1 #ifndef _LINUX_BLKDEV_H
2 #define _LINUX_BLKDEV_H
4 #include <linux/sched.h>
8 #include <linux/major.h>
9 #include <linux/genhd.h>
10 #include <linux/list.h>
11 #include <linux/timer.h>
12 #include <linux/workqueue.h>
13 #include <linux/pagemap.h>
14 #include <linux/backing-dev.h>
15 #include <linux/wait.h>
16 #include <linux/mempool.h>
17 #include <linux/bio.h>
18 #include <linux/stringify.h>
19 #include <linux/gfp.h>
20 #include <linux/bsg.h>
21 #include <linux/smp.h>
23 #include <asm/scatterlist.h>
26 struct scsi_ioctl_command
;
29 struct elevator_queue
;
30 struct request_pm_state
;
37 #define BLKDEV_MIN_RQ 4
38 #define BLKDEV_MAX_RQ 128 /* Default maximum */
41 * Maximum number of blkcg policies allowed to be registered concurrently.
42 * Defined here to simplify include dependency.
44 #define BLKCG_MAX_POLS 2
47 typedef void (rq_end_io_fn
)(struct request
*, int);
49 #define BLK_RL_SYNCFULL (1U << 0)
50 #define BLK_RL_ASYNCFULL (1U << 1)
53 struct request_queue
*q
; /* the queue this rl belongs to */
54 #ifdef CONFIG_BLK_CGROUP
55 struct blkcg_gq
*blkg
; /* blkg this request pool belongs to */
58 * count[], starved[], and wait[] are indexed by
59 * BLK_RW_SYNC/BLK_RW_ASYNC
64 wait_queue_head_t wait
[2];
69 * request command types
71 enum rq_cmd_type_bits
{
72 REQ_TYPE_FS
= 1, /* fs request */
73 REQ_TYPE_BLOCK_PC
, /* scsi command */
74 REQ_TYPE_SENSE
, /* sense request */
75 REQ_TYPE_PM_SUSPEND
, /* suspend request */
76 REQ_TYPE_PM_RESUME
, /* resume request */
77 REQ_TYPE_PM_SHUTDOWN
, /* shutdown request */
78 REQ_TYPE_SPECIAL
, /* driver defined type */
80 * for ATA/ATAPI devices. this really doesn't belong here, ide should
81 * use REQ_TYPE_SPECIAL and use rq->cmd[0] with the range of driver
82 * private REQ_LB opcodes to differentiate what type of request this is
84 REQ_TYPE_ATA_TASKFILE
,
88 #define BLK_MAX_CDB 16
91 * try to put the fields that are referenced together in the same cacheline.
92 * if you modify this structure, be sure to check block/blk-core.c:blk_rq_init()
96 struct list_head queuelist
;
97 struct call_single_data csd
;
99 struct request_queue
*q
;
101 unsigned int cmd_flags
;
102 enum rq_cmd_type_bits cmd_type
;
103 unsigned long atomic_flags
;
107 /* the following two fields are internal, NEVER access directly */
108 unsigned int __data_len
; /* total data len */
109 sector_t __sector
; /* sector cursor */
114 struct hlist_node hash
; /* merge hash */
116 * The rb_node is only used inside the io scheduler, requests
117 * are pruned when moved to the dispatch queue. So let the
118 * completion_data share space with the rb_node.
121 struct rb_node rb_node
; /* sort/lookup */
122 void *completion_data
;
126 * Three pointers are available for the IO schedulers, if they need
127 * more they have to dynamically allocate it. Flush requests are
128 * never put on the IO scheduler. So let the flush fields share
129 * space with the elevator data.
139 struct list_head list
;
140 rq_end_io_fn
*saved_end_io
;
144 struct gendisk
*rq_disk
;
145 struct hd_struct
*part
;
146 unsigned long start_time
;
147 #ifdef CONFIG_BLK_CGROUP
148 struct request_list
*rl
; /* rl this rq is alloced from */
149 unsigned long long start_time_ns
;
150 unsigned long long io_start_time_ns
; /* when passed to hardware */
152 /* Number of scatter-gather DMA addr+len pairs after
153 * physical address coalescing is performed.
155 unsigned short nr_phys_segments
;
156 #if defined(CONFIG_BLK_DEV_INTEGRITY)
157 unsigned short nr_integrity_segments
;
160 unsigned short ioprio
;
164 void *special
; /* opaque pointer available for LLD use */
165 char *buffer
; /* kaddr of the current segment if available */
171 * when request is used as a packet command carrier
173 unsigned char __cmd
[BLK_MAX_CDB
];
175 unsigned short cmd_len
;
177 unsigned int extra_len
; /* length of alignment and padding */
178 unsigned int sense_len
;
179 unsigned int resid_len
; /* residual count */
182 unsigned long deadline
;
183 struct list_head timeout_list
;
184 unsigned int timeout
;
188 * completion callback.
190 rq_end_io_fn
*end_io
;
194 struct request
*next_rq
;
197 static inline unsigned short req_get_ioprio(struct request
*req
)
203 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
204 * requests. Some step values could eventually be made generic.
