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 */
56 * count[], starved[], and wait[] are indexed by
57 * BLK_RW_SYNC/BLK_RW_ASYNC
62 wait_queue_head_t wait
[2];
67 * request command types
69 enum rq_cmd_type_bits
{
70 REQ_TYPE_FS
= 1, /* fs request */
71 REQ_TYPE_BLOCK_PC
, /* scsi command */
72 REQ_TYPE_SENSE
, /* sense request */
73 REQ_TYPE_PM_SUSPEND
, /* suspend request */
74 REQ_TYPE_PM_RESUME
, /* resume request */
75 REQ_TYPE_PM_SHUTDOWN
, /* shutdown request */
76 REQ_TYPE_SPECIAL
, /* driver defined type */
78 * for ATA/ATAPI devices. this really doesn't belong here, ide should
79 * use REQ_TYPE_SPECIAL and use rq->cmd[0] with the range of driver
80 * private REQ_LB opcodes to differentiate what type of request this is
82 REQ_TYPE_ATA_TASKFILE
,
86 #define BLK_MAX_CDB 16
89 * try to put the fields that are referenced together in the same cacheline.
90 * if you modify this structure, be sure to check block/blk-core.c:blk_rq_init()
94 struct list_head queuelist
;
95 struct call_single_data csd
;
97 struct request_queue
*q
;
99 unsigned int cmd_flags
;
100 enum rq_cmd_type_bits cmd_type
;
101 unsigned long atomic_flags
;
105 /* the following two fields are internal, NEVER access directly */
106 unsigned int __data_len
; /* total data len */
107 sector_t __sector
; /* sector cursor */
112 struct hlist_node hash
; /* merge hash */
114 * The rb_node is only used inside the io scheduler, requests
115 * are pruned when moved to the dispatch queue. So let the
116 * completion_data share space with the rb_node.
119 struct rb_node rb_node
; /* sort/lookup */
120 void *completion_data
;
124 * Three pointers are available for the IO schedulers, if they need
125 * more they have to dynamically allocate it. Flush requests are
126 * never put on the IO scheduler. So let the flush fields share
127 * space with the elevator data.
137 struct list_head list
;
138 rq_end_io_fn
*saved_end_io
;
142 struct gendisk
*rq_disk
;
143 struct hd_struct
*part
;
144 unsigned long start_time
;
145 #ifdef CONFIG_BLK_CGROUP
146 unsigned long long start_time_ns
;
147 unsigned long long io_start_time_ns
; /* when passed to hardware */
149 /* Number of scatter-gather DMA addr+len pairs after
150 * physical address coalescing is performed.
152 unsigned short nr_phys_segments
;
153 #if defined(CONFIG_BLK_DEV_INTEGRITY)
154 unsigned short nr_integrity_segments
;
157 unsigned short ioprio
;
161 void *special
; /* opaque pointer available for LLD use */
162 char *buffer
; /* kaddr of the current segment if available */
168 * when request is used as a packet command carrier
170 unsigned char __cmd
[BLK_MAX_CDB
];
172 unsigned short cmd_len
;
174 unsigned int extra_len
; /* length of alignment and padding */
175 unsigned int sense_len
;
176 unsigned int resid_len
; /* residual count */
179 unsigned long deadline
;
180 struct list_head timeout_list
;
181 unsigned int timeout
;
185 * completion callback.
187 rq_end_io_fn
*end_io
;
191 struct request
*next_rq
;
194 static inline unsigned short req_get_ioprio(struct request
*req
)
200 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
201 * requests. Some step values could eventually be made generic.
