timeout_fn *timeout;
320ae51f	116
30a91cb4 CH	117	softirq_done_fn *complete;
30a91cb4 CH	118
320ae51f JA	119	/*
	120	* Called when the block layer side of a hardware queue has been
	121	* set up, allowing the driver to allocate/init matching structures.
	122	* Ditto for exit/teardown.
	123	*/
	124	init_hctx_fn *init_hctx;
	125	exit_hctx_fn *exit_hctx;
e9b267d9 CH	126
	127	/*
	128	* Called for every command allocated by the block layer to allow
	129	* the driver to set up driver specific data.
f70ced09 ML	130	*
	131	* Tag greater than or equal to queue_depth is for setting up
	132	* flush request.
	133	*
e9b267d9 CH	134	* Ditto for exit/teardown.
	135	*/
	136	init_request_fn *init_request;
	137	exit_request_fn *exit_request;
320ae51f JA	138	};
	139
	140	enum {
	141	BLK_MQ_RQ_QUEUE_OK = 0, /* queued fine */
	142	BLK_MQ_RQ_QUEUE_BUSY = 1, /* requeue IO for later */
	143	BLK_MQ_RQ_QUEUE_ERROR = 2, /* end IO with error */
	144
	145	BLK_MQ_F_SHOULD_MERGE = 1 << 0,
8a58d1f1 JA	146	BLK_MQ_F_TAG_SHARED = 1 << 1,
	147	BLK_MQ_F_SG_MERGE = 1 << 2,
	148	BLK_MQ_F_SYSFS_UP = 1 << 3,
320ae51f	149
5d12f905	150	BLK_MQ_S_STOPPED = 0,
0d2602ca	151	BLK_MQ_S_TAG_ACTIVE = 1,
320ae51f	152
a4391c64	153	BLK_MQ_MAX_DEPTH = 10240,
506e931f JA	154
506e931f JA	155	BLK_MQ_CPU_WORK_BATCH = 8,
320ae51f JA	156	};
320ae51f JA	157
24d2f903	158	struct request_queue blk_mq_init_queue(struct blk_mq_tag_set );
17497acb	159	void blk_mq_finish_init(struct request_queue *q);
320ae51f JA	160	int blk_mq_register_disk(struct gendisk *);
320ae51f JA	161	void blk_mq_unregister_disk(struct gendisk *);
320ae51f	162
24d2f903 CH	163	int blk_mq_alloc_tag_set(struct blk_mq_tag_set *set);
	164	void blk_mq_free_tag_set(struct blk_mq_tag_set *set);
	165
320ae51f JA	166	void blk_mq_flush_plug_list(struct blk_plug *plug, bool from_schedule);
320ae51f JA	167
feb71dae	168	void blk_mq_insert_request(struct request *, bool, bool, bool);
320ae51f JA	169	void blk_mq_run_queues(struct request_queue *q, bool async);
	170	void blk_mq_free_request(struct request *rq);
	171	bool blk_mq_can_queue(struct blk_mq_hw_ctx *);
4ce01dd1 CH	172	struct request blk_mq_alloc_request(struct request_queue q, int rw,
4ce01dd1 CH	173	gfp_t gfp, bool reserved);
0e62f51f	174	struct request blk_mq_tag_to_rq(struct blk_mq_tags tags, unsigned int tag);
320ae51f JA	175
320ae51f JA	176	struct blk_mq_hw_ctx blk_mq_map_queue(struct request_queue , const int ctx_index);
f14bbe77	177	struct blk_mq_hw_ctx blk_mq_alloc_single_hw_queue(struct blk_mq_tag_set , unsigned int, int);
320ae51f	178
e2490073	179	void blk_mq_start_request(struct request *rq);
c8a446ad CH	180	void blk_mq_end_request(struct request *rq, int error);
c8a446ad CH	181	void __blk_mq_end_request(struct request *rq, int error);
320ae51f	182
ed0791b2	183	void blk_mq_requeue_request(struct request *rq);
6fca6a61 CH	184	void blk_mq_add_to_requeue_list(struct request *rq, bool at_head);
6fca6a61 CH	185	void blk_mq_kick_requeue_list(struct request_queue *q);
30a91cb4 CH	186	void blk_mq_complete_request(struct request *rq);
30a91cb4 CH	187
320ae51f JA	188	void blk_mq_stop_hw_queue(struct blk_mq_hw_ctx *hctx);
320ae51f JA	189	void blk_mq_start_hw_queue(struct blk_mq_hw_ctx *hctx);
280d45f6	190	void blk_mq_stop_hw_queues(struct request_queue *q);
2f268556	191	void blk_mq_start_hw_queues(struct request_queue *q);
1b4a3258	192	void blk_mq_start_stopped_hw_queues(struct request_queue *q, bool async);
70f4db63	193	void blk_mq_delay_queue(struct blk_mq_hw_ctx *hctx, unsigned long msecs);
81481eb4 CH	194	void blk_mq_tag_busy_iter(struct blk_mq_hw_ctx hctx, busy_iter_fn fn,
81481eb4 CH	195	void *priv);
320ae51f JA	196
	197	/*
	198	* Driver command data is immediately after the request. So subtract request
	199	* size to get back to the original request.
	200	*/
	201	static inline struct request blk_mq_rq_from_pdu(void pdu)
	202	{
	203	return pdu - sizeof(struct request);
	204	}
	205	static inline void blk_mq_rq_to_pdu(struct request rq)
	206	{
	207	return (void ) rq + sizeof(rq);
	208	}
	209
320ae51f	210	#define queue_for_each_hw_ctx(q, hctx, i) \
0d0b7d42 JA	211	for ((i) = 0; (i) < (q)->nr_hw_queues && \
0d0b7d42 JA	212	({ hctx = (q)->queue_hw_ctx[i]; 1; }); (i)++)
320ae51f JA	213
320ae51f JA	214	#define queue_for_each_ctx(q, ctx, i) \
0d0b7d42 JA	215	for ((i) = 0; (i) < (q)->nr_queues && \
0d0b7d42 JA	216	({ ctx = per_cpu_ptr((q)->queue_ctx, (i)); 1; }); (i)++)
320ae51f JA	217
320ae51f JA	218	#define hctx_for_each_ctx(hctx, ctx, i) \
0d0b7d42 JA	219	for ((i) = 0; (i) < (hctx)->nr_ctx && \
0d0b7d42 JA	220	({ ctx = (hctx)->ctxs[(i)]; 1; }); (i)++)
320ae51f JA	221
	222	#define blk_ctx_sum(q, sum) \
	223	({ \
	224	struct blk_mq_ctx *__x; \
	225	unsigned int __ret = 0, __i; \
	226	\
	227	queue_for_each_ctx((q), __x, __i) \
	228	__ret += sum; \
	229	__ret; \
	230	})
	231
	232	#endif