[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_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;

	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;
};

typedef int (queue_rq_fn)(struct blk_mq_hw_ctx *, struct request *, bool);
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.
	 * 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 *);
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;
	7
	8	struct blk_mq_cpu_notifier {
	9	struct list_head list;
	10	void *data;
e814e71b	11	int (notify)(void data, unsigned long action, unsigned int cpu);
320ae51f JA	12	};
320ae51f JA	13
1429d7c9 JA	14	struct blk_mq_ctxmap {
	15	unsigned int map_size;
	16	unsigned int bits_per_word;
	17	struct blk_align_bitmap *map;
	18	};
	19
320ae51f JA	20	struct blk_mq_hw_ctx {
	21	struct {
	22	spinlock_t lock;
	23	struct list_head dispatch;
	24	} ____cacheline_aligned_in_smp;
	25
	26	unsigned long state; /* BLK_MQ_S_* flags */
70f4db63 CH	27	struct delayed_work run_work;
70f4db63 CH	28	struct delayed_work delay_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;
	36	unsigned int queue_num;
	37
	38	void *driver_data;
	39
1429d7c9 JA	40	struct blk_mq_ctxmap ctx_map;
1429d7c9 JA	41
4bb659b1 JA	42	unsigned int nr_ctx;
	43	struct blk_mq_ctx **ctxs;
	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;
	51	#define BLK_MQ_MAX_DISPATCH_ORDER 10
	52	unsigned long dispatched[BLK_MQ_MAX_DISPATCH_ORDER];
	53
320ae51f JA	54	unsigned int numa_node;
	55	unsigned int cmd_size; /* per-request extra data */
	56
0d2602ca JA	57	atomic_t nr_active;
0d2602ca JA	58
320ae51f JA	59	struct blk_mq_cpu_notifier cpu_notifier;
	60	struct kobject kobj;
	61	};
	62
24d2f903	63	struct blk_mq_tag_set {
320ae51f JA	64	struct blk_mq_ops *ops;
320ae51f JA	65	unsigned int nr_hw_queues;
e3a2b3f9	66	unsigned int queue_depth; /* max hw supported */
320ae51f JA	67	unsigned int reserved_tags;
	68	unsigned int cmd_size; /* per-request extra data */
	69	int numa_node;
	70	unsigned int timeout;
	71	unsigned int flags; /* BLK_MQ_F_* */
24d2f903 CH	72	void *driver_data;
	73
	74	struct blk_mq_tags **tags;
0d2602ca JA	75
	76	struct mutex tag_list_lock;
	77	struct list_head tag_list;
320ae51f JA	78	};
320ae51f JA	79
bf572297	80	typedef int (queue_rq_fn)(struct blk_mq_hw_ctx , struct request , bool);
320ae51f	81	typedef struct blk_mq_hw_ctx (map_queue_fn)(struct request_queue , const int);
0152fb6b	82	typedef enum blk_eh_timer_return (timeout_fn)(struct request *, bool);
320ae51f JA	83	typedef int (init_hctx_fn)(struct blk_mq_hw_ctx , void , unsigned int);
320ae51f JA	84	typedef void (exit_hctx_fn)(struct blk_mq_hw_ctx *, unsigned int);
24d2f903 CH	85	typedef int (init_request_fn)(void , struct request , unsigned int,
	86	unsigned int, unsigned int);
	87	typedef void (exit_request_fn)(void , struct request , unsigned int,
	88	unsigned int);
320ae51f	89
81481eb4 CH	90	typedef void (busy_iter_fn)(struct blk_mq_hw_ctx , struct request , void *,
	91	bool);
	92
320ae51f JA	93	struct blk_mq_ops {
	94	/*
	95	* Queue request
	96	*/
	97	queue_rq_fn *queue_rq;
	98
	99	/*
	100	* Map to specific hardware queue
	101	*/
	102	map_queue_fn *map_queue;
	103
	104	/*
	105	* Called on request timeout
	106	*/
0152fb6b	107	timeout_fn *timeout;
320ae51f	108
30a91cb4 CH	109	softirq_done_fn *complete;
30a91cb4 CH	110
320ae51f JA	111	/*
	112	* Called when the block layer side of a hardware queue has been
	113	* set up, allowing the driver to allocate/init matching structures.
	114	* Ditto for exit/teardown.
	115	*/
	116	init_hctx_fn *init_hctx;
	117	exit_hctx_fn *exit_hctx;
e9b267d9 CH	118
	119	/*
	120	* Called for every command allocated by the block layer to allow
	121	* the driver to set up driver specific data.
	122	* Ditto for exit/teardown.