206 struct request_pm_state
208 /* PM state machine step value, currently driver specific */
210 /* requested PM state value (S1, S2, S3, S4, ...) */
212 void* data
; /* for driver use */
215 #include <linux/elevator.h>
217 typedef void (request_fn_proc
) (struct request_queue
*q
);
218 typedef void (make_request_fn
) (struct request_queue
*q
, struct bio
*bio
);
219 typedef int (prep_rq_fn
) (struct request_queue
*, struct request
*);
220 typedef void (unprep_rq_fn
) (struct request_queue
*, struct request
*);
223 struct bvec_merge_data
{
224 struct block_device
*bi_bdev
;
229 typedef int (merge_bvec_fn
) (struct request_queue
*, struct bvec_merge_data
*,
231 typedef void (softirq_done_fn
)(struct request
*);
232 typedef int (dma_drain_needed_fn
)(struct request
*);
233 typedef int (lld_busy_fn
) (struct request_queue
*q
);
234 typedef int (bsg_job_fn
) (struct bsg_job
*);
236 enum blk_eh_timer_return
{
242 typedef enum blk_eh_timer_return (rq_timed_out_fn
)(struct request
*);
244 enum blk_queue_state
{
249 struct blk_queue_tag
{
250 struct request
**tag_index
; /* map of busy tags */
251 unsigned long *tag_map
; /* bit map of free/busy tags */
252 int busy
; /* current depth */
253 int max_depth
; /* what we will send to device */
254 int real_max_depth
; /* what the array can hold */
255 atomic_t refcnt
; /* map can be shared */
258 #define BLK_SCSI_MAX_CMDS (256)
259 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
261 struct queue_limits
{
262 unsigned long bounce_pfn
;
263 unsigned long seg_boundary_mask
;
265 unsigned int max_hw_sectors
;
266 unsigned int max_sectors
;
267 unsigned int max_segment_size
;
268 unsigned int physical_block_size
;
269 unsigned int alignment_offset
;
272 unsigned int max_discard_sectors
;
273 unsigned int discard_granularity
;
274 unsigned int discard_alignment
;
276 unsigned short logical_block_size
;
277 unsigned short max_segments
;
278 unsigned short max_integrity_segments
;
280 unsigned char misaligned
;
281 unsigned char discard_misaligned
;
282 unsigned char cluster
;
283 unsigned char discard_zeroes_data
;
286 struct request_queue
{
288 * Together with queue_head for cacheline sharing
290 struct list_head queue_head
;
291 struct request
*last_merge
;
292 struct elevator_queue
*elevator
;
293 int nr_rqs
[2]; /* # allocated [a]sync rqs */
294 int nr_rqs_elvpriv
; /* # allocated rqs w/ elvpriv */
297 * If blkcg is not used, @q->root_rl serves all requests. If blkcg
298 * is used, root blkg allocates from @q->root_rl and all other
299 * blkgs from their own blkg->rl. Which one to use should be
300 * determined using bio_request_list().
302 struct request_list root_rl
;
304 request_fn_proc
*request_fn
;
305 make_request_fn
*make_request_fn
;
306 prep_rq_fn
*prep_rq_fn
;
307 unprep_rq_fn
*unprep_rq_fn
;
308 merge_bvec_fn
*merge_bvec_fn
;
309 softirq_done_fn
*softirq_done_fn
;
310 rq_timed_out_fn
*rq_timed_out_fn
;
311 dma_drain_needed_fn
*dma_drain_needed
;
312 lld_busy_fn
*lld_busy_fn
;
315 * Dispatch queue sorting
318 struct request
*boundary_rq
;
321 * Delayed queue handling
323 struct delayed_work delay_work
;
325 struct backing_dev_info backing_dev_info
;
328 * The queue owner gets to use this for whatever they like.
329 * ll_rw_blk doesn't touch it.
334 * various queue flags, see QUEUE_* below
336 unsigned long queue_flags
;
339 * ida allocated id for this queue. Used to index queues from
345 * queue needs bounce pages for pages above this limit
350 * protects queue structures from reentrancy. ->__queue_lock should
351 * _never_ be used directly, it is queue private. always use
354 spinlock_t __queue_lock
;
355 spinlock_t
*queue_lock
;
365 unsigned long nr_requests
; /* Max # of requests */
366 unsigned int nr_congestion_on
;
367 unsigned int nr_congestion_off
;
368 unsigned int nr_batching
;
370 unsigned int dma_drain_size
;
371 void *dma_drain_buffer
;
372 unsigned int dma_pad_mask
;
373 unsigned int dma_alignment
;
375 struct blk_queue_tag
*queue_tags
;
376 struct list_head tag_busy_list
;
378 unsigned int nr_sorted
;
379 unsigned int in_flight
[2];
381 unsigned int rq_timeout
;
382 struct timer_list timeout
;
383 struct list_head timeout_list
;
385 struct list_head icq_list
;
386 #ifdef CONFIG_BLK_CGROUP
387 DECLARE_BITMAP (blkcg_pols
, BLKCG_MAX_POLS
);
388 struct blkcg_gq
*root_blkg
;
389 struct list_head blkg_list
;
392 struct queue_limits limits
;
397 unsigned int sg_timeout
;
398 unsigned int sg_reserved_size
;
400 #ifdef CONFIG_BLK_DEV_IO_TRACE
401 struct blk_trace
*blk_trace
;
404 * for flush operations
406 unsigned int flush_flags
;
407 unsigned int flush_not_queueable
:1;
408 unsigned int flush_queue_delayed
:1;
409 unsigned int flush_pending_idx
:1;
410 unsigned int flush_running_idx
:1;
411 unsigned long flush_pending_since
;
412 struct list_head flush_queue
[2];
413 struct list_head flush_data_in_flight
;
414 struct request flush_rq
;
416 struct mutex sysfs_lock
;
420 #if defined(CONFIG_BLK_DEV_BSG)
421 bsg_job_fn
*bsg_job_fn
;
423 struct bsg_class_device bsg_dev
;
426 #ifdef CONFIG_BLK_CGROUP
427 struct list_head all_q_node
;
429 #ifdef CONFIG_BLK_DEV_THROTTLING
431 struct throtl_data
*td
;