203 struct request_pm_state
205 /* PM state machine step value, currently driver specific */
207 /* requested PM state value (S1, S2, S3, S4, ...) */
209 void* data
; /* for driver use */
212 #include <linux/elevator.h>
214 typedef void (request_fn_proc
) (struct request_queue
*q
);
215 typedef void (make_request_fn
) (struct request_queue
*q
, struct bio
*bio
);
216 typedef int (prep_rq_fn
) (struct request_queue
*, struct request
*);
217 typedef void (unprep_rq_fn
) (struct request_queue
*, struct request
*);
220 struct bvec_merge_data
{
221 struct block_device
*bi_bdev
;
226 typedef int (merge_bvec_fn
) (struct request_queue
*, struct bvec_merge_data
*,
228 typedef void (softirq_done_fn
)(struct request
*);
229 typedef int (dma_drain_needed_fn
)(struct request
*);
230 typedef int (lld_busy_fn
) (struct request_queue
*q
);
231 typedef int (bsg_job_fn
) (struct bsg_job
*);
233 enum blk_eh_timer_return
{
239 typedef enum blk_eh_timer_return (rq_timed_out_fn
)(struct request
*);
241 enum blk_queue_state
{
246 struct blk_queue_tag
{
247 struct request
**tag_index
; /* map of busy tags */
248 unsigned long *tag_map
; /* bit map of free/busy tags */
249 int busy
; /* current depth */
250 int max_depth
; /* what we will send to device */
251 int real_max_depth
; /* what the array can hold */
252 atomic_t refcnt
; /* map can be shared */
255 #define BLK_SCSI_MAX_CMDS (256)
256 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
258 struct queue_limits
{
259 unsigned long bounce_pfn
;
260 unsigned long seg_boundary_mask
;
262 unsigned int max_hw_sectors
;
263 unsigned int max_sectors
;
264 unsigned int max_segment_size
;
265 unsigned int physical_block_size
;
266 unsigned int alignment_offset
;
269 unsigned int max_discard_sectors
;
270 unsigned int discard_granularity
;
271 unsigned int discard_alignment
;
273 unsigned short logical_block_size
;
274 unsigned short max_segments
;
275 unsigned short max_integrity_segments
;
277 unsigned char misaligned
;
278 unsigned char discard_misaligned
;
279 unsigned char cluster
;
280 unsigned char discard_zeroes_data
;
283 struct request_queue
{
285 * Together with queue_head for cacheline sharing
287 struct list_head queue_head
;
288 struct request
*last_merge
;
289 struct elevator_queue
*elevator
;
290 int nr_rqs
[2]; /* # allocated [a]sync rqs */
291 int nr_rqs_elvpriv
; /* # allocated rqs w/ elvpriv */
294 * the queue request freelist, one for reads and one for writes
296 struct request_list rq
;
298 request_fn_proc
*request_fn
;
299 make_request_fn
*make_request_fn
;
300 prep_rq_fn
*prep_rq_fn
;
301 unprep_rq_fn
*unprep_rq_fn
;
302 merge_bvec_fn
*merge_bvec_fn
;
303 softirq_done_fn
*softirq_done_fn
;
304 rq_timed_out_fn
*rq_timed_out_fn
;
305 dma_drain_needed_fn
*dma_drain_needed
;
306 lld_busy_fn
*lld_busy_fn
;
309 * Dispatch queue sorting
312 struct request
*boundary_rq
;
315 * Delayed queue handling
317 struct delayed_work delay_work
;
319 struct backing_dev_info backing_dev_info
;
322 * The queue owner gets to use this for whatever they like.
323 * ll_rw_blk doesn't touch it.