	123	*/
	124	init_request_fn *init_request;
	125	exit_request_fn *exit_request;
320ae51f JA	126	};
	127
	128	enum {
	129	BLK_MQ_RQ_QUEUE_OK = 0, /* queued fine */
	130	BLK_MQ_RQ_QUEUE_BUSY = 1, /* requeue IO for later */
	131	BLK_MQ_RQ_QUEUE_ERROR = 2, /* end IO with error */
	132
	133	BLK_MQ_F_SHOULD_MERGE = 1 << 0,
8a58d1f1 JA	134	BLK_MQ_F_TAG_SHARED = 1 << 1,
	135	BLK_MQ_F_SG_MERGE = 1 << 2,
	136	BLK_MQ_F_SYSFS_UP = 1 << 3,
320ae51f	137
5d12f905	138	BLK_MQ_S_STOPPED = 0,
0d2602ca	139	BLK_MQ_S_TAG_ACTIVE = 1,
320ae51f	140
a4391c64	141	BLK_MQ_MAX_DEPTH = 10240,
506e931f JA	142
506e931f JA	143	BLK_MQ_CPU_WORK_BATCH = 8,
320ae51f JA	144	};
320ae51f JA	145
24d2f903	146	struct request_queue blk_mq_init_queue(struct blk_mq_tag_set );
320ae51f JA	147	int blk_mq_register_disk(struct gendisk *);
320ae51f JA	148	void blk_mq_unregister_disk(struct gendisk *);
320ae51f	149
24d2f903 CH	150	int blk_mq_alloc_tag_set(struct blk_mq_tag_set *set);
	151	void blk_mq_free_tag_set(struct blk_mq_tag_set *set);
	152
320ae51f JA	153	void blk_mq_flush_plug_list(struct blk_plug *plug, bool from_schedule);
320ae51f JA	154
feb71dae	155	void blk_mq_insert_request(struct request *, bool, bool, bool);
320ae51f JA	156	void blk_mq_run_queues(struct request_queue *q, bool async);
	157	void blk_mq_free_request(struct request *rq);
	158	bool blk_mq_can_queue(struct blk_mq_hw_ctx *);
4ce01dd1 CH	159	struct request blk_mq_alloc_request(struct request_queue q, int rw,
4ce01dd1 CH	160	gfp_t gfp, bool reserved);
0e62f51f	161	struct request blk_mq_tag_to_rq(struct blk_mq_tags tags, unsigned int tag);
320ae51f JA	162
320ae51f JA	163	struct blk_mq_hw_ctx blk_mq_map_queue(struct request_queue , const int ctx_index);
f14bbe77	164	struct blk_mq_hw_ctx blk_mq_alloc_single_hw_queue(struct blk_mq_tag_set , unsigned int, int);
320ae51f	165
e2490073	166	void blk_mq_start_request(struct request *rq);
c8a446ad CH	167	void blk_mq_end_request(struct request *rq, int error);
c8a446ad CH	168	void __blk_mq_end_request(struct request *rq, int error);
320ae51f	169
ed0791b2	170	void blk_mq_requeue_request(struct request *rq);
6fca6a61 CH	171	void blk_mq_add_to_requeue_list(struct request *rq, bool at_head);
6fca6a61 CH	172	void blk_mq_kick_requeue_list(struct request_queue *q);
30a91cb4 CH	173	void blk_mq_complete_request(struct request *rq);
30a91cb4 CH	174
320ae51f JA	175	void blk_mq_stop_hw_queue(struct blk_mq_hw_ctx *hctx);
320ae51f JA	176	void blk_mq_start_hw_queue(struct blk_mq_hw_ctx *hctx);
280d45f6	177	void blk_mq_stop_hw_queues(struct request_queue *q);
2f268556	178	void blk_mq_start_hw_queues(struct request_queue *q);
1b4a3258	179	void blk_mq_start_stopped_hw_queues(struct request_queue *q, bool async);
70f4db63	180	void blk_mq_delay_queue(struct blk_mq_hw_ctx *hctx, unsigned long msecs);
81481eb4 CH	181	void blk_mq_tag_busy_iter(struct blk_mq_hw_ctx hctx, busy_iter_fn fn,
81481eb4 CH	182	void *priv);
320ae51f JA	183
	184	/*
	185	* Driver command data is immediately after the request. So subtract request
	186	* size to get back to the original request.
	187	*/
	188	static inline struct request blk_mq_rq_from_pdu(void pdu)
	189	{
	190	return pdu - sizeof(struct request);
	191	}
	192	static inline void blk_mq_rq_to_pdu(struct request rq)
	193	{
	194	return (void ) rq + sizeof(rq);
	195	}
	196
320ae51f	197	#define queue_for_each_hw_ctx(q, hctx, i) \
0d0b7d42 JA	198	for ((i) = 0; (i) < (q)->nr_hw_queues && \
0d0b7d42 JA	199	({ hctx = (q)->queue_hw_ctx[i]; 1; }); (i)++)
320ae51f JA	200
320ae51f JA	201	#define queue_for_each_ctx(q, ctx, i) \
0d0b7d42 JA	202	for ((i) = 0; (i) < (q)->nr_queues && \
0d0b7d42 JA	203	({ ctx = per_cpu_ptr((q)->queue_ctx, (i)); 1; }); (i)++)
320ae51f JA	204
320ae51f JA	205	#define hctx_for_each_ctx(hctx, ctx, i) \
0d0b7d42 JA	206	for ((i) = 0; (i) < (hctx)->nr_ctx && \
0d0b7d42 JA	207	({ ctx = (hctx)->ctxs[(i)]; 1; }); (i)++)
320ae51f JA	208
	209	#define blk_ctx_sum(q, sum) \
	210	({ \
	211	struct blk_mq_ctx *__x; \
	212	unsigned int __ret = 0, __i; \
	213	\
	214	queue_for_each_ctx((q), __x, __i) \
	215	__ret += sum; \
	216	__ret; \
	217	})
	218
	219	#endif