435 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
436 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
437 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
438 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
439 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
440 #define QUEUE_FLAG_BYPASS 6 /* act as dumb FIFO queue */
441 #define QUEUE_FLAG_BIDI 7 /* queue supports bidi requests */
442 #define QUEUE_FLAG_NOMERGES 8 /* disable merge attempts */
443 #define QUEUE_FLAG_SAME_COMP 9 /* complete on same CPU-group */
444 #define QUEUE_FLAG_FAIL_IO 10 /* fake timeout */
445 #define QUEUE_FLAG_STACKABLE 11 /* supports request stacking */
446 #define QUEUE_FLAG_NONROT 12 /* non-rotational device (SSD) */
447 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
448 #define QUEUE_FLAG_IO_STAT 13 /* do IO stats */
449 #define QUEUE_FLAG_DISCARD 14 /* supports DISCARD */
450 #define QUEUE_FLAG_NOXMERGES 15 /* No extended merges */
451 #define QUEUE_FLAG_ADD_RANDOM 16 /* Contributes to random pool */
452 #define QUEUE_FLAG_SECDISCARD 17 /* supports SECDISCARD */
453 #define QUEUE_FLAG_SAME_FORCE 18 /* force complete on same CPU */
455 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
456 (1 << QUEUE_FLAG_STACKABLE) | \
457 (1 << QUEUE_FLAG_SAME_COMP) | \
458 (1 << QUEUE_FLAG_ADD_RANDOM))
460 static inline void queue_lockdep_assert_held(struct request_queue
*q
)
463 lockdep_assert_held(q
->queue_lock
);
466 static inline void queue_flag_set_unlocked(unsigned int flag
,
467 struct request_queue
*q
)
469 __set_bit(flag
, &q
->queue_flags
);
472 static inline int queue_flag_test_and_clear(unsigned int flag
,
473 struct request_queue
*q
)
475 queue_lockdep_assert_held(q
);
477 if (test_bit(flag
, &q
->queue_flags
)) {
478 __clear_bit(flag
, &q
->queue_flags
);
485 static inline int queue_flag_test_and_set(unsigned int flag
,
486 struct request_queue
*q
)
488 queue_lockdep_assert_held(q
);
490 if (!test_bit(flag
, &q
->queue_flags
)) {
491 __set_bit(flag
, &q
->queue_flags
);
498 static inline void queue_flag_set(unsigned int flag
, struct request_queue
*q
)
500 queue_lockdep_assert_held(q
);
501 __set_bit(flag
, &q
->queue_flags
);
504 static inline void queue_flag_clear_unlocked(unsigned int flag
,
505 struct request_queue
*q
)
507 __clear_bit(flag
, &q
->queue_flags
);
510 static inline int queue_in_flight(struct request_queue
*q
)
512 return q
->in_flight
[0] + q
->in_flight
[1];
515 static inline void queue_flag_clear(unsigned int flag
, struct request_queue
*q
)
517 queue_lockdep_assert_held(q
);
518 __clear_bit(flag
, &q
->queue_flags
);
521 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
522 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
523 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
524 #define blk_queue_bypass(q) test_bit(QUEUE_FLAG_BYPASS, &(q)->queue_flags)
525 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
526 #define blk_queue_noxmerges(q) \
527 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
528 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
529 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
530 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
531 #define blk_queue_stackable(q) \
532 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
533 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
534 #define blk_queue_secdiscard(q) (blk_queue_discard(q) && \
535 test_bit(QUEUE_FLAG_SECDISCARD, &(q)->queue_flags))
537 #define blk_noretry_request(rq) \
538 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
539 REQ_FAILFAST_DRIVER))
541 #define blk_account_rq(rq) \
542 (((rq)->cmd_flags & REQ_STARTED) && \
543 ((rq)->cmd_type == REQ_TYPE_FS || \
544 ((rq)->cmd_flags & REQ_DISCARD)))
546 #define blk_pm_request(rq) \
547 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
548 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
550 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
551 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
552 /* rq->queuelist of dequeued request must be list_empty() */
553 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
555 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
557 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
559 static inline unsigned int blk_queue_cluster(struct request_queue
*q
)
561 return q
->limits
.cluster
;
565 * We regard a request as sync, if either a read or a sync write
567 static inline bool rw_is_sync(unsigned int rw_flags
)
569 return !(rw_flags
& REQ_WRITE
) || (rw_flags
& REQ_SYNC
);
572 static inline bool rq_is_sync(struct request
*rq
)
574 return rw_is_sync(rq
->cmd_flags
);
577 static inline bool blk_rl_full(struct request_list
*rl
, bool sync
)
579 unsigned int flag
= sync
? BLK_RL_SYNCFULL
: BLK_RL_ASYNCFULL
;
581 return rl
->flags
& flag
;
584 static inline void blk_set_rl_full(struct request_list
*rl
, bool sync
)
586 unsigned int flag
= sync
? BLK_RL_SYNCFULL
: BLK_RL_ASYNCFULL
;
591 static inline void blk_clear_rl_full(struct request_list
*rl
, bool sync
)
593 unsigned int flag
= sync
? BLK_RL_SYNCFULL
: BLK_RL_ASYNCFULL
;
600 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
601 * it already be started by driver.