328 * various queue flags, see QUEUE_* below
330 unsigned long queue_flags
;
333 * ida allocated id for this queue. Used to index queues from
339 * queue needs bounce pages for pages above this limit
344 * protects queue structures from reentrancy. ->__queue_lock should
345 * _never_ be used directly, it is queue private. always use
348 spinlock_t __queue_lock
;
349 spinlock_t
*queue_lock
;
359 unsigned long nr_requests
; /* Max # of requests */
360 unsigned int nr_congestion_on
;
361 unsigned int nr_congestion_off
;
362 unsigned int nr_batching
;
364 unsigned int dma_drain_size
;
365 void *dma_drain_buffer
;
366 unsigned int dma_pad_mask
;
367 unsigned int dma_alignment
;
369 struct blk_queue_tag
*queue_tags
;
370 struct list_head tag_busy_list
;
372 unsigned int nr_sorted
;
373 unsigned int in_flight
[2];
375 unsigned int rq_timeout
;
376 struct timer_list timeout
;
377 struct list_head timeout_list
;
379 struct list_head icq_list
;
380 #ifdef CONFIG_BLK_CGROUP
381 DECLARE_BITMAP (blkcg_pols
, BLKCG_MAX_POLS
);
382 struct blkcg_gq
*root_blkg
;
383 struct list_head blkg_list
;
386 struct queue_limits limits
;
391 unsigned int sg_timeout
;
392 unsigned int sg_reserved_size
;
394 #ifdef CONFIG_BLK_DEV_IO_TRACE
395 struct blk_trace
*blk_trace
;
398 * for flush operations
400 unsigned int flush_flags
;
401 unsigned int flush_not_queueable
:1;
402 unsigned int flush_queue_delayed
:1;
403 unsigned int flush_pending_idx
:1;
404 unsigned int flush_running_idx
:1;
405 unsigned long flush_pending_since
;
406 struct list_head flush_queue
[2];
407 struct list_head flush_data_in_flight
;
408 struct request flush_rq
;
410 struct mutex sysfs_lock
;
414 #if defined(CONFIG_BLK_DEV_BSG)
415 bsg_job_fn
*bsg_job_fn
;
417 struct bsg_class_device bsg_dev
;
420 #ifdef CONFIG_BLK_CGROUP
421 struct list_head all_q_node
;
423 #ifdef CONFIG_BLK_DEV_THROTTLING
425 struct throtl_data
*td
;
429 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
430 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
431 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
432 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
433 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
434 #define QUEUE_FLAG_BYPASS 6 /* act as dumb FIFO queue */
435 #define QUEUE_FLAG_BIDI 7 /* queue supports bidi requests */
436 #define QUEUE_FLAG_NOMERGES 8 /* disable merge attempts */
437 #define QUEUE_FLAG_SAME_COMP 9 /* complete on same CPU-group */
438 #define QUEUE_FLAG_FAIL_IO 10 /* fake timeout */
439 #define QUEUE_FLAG_STACKABLE 11 /* supports request stacking */
440 #define QUEUE_FLAG_NONROT 12 /* non-rotational device (SSD) */
441 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
442 #define QUEUE_FLAG_IO_STAT 13 /* do IO stats */
443 #define QUEUE_FLAG_DISCARD 14 /* supports DISCARD */
444 #define QUEUE_FLAG_NOXMERGES 15 /* No extended merges */
445 #define QUEUE_FLAG_ADD_RANDOM 16 /* Contributes to random pool */
446 #define QUEUE_FLAG_SECDISCARD 17 /* supports SECDISCARD */
447 #define QUEUE_FLAG_SAME_FORCE 18 /* force complete on same CPU */
449 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
450 (1 << QUEUE_FLAG_STACKABLE) | \
451 (1 << QUEUE_FLAG_SAME_COMP) | \
452 (1 << QUEUE_FLAG_ADD_RANDOM))
454 static inline void queue_lockdep_assert_held(struct request_queue
*q
)
457 lockdep_assert_held(q
->queue_lock
);
460 static inline void queue_flag_set_unlocked(unsigned int flag
,
461 struct request_queue
*q
)
463 __set_bit(flag
, &q
->queue_flags
);
466 static inline int queue_flag_test_and_clear(unsigned int flag