603 #define RQ_NOMERGE_FLAGS \
604 (REQ_NOMERGE | REQ_STARTED | REQ_SOFTBARRIER | REQ_FLUSH | REQ_FUA | REQ_DISCARD)
605 #define rq_mergeable(rq) \
606 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
607 (((rq)->cmd_flags & REQ_DISCARD) || \
608 (rq)->cmd_type == REQ_TYPE_FS))
611 * q->prep_rq_fn return values
613 #define BLKPREP_OK 0 /* serve it */
614 #define BLKPREP_KILL 1 /* fatal error, kill */
615 #define BLKPREP_DEFER 2 /* leave on queue */
617 extern unsigned long blk_max_low_pfn
, blk_max_pfn
;
620 * standard bounce addresses:
622 * BLK_BOUNCE_HIGH : bounce all highmem pages
623 * BLK_BOUNCE_ANY : don't bounce anything
624 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
627 #if BITS_PER_LONG == 32
628 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
630 #define BLK_BOUNCE_HIGH -1ULL
632 #define BLK_BOUNCE_ANY (-1ULL)
633 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
636 * default timeout for SG_IO if none specified
638 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
639 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
642 extern int init_emergency_isa_pool(void);
643 extern void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
);
645 static inline int init_emergency_isa_pool(void)
649 static inline void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
)
652 #endif /* CONFIG_MMU */
658 unsigned long offset
;
663 struct req_iterator
{
668 /* This should not be used directly - use rq_for_each_segment */
669 #define for_each_bio(_bio) \
670 for (; _bio; _bio = _bio->bi_next)
671 #define __rq_for_each_bio(_bio, rq) \
673 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
675 #define rq_for_each_segment(bvl, _rq, _iter) \
676 __rq_for_each_bio(_iter.bio, _rq) \
677 bio_for_each_segment(bvl, _iter.bio, _iter.i)
679 #define rq_iter_last(rq, _iter) \
680 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
682 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
683 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
685 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
686 extern void rq_flush_dcache_pages(struct request
*rq
);
688 static inline void rq_flush_dcache_pages(struct request
*rq
)
693 extern int blk_register_queue(struct gendisk
*disk
);
694 extern void blk_unregister_queue(struct gendisk
*disk
);
695 extern void generic_make_request(struct bio
*bio
);
696 extern void blk_rq_init(struct request_queue
*q
, struct request
*rq
);
697 extern void blk_put_request(struct request
*);
698 extern void __blk_put_request(struct request_queue
*, struct request
*);
699 extern struct request
*blk_get_request(struct request_queue
*, int, gfp_t
);
700 extern struct request
*blk_make_request(struct request_queue
*, struct bio
*,
702 extern void blk_requeue_request(struct request_queue
*, struct request
*);
703 extern void blk_add_request_payload(struct request
*rq
, struct page
*page
,
705 extern int blk_rq_check_limits(struct request_queue
*q
, struct request
*rq
);
706 extern int blk_lld_busy(struct request_queue
*q
);
707 extern int blk_rq_prep_clone(struct request
*rq
, struct request
*rq_src
,
708 struct bio_set
*bs
, gfp_t gfp_mask
,
709 int (*bio_ctr
)(struct bio
*, struct bio
*, void *),
711 extern void blk_rq_unprep_clone(struct request
*rq
);
712 extern int blk_insert_cloned_request(struct request_queue
*q
,
714 extern void blk_delay_queue(struct request_queue
*, unsigned long);
715 extern void blk_recount_segments(struct request_queue
*, struct bio
*);
716 extern int scsi_verify_blk_ioctl(struct block_device
*, unsigned int);
717 extern int scsi_cmd_blk_ioctl(struct block_device
*, fmode_t
,
718 unsigned int, void __user
*);
719 extern int scsi_cmd_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
720 unsigned int, void __user
*);
721 extern int sg_scsi_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
722 struct scsi_ioctl_command __user
*);
724 extern void blk_queue_bio(struct request_queue
*q
, struct bio
*bio
);
727 * A queue has just exitted congestion. Note this in the global counter of
728 * congested queues, and wake up anyone who was waiting for requests to be
731 static inline void blk_clear_queue_congested(struct request_queue
*q
, int sync
)
733 clear_bdi_congested(&q
->backing_dev_info
, sync
);
737 * A queue has just entered congestion. Flag that in the queue's VM-visible
738 * state flags and increment the global gounter of congested queues.
740 static inline void blk_set_queue_congested(struct request_queue
*q
, int sync
)
742 set_bdi_congested(&q
->backing_dev_info
, sync
);
745 extern void blk_start_queue(struct request_queue
*q
);
746 extern void blk_stop_queue(struct request_queue
*q
);
747 extern void blk_sync_queue(struct request_queue
*q
);
748 extern void __blk_stop_queue(struct request_queue
*q
);
749 extern void __blk_run_queue(struct request_queue
*q
);
750 extern void blk_run_queue(struct request_queue
*);
751 extern void blk_run_queue_async(struct request_queue
*q
);
752 extern int blk_rq_map_user(struct request_queue
*, struct request
*,
753 struct rq_map_data
*, void __user
*, unsigned long,
755 extern int blk_rq_unmap_user(struct bio
*);
756 extern int blk_rq_map_kern(struct request_queue
*, struct request
*, void *, unsigned int, gfp_t
);
757 extern int blk_rq_map_user_iov(struct request_queue
*, struct request
*,
758 struct rq_map_data
*, struct sg_iovec
*, int,
759 unsigned int, gfp_t
);
760 extern int blk_execute_rq(struct request_queue
*, struct gendisk
*,
761 struct request
*, int);
762 extern void blk_execute_rq_nowait(struct request_queue
*, struct gendisk
*,
763 struct request
*, int, rq_end_io_fn
*);
765 static inline struct request_queue
*bdev_get_queue(struct block_device
*bdev
)
767 return bdev
->bd_disk
->queue
;
771 * blk_rq_pos() : the current sector
772 * blk_rq_bytes() : bytes left in the entire request
773 * blk_rq_cur_bytes() : bytes left in the current segment
774 * blk_rq_err_bytes() : bytes left till the next error boundary
775 * blk_rq_sectors() : sectors left in the entire request
776 * blk_rq_cur_sectors() : sectors left in the current segment
778 static inline sector_t
blk_rq_pos(const struct request
*rq
)
783 static inline unsigned int blk_rq_bytes(const struct request
*rq
)
785 return rq
->__data_len
;
788 static inline int blk_rq_cur_bytes(const struct request
*rq
)
790 return rq
->bio
? bio_cur_bytes(rq
->bio
) : 0;
793 extern unsigned int blk_rq_err_bytes(const struct request
*rq
);
795 static inline unsigned int blk_rq_sectors(const struct request
*rq
)
797 return blk_rq_bytes(rq
) >> 9;
800 static inline unsigned int blk_rq_cur_sectors(const struct request
*rq
)
802 return blk_rq_cur_bytes(rq
) >> 9;
806 * Request issue related functions.
808 extern struct request
*blk_peek_request(struct request_queue
*q
);
809 extern void blk_start_request(struct request
*rq
);
810 extern struct request
*blk_fetch_request(struct request_queue
*q
);
813 * Request completion related functions.
815 * blk_update_request() completes given number of bytes and updates
816 * the request without completing it.
818 * blk_end_request() and friends. __blk_end_request() must be called
819 * with the request queue spinlock acquired.
821 * Several drivers define their own end_request and call
822 * blk_end_request() for parts of the original function.
823 * This prevents code duplication in drivers.
825 extern bool blk_update_request(struct request
*rq
, int error
,
826 unsigned int nr_bytes
);
827 extern bool blk_end_request(struct request
*rq
, int error
,
828 unsigned int nr_bytes
);
829 extern void blk_end_request_all(struct request
*rq
, int error
);
830 extern bool blk_end_request_cur(struct request
*rq
, int error
);
831 extern bool blk_end_request_err(struct request
*rq
, int error
);
832 extern bool __blk_end_request(struct request
*rq
, int error
,
833 unsigned int nr_bytes
);
834 extern void __blk_end_request_all(struct request
*rq
, int error
);
835 extern bool __blk_end_request_cur(struct request
*rq
, int error
);
836 extern bool __blk_end_request_err(struct request
*rq
, int error
);
838 extern void blk_complete_request(struct request
*);
839 extern void __blk_complete_request(struct request
*);
840 extern void blk_abort_request(struct request
*);
841 extern void blk_unprep_request(struct request
*);
844 * Access functions for manipulating queue properties
846 extern struct request_queue
*blk_init_queue_node(request_fn_proc
*rfn
,
847 spinlock_t
*lock
, int node_id
);
848 extern struct request_queue
*blk_init_queue(request_fn_proc
*, spinlock_t
*);
849 extern struct request_queue
*blk_init_allocated_queue(struct request_queue
*,
850 request_fn_proc
*, spinlock_t
*);
851 extern void blk_cleanup_queue(struct request_queue
*);
852 extern void blk_queue_make_request(struct request_queue
*, make_request_fn
*);
853 extern void blk_queue_bounce_limit(struct request_queue
*, u64
);
854 extern void blk_limits_max_hw_sectors(struct queue_limits
*, unsigned int);
855 extern void blk_queue_max_hw_sectors(struct request_queue
*, unsigned int);
856 extern void blk_queue_max_segments(struct request_queue
*, unsigned short);
857 extern void blk_queue_max_segment_size(struct request_queue
*, unsigned int);
858 extern void blk_queue_max_discard_sectors(struct request_queue
*q
,
859 unsigned int max_discard_sectors
);
860 extern void blk_queue_logical_block_size(struct request_queue
*, unsigned short);
861 extern void blk_queue_physical_block_size(struct request_queue
*, unsigned int);
862 extern void blk_queue_alignment_offset(struct request_queue
*q
,
863 unsigned int alignment
);
864 extern void blk_limits_io_min(struct queue_limits
*limits
, unsigned int min
);
865 extern void blk_queue_io_min(struct request_queue
*q
, unsigned int min
);
866 extern void blk_limits_io_opt(struct queue_limits
*limits
, unsigned int opt
);
867 extern void blk_queue_io_opt(struct request_queue
*q
, unsigned int opt
);
868 extern void blk_set_default_limits(struct queue_limits
*lim
);
869 extern void blk_set_stacking_limits(struct queue_limits
*lim
);
870 extern int blk_stack_limits(struct queue_limits
*t
, struct queue_limits
*b
,
872 extern int bdev_stack_limits(struct queue_limits
*t
, struct block_device
*bdev
,
874 extern void disk_stack_limits(struct gendisk
*disk
, struct block_device
*bdev
,
876 extern void blk_queue_stack_limits(struct request_queue
*t
, struct request_queue
*b
);
877 extern void blk_queue_dma_pad(struct request_queue
*, unsigned int);
878 extern void blk_queue_update_dma_pad(struct request_queue
*, unsigned int);
879 extern int blk_queue_dma_drain(struct request_queue
*q
,
880 dma_drain_needed_fn
*dma_drain_needed
,
881 void *buf
, unsigned int size
);
882 extern void blk_queue_lld_busy(struct request_queue
*q
, lld_busy_fn
*fn
);
883 extern void blk_queue_segment_boundary(struct request_queue
*, unsigned long);
884 extern void blk_queue_prep_rq(struct request_queue
*, prep_rq_fn
*pfn
);
885 extern void blk_queue_unprep_rq(struct request_queue
*, unprep_rq_fn
*ufn
);
886 extern void blk_queue_merge_bvec(struct request_queue
*, merge_bvec_fn
*);
887 extern void blk_queue_dma_alignment(struct request_queue
*, int);
888 extern void blk_queue_update_dma_alignment(struct request_queue
*, int);
889 extern void blk_queue_softirq_done(struct request_queue
*, softirq_done_fn
*);
890 extern void blk_queue_rq_timed_out(struct request_queue
*, rq_timed_out_fn
*);
891 extern void blk_queue_rq_timeout(struct request_queue
*, unsigned int);
892 extern void blk_queue_flush(struct request_queue
*q
, unsigned int flush
);
893 extern void blk_queue_flush_queueable(struct request_queue
*q
, bool queueable
);
894 extern struct backing_dev_info
*blk_get_backing_dev_info(struct block_device
*bdev
);
896 extern int blk_rq_map_sg(struct request_queue
*, struct request
*, struct scatterlist
*);
897 extern int blk_bio_map_sg(struct request_queue
*q
, struct bio
*bio
,
898 struct scatterlist
*sglist
);
899 extern void blk_dump_rq_flags(struct request
*, char *);
900 extern long nr_blockdev_pages(void);
902 bool __must_check
blk_get_queue(struct request_queue
*);
903 struct request_queue
*blk_alloc_queue(gfp_t
);
904 struct request_queue
*blk_alloc_queue_node(gfp_t
, int);
905 extern void blk_put_queue(struct request_queue
*);
908 * blk_plug permits building a queue of related requests by holding the I/O
909 * fragments for a short period. This allows merging of sequential requests
910 * into single larger request. As the requests are moved from a per-task list to
911 * the device's request_queue in a batch, this results in improved scalability
912 * as the lock contention for request_queue lock is reduced.