,
467 struct request_queue
*q
)
469 queue_lockdep_assert_held(q
);
471 if (test_bit(flag
, &q
->queue_flags
)) {
472 __clear_bit(flag
, &q
->queue_flags
);
479 static inline int queue_flag_test_and_set(unsigned int flag
,
480 struct request_queue
*q
)
482 queue_lockdep_assert_held(q
);
484 if (!test_bit(flag
, &q
->queue_flags
)) {
485 __set_bit(flag
, &q
->queue_flags
);
492 static inline void queue_flag_set(unsigned int flag
, struct request_queue
*q
)
494 queue_lockdep_assert_held(q
);
495 __set_bit(flag
, &q
->queue_flags
);
498 static inline void queue_flag_clear_unlocked(unsigned int flag
,
499 struct request_queue
*q
)
501 __clear_bit(flag
, &q
->queue_flags
);
504 static inline int queue_in_flight(struct request_queue
*q
)
506 return q
->in_flight
[0] + q
->in_flight
[1];
509 static inline void queue_flag_clear(unsigned int flag
, struct request_queue
*q
)
511 queue_lockdep_assert_held(q
);
512 __clear_bit(flag
, &q
->queue_flags
);
515 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
516 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
517 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
518 #define blk_queue_bypass(q) test_bit(QUEUE_FLAG_BYPASS, &(q)->queue_flags)
519 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
520 #define blk_queue_noxmerges(q) \
521 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
522 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
523 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
524 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
525 #define blk_queue_stackable(q) \
526 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
527 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
528 #define blk_queue_secdiscard(q) (blk_queue_discard(q) && \
529 test_bit(QUEUE_FLAG_SECDISCARD, &(q)->queue_flags))
531 #define blk_noretry_request(rq) \
532 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
533 REQ_FAILFAST_DRIVER))
535 #define blk_account_rq(rq) \
536 (((rq)->cmd_flags & REQ_STARTED) && \
537 ((rq)->cmd_type == REQ_TYPE_FS || \
538 ((rq)->cmd_flags & REQ_DISCARD)))
540 #define blk_pm_request(rq) \
541 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
542 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
544 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
545 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
546 /* rq->queuelist of dequeued request must be list_empty() */
547 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
549 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
551 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
553 static inline unsigned int blk_queue_cluster(struct request_queue
*q
)
555 return q
->limits
.cluster
;
559 * We regard a request as sync, if either a read or a sync write
561 static inline bool rw_is_sync(unsigned int rw_flags
)
563 return !(rw_flags
& REQ_WRITE
) || (rw_flags
& REQ_SYNC
);
566 static inline bool rq_is_sync(struct request
*rq
)
568 return rw_is_sync(rq
->cmd_flags
);
571 static inline bool blk_rl_full(struct request_list
*rl
, bool sync
)
573 unsigned int flag
= sync
? BLK_RL_SYNCFULL
: BLK_RL_ASYNCFULL
;
575 return rl
->flags
& flag
;
578 static inline void blk_set_rl_full(struct request_list
*rl
, bool sync
)
580 unsigned int flag
= sync
? BLK_RL_SYNCFULL
: BLK_RL_ASYNCFULL
;
585 static inline void blk_clear_rl_full(struct request_list
*rl
, bool sync
)
587 unsigned int flag
= sync
? BLK_RL_SYNCFULL
: BLK_RL_ASYNCFULL
;
594 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
595 * it already be started by driver.