914 * It is ok not to disable preemption when adding the request to the plug list
915 * or when attempting a merge, because blk_schedule_flush_list() will only flush
916 * the plug list when the task sleeps by itself. For details, please see
917 * schedule() where blk_schedule_flush_plug() is called.
920 unsigned long magic
; /* detect uninitialized use-cases */
921 struct list_head list
; /* requests */
922 struct list_head cb_list
; /* md requires an unplug callback */
923 unsigned int should_sort
; /* list to be sorted before flushing? */
925 #define BLK_MAX_REQUEST_COUNT 16
928 typedef void (*blk_plug_cb_fn
)(struct blk_plug_cb
*, bool);
930 struct list_head list
;
931 blk_plug_cb_fn callback
;
934 extern struct blk_plug_cb
*blk_check_plugged(blk_plug_cb_fn unplug
,
935 void *data
, int size
);
936 extern void blk_start_plug(struct blk_plug
*);
937 extern void blk_finish_plug(struct blk_plug
*);
938 extern void blk_flush_plug_list(struct blk_plug
*, bool);
940 static inline void blk_flush_plug(struct task_struct
*tsk
)
942 struct blk_plug
*plug
= tsk
->plug
;
945 blk_flush_plug_list(plug
, false);
948 static inline void blk_schedule_flush_plug(struct task_struct
*tsk
)
950 struct blk_plug
*plug
= tsk
->plug
;
953 blk_flush_plug_list(plug
, true);
956 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
958 struct blk_plug
*plug
= tsk
->plug
;
960 return plug
&& (!list_empty(&plug
->list
) || !list_empty(&plug
->cb_list
));
966 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
967 extern int blk_queue_start_tag(struct request_queue
*, struct request
*);
968 extern struct request
*blk_queue_find_tag(struct request_queue
*, int);
969 extern void blk_queue_end_tag(struct request_queue
*, struct request
*);
970 extern int blk_queue_init_tags(struct request_queue
*, int, struct blk_queue_tag
*);
971 extern void blk_queue_free_tags(struct request_queue
*);
972 extern int blk_queue_resize_tags(struct request_queue
*, int);
973 extern void blk_queue_invalidate_tags(struct request_queue
*);
974 extern struct blk_queue_tag
*blk_init_tags(int);
975 extern void blk_free_tags(struct blk_queue_tag
*);
977 static inline struct request
*blk_map_queue_find_tag(struct blk_queue_tag
*bqt
,
980 if (unlikely(bqt
== NULL
|| tag
>= bqt
->real_max_depth
))
982 return bqt
->tag_index
[tag
];
985 #define BLKDEV_DISCARD_SECURE 0x01 /* secure discard */
987 extern int blkdev_issue_flush(struct block_device
*, gfp_t
, sector_t
*);
988 extern int blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
989 sector_t nr_sects
, gfp_t gfp_mask
, unsigned long flags
);
990 extern int blkdev_issue_zeroout(struct block_device
*bdev
, sector_t sector
,
991 sector_t nr_sects
, gfp_t gfp_mask
);
992 static inline int sb_issue_discard(struct super_block
*sb
, sector_t block
,
993 sector_t nr_blocks
, gfp_t gfp_mask
, unsigned long flags
)
995 return blkdev_issue_discard(sb
->s_bdev
, block
<< (sb
->s_blocksize_bits
- 9),
996 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
999 static inline int sb_issue_zeroout(struct super_block
*sb
, sector_t block
,
1000 sector_t nr_blocks
, gfp_t gfp_mask
)
1002 return blkdev_issue_zeroout(sb
->s_bdev
,
1003 block
<< (sb
->s_blocksize_bits
- 9),
1004 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
1008 extern int blk_verify_command(unsigned char *cmd
, fmode_t has_write_perm
);
1010 enum blk_default_limits
{
1011 BLK_MAX_SEGMENTS
= 128,
1012 BLK_SAFE_MAX_SECTORS
= 255,
1013 BLK_DEF_MAX_SECTORS
= 1024,
1014 BLK_MAX_SEGMENT_SIZE
= 65536,
1015 BLK_SEG_BOUNDARY_MASK
= 0xFFFFFFFFUL
,
1018 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
1020 static inline unsigned long queue_bounce_pfn(struct request_queue
*q
)
1022 return q
->limits
.bounce_pfn
;
1025 static inline unsigned long queue_segment_boundary(struct request_queue
*q
)
1027 return q
->limits
.seg_boundary_mask
;
1030 static inline unsigned int queue_max_sectors(struct request_queue
*q
)
1032 return q
->limits
.max_sectors
;
1035 static inline unsigned int queue_max_hw_sectors(struct request_queue
*q
)
1037 return q
->limits
.max_hw_sectors
;
1040 static inline unsigned short queue_max_segments(struct request_queue
*q
)
1042 return q
->limits
.max_segments
;
1045 static inline unsigned int queue_max_segment_size(struct request_queue
*q
)
1047 return q
->limits
.max_segment_size
;
1050 static inline unsigned short queue_logical_block_size(struct request_queue
*q
)
1054 if (q
&& q
->limits
.logical_block_size
)
1055 retval
= q
->limits
.logical_block_size
;
1060 static inline unsigned short bdev_logical_block_size(struct block_device
*bdev
)
1062 return queue_logical_block_size(bdev_get_queue(bdev
));
1065 static inline unsigned int queue_physical_block_size(struct request_queue
*q
)
1067 return q
->limits
.physical_block_size
;
1070 static inline unsigned int bdev_physical_block_size(struct block_device
*bdev
)
1072 return queue_physical_block_size(bdev_get_queue(bdev
));
1075 static inline unsigned int queue_io_min(struct request_queue
*q
)
1077 return q
->limits
.io_min
;
1080 static inline int bdev_io_min(struct block_device
*bdev
)
1082 return queue_io_min(bdev_get_queue(bdev
));
1085 static inline unsigned int queue_io_opt(struct request_queue
*q
)
1087 return q
->limits
.io_opt
;
1090 static inline int bdev_io_opt(struct block_device
*bdev
)
1092 return queue_io_opt(bdev_get_queue(bdev
));
1095 static inline int queue_alignment_offset(struct request_queue
*q
)
1097 if (q
->limits
.misaligned
)
1100 return q
->limits
.alignment_offset
;
1103 static inline int queue_limit_alignment_offset(struct queue_limits
*lim
, sector_t sector
)
1105 unsigned int granularity
= max(lim
->physical_block_size
, lim
->io_min
);
1106 unsigned int alignment
= (sector
<< 9) & (granularity
- 1);
1108 return (granularity
+ lim
->alignment_offset
- alignment
)
1109 & (granularity
- 1);
1112 static inline int bdev_alignment_offset(struct block_device
*bdev
)
1114 struct request_queue
*q
= bdev_get_queue(bdev
);
1116 if (q
->limits
.misaligned
)
1119 if (bdev
!= bdev
->bd_contains
)
1120 return bdev
->bd_part
->alignment_offset
;
1122 return q
->limits
.alignment_offset
;
1125 static inline int queue_discard_alignment(struct request_queue
*q
)
1127 if (q
->limits
.discard_misaligned
)
1130 return q
->limits
.discard_alignment
;
1133 static inline int queue_limit_discard_alignment(struct queue_limits
*lim
, sector_t sector
)
1135 unsigned int alignment
= (sector
<< 9) & (lim
->discard_granularity
- 1);
1137 if (!lim
->max_discard_sectors
)
1140 return (lim
->discard_granularity
+ lim
->discard_alignment
- alignment
)
1141 & (lim
->discard_granularity
- 1);
1144 static inline int bdev_discard_alignment(struct block_device
*bdev
)
1146 struct request_queue
*q
= bdev_get_queue(bdev
);
1148 if (bdev
!= bdev
->bd_contains
)
1149 return bdev
->bd_part
->discard_alignment
;
1151 return q
->limits
.discard_alignment
;
1154 static inline unsigned int queue_discard_zeroes_data(struct request_queue
*q
)
1156 if (q
->limits
.max_discard_sectors
&& q
->limits
.discard_zeroes_data
== 1)
1162 static inline unsigned int bdev_discard_zeroes_data(struct block_device
*bdev
)
1164 return queue_discard_zeroes_data(bdev_get_queue(bdev
));
1167 static inline int queue_dma_alignment(struct request_queue
*q
)
1169 return q
? q
->dma_alignment
: 511;
1172 static inline int blk_rq_aligned(struct request_queue
*q
, unsigned long addr
,
1175 unsigned int alignment
= queue_dma_alignment(q
) | q
->dma_pad_mask
;
1176 return !(addr
& alignment
) && !(len
& alignment
);
1179 /* assumes size > 256 */
1180 static inline unsigned int blksize_bits(unsigned int size
)
1182 unsigned int bits
= 8;
1186 } while (size
> 256);
1190 static inline unsigned int block_size(struct block_device
*bdev
)
1192 return bdev
->bd_block_size
;
1195 static inline bool queue_flush_queueable(struct request_queue
*q
)
1197 return !q
->flush_not_queueable
;
1200 typedef struct {struct page
*v
;} Sector
;
1202 unsigned char *read_dev_sector(struct block_device
*, sector_t
, Sector
*);
1204 static inline void put_dev_sector(Sector p
)
1206 page_cache_release(p
.v
);
1210 int kblockd_schedule_work(struct request_queue
*q
, struct work_struct
*work
);
1212 #ifdef CONFIG_BLK_CGROUP
1214 * This should not be using sched_clock(). A real patch is in progress
1215 * to fix this up, until that is in place we need to disable preemption
1216 * around sched_clock() in this function and set_io_start_time_ns().
1218 static inline void set_start_time_ns(struct request
*req
)
1221 req
->start_time_ns
= sched_clock();
1225 static inline void set_io_start_time_ns(struct request
*req
)
1228 req
->io_start_time_ns
= sched_clock();
1232 static inline uint64_t rq_start_time_ns(struct request
*req
)
1234 return req
->start_time_ns
;
1237 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1239 return req
->io_start_time_ns
;
1242 static inline void set_start_time_ns(struct request
*req
) {}
1243 static inline void set_io_start_time_ns(struct request
*req
) {}
1244 static inline uint64_t rq_start_time_ns(struct request
*req
)
1248 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1254 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1255 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1256 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1257 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1259 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1261 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1262 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1264 struct blk_integrity_exchg
{
1268 unsigned int data_size
;
1269 unsigned short sector_size
;
1270 const char *disk_name
;
1273 typedef void (integrity_gen_fn
) (struct blk_integrity_exchg
*);
1274 typedef int (integrity_vrfy_fn
) (struct blk_integrity_exchg
*);
1275 typedef void (integrity_set_tag_fn
) (void *, void *, unsigned int);
1276 typedef void (integrity_get_tag_fn
) (void *, void *, unsigned int);
1278 struct blk_integrity
{
1279 integrity_gen_fn
*generate_fn
;
1280 integrity_vrfy_fn
*verify_fn
;
1281 integrity_set_tag_fn
*set_tag_fn
;
1282 integrity_get_tag_fn
*get_tag_fn
;
1284 unsigned short flags
;
1285 unsigned short tuple_size
;
1286 unsigned short sector_size
;
1287 unsigned short tag_size
;
1291 struct kobject kobj
;
1294 extern bool blk_integrity_is_initialized(struct gendisk
*);
1295 extern int blk_integrity_register(struct gendisk
*, struct blk_integrity
*);
1296 extern void blk_integrity_unregister(struct gendisk
*);
1297 extern int blk_integrity_compare(struct gendisk
*, struct gendisk
*);
1298 extern int blk_rq_map_integrity_sg(struct request_queue
*, struct bio
*,
1299 struct scatterlist
*);
1300 extern int blk_rq_count_integrity_sg(struct request_queue
*, struct bio
*);
1301 extern int blk_integrity_merge_rq(struct request_queue
*, struct request
*,
1303 extern int blk_integrity_merge_bio(struct request_queue
*, struct request
*,
1307 struct blk_integrity
*bdev_get_integrity(struct block_device
*bdev
)
1309 return bdev
->bd_disk
->integrity
;
1312 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1314 return disk
->integrity
;
1317 static inline int blk_integrity_rq(struct request
*rq
)
1319 if (rq
->bio
== NULL
)
1322 return bio_integrity(rq
->bio
);
1325 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1328 q
->limits
.max_integrity_segments
= segs
;
1331 static inline unsigned short
1332 queue_max_integrity_segments(struct request_queue
*q
)
1334 return q
->limits
.max_integrity_segments
;
1337 #else /* CONFIG_BLK_DEV_INTEGRITY */
1340 struct block_device
;
1342 struct blk_integrity
;
1344 static inline int blk_integrity_rq(struct request
*rq
)
1348 static inline int blk_rq_count_integrity_sg(struct request_queue
*q
,
1353 static inline int blk_rq_map_integrity_sg(struct request_queue
*q
,
1355 struct scatterlist
*s
)
1359 static inline struct blk_integrity
*bdev_get_integrity(struct block_device
*b
)
1363 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1367 static inline int blk_integrity_compare(struct gendisk
*a
, struct gendisk
*b
)
1371 static inline int blk_integrity_register(struct gendisk
*d
,
1372 struct blk_integrity
*b
)
1376 static inline void blk_integrity_unregister(struct gendisk
*d
)
1379 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1383 static inline unsigned short queue_max_integrity_segments(struct request_queue
*q
)
1387 static inline int blk_integrity_merge_rq(struct request_queue
*rq
,
1393 static inline int blk_integrity_merge_bio(struct request_queue
*rq
,
1399 static inline bool blk_integrity_is_initialized(struct gendisk
*g
)
1404 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1406 struct block_device_operations
{
1407 int (*open
) (struct block_device
*, fmode_t
);
1408 int (*release
) (struct gendisk
*, fmode_t
);
1409 int (*ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1410 int (*compat_ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1411 int (*direct_access
) (struct block_device
*, sector_t
,
1412 void **, unsigned long *);
1413 unsigned int (*check_events
) (struct gendisk
*disk
,
1414 unsigned int clearing
);
1415 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1416 int (*media_changed
) (struct gendisk
*);
1417 void (*unlock_native_capacity
) (struct gendisk
*);
1418 int (*revalidate_disk
) (struct gendisk
*);
1419 int (*getgeo
)(struct block_device
*, struct hd_geometry
*);
1420 /* this callback is with swap_lock and sometimes page table lock held */
1421 void (*swap_slot_free_notify
) (struct block_device
*, unsigned long);
1422 struct module
*owner
;
1425 extern int __blkdev_driver_ioctl(struct block_device
*, fmode_t
, unsigned int,
1427 #else /* CONFIG_BLOCK */
1429 * stubs for when the block layer is configured out
1431 #define buffer_heads_over_limit 0
1433 static inline long nr_blockdev_pages(void)
1441 static inline void blk_start_plug(struct blk_plug
*plug
)
1445 static inline void blk_finish_plug(struct blk_plug
*plug
)
1449 static inline void blk_flush_plug(struct task_struct
*task
)
1453 static inline void blk_schedule_flush_plug(struct task_struct
*task
)
1458 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
1463 #endif /* CONFIG_BLOCK */