597 #define RQ_NOMERGE_FLAGS \
598 (REQ_NOMERGE | REQ_STARTED | REQ_SOFTBARRIER | REQ_FLUSH | REQ_FUA)
599 #define rq_mergeable(rq) \
600 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
601 (((rq)->cmd_flags & REQ_DISCARD) || \
602 (rq)->cmd_type == REQ_TYPE_FS))
605 * q->prep_rq_fn return values
607 #define BLKPREP_OK 0 /* serve it */
608 #define BLKPREP_KILL 1 /* fatal error, kill */
609 #define BLKPREP_DEFER 2 /* leave on queue */
611 extern unsigned long blk_max_low_pfn
, blk_max_pfn
;
614 * standard bounce addresses:
616 * BLK_BOUNCE_HIGH : bounce all highmem pages
617 * BLK_BOUNCE_ANY : don't bounce anything
618 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
621 #if BITS_PER_LONG == 32
622 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
624 #define BLK_BOUNCE_HIGH -1ULL
626 #define BLK_BOUNCE_ANY (-1ULL)
627 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
630 * default timeout for SG_IO if none specified
632 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
633 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
636 extern int init_emergency_isa_pool(void);
637 extern void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
);
639 static inline int init_emergency_isa_pool(void)
643 static inline void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
)
646 #endif /* CONFIG_MMU */
652 unsigned long offset
;
657 struct req_iterator
{
662 /* This should not be used directly - use rq_for_each_segment */
663 #define for_each_bio(_bio) \
664 for (; _bio; _bio = _bio->bi_next)
665 #define __rq_for_each_bio(_bio, rq) \
667 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
669 #define rq_for_each_segment(bvl, _rq, _iter) \
670 __rq_for_each_bio(_iter.bio, _rq) \
671 bio_for_each_segment(bvl, _iter.bio, _iter.i)
673 #define rq_iter_last(rq, _iter) \
674 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
676 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
677 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
679 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
680 extern void rq_flush_dcache_pages(struct request
*rq
);
682 static inline void rq_flush_dcache_pages(struct request
*rq
)
687 extern int blk_register_queue(struct gendisk
*disk
);
688 extern void blk_unregister_queue(struct gendisk
*disk
);
689 extern void generic_make_request(struct bio
*bio
);
690 extern void blk_rq_init(struct request_queue
*q
, struct request
*rq
);
691 extern void blk_put_request(struct request
*);
692 extern void __blk_put_request(struct request_queue
*, struct request
*);
693 extern struct request
*blk_get_request(struct request_queue
*, int, gfp_t
);
694 extern struct request
*blk_make_request(struct request_queue
*, struct bio
*,
696 extern void blk_requeue_request(struct request_queue
*, struct request
*);
697 extern void blk_add_request_payload(struct request
*rq
, struct page
*page
,
699 extern int blk_rq_check_limits(struct request_queue
*q
, struct request
*rq
);
700 extern int blk_lld_busy(struct request_queue
*q
);
701 extern int blk_rq_prep_clone(struct request
*rq
, struct request
*rq_src
,
702 struct bio_set
*bs
, gfp_t gfp_mask
,
703 int (*bio_ctr
)(struct bio
*, struct bio
*, void *),
705 extern void blk_rq_unprep_clone(struct request
*rq
);
706 extern int blk_insert_cloned_request(struct request_queue
*q
,
708 extern void blk_delay_queue(struct request_queue
*, unsigned long);
709 extern void blk_recount_segments(struct request_queue
*, struct bio
*);
710 extern int scsi_verify_blk_ioctl(struct block_device
*, unsigned int);
711 extern int scsi_cmd_blk_ioctl(struct block_device
*, fmode_t
,
712 unsigned int, void __user
*);
713 extern int scsi_cmd_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
714 unsigned int, void __user
*);
715 extern int sg_scsi_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
716 struct scsi_ioctl_command __user
*);
718 extern void blk_queue_bio(struct request_queue
*q
, struct bio
*bio
);
721 * A queue has just exitted congestion. Note this in the global counter of
722 * congested queues, and wake up anyone who was waiting for requests to be
725 static inline void blk_clear_queue_congested(struct request_queue
*q
, int sync
)
727 clear_bdi_congested(&q
->backing_dev_info
, sync
);
731 * A queue has just entered congestion. Flag that in the queue's VM-visible
732 * state flags and increment the global gounter of congested queues.
734 static inline void blk_set_queue_congested(struct request_queue
*q
, int sync
)
736 set_bdi_congested(&q
->backing_dev_info
, sync
);
739 extern void blk_start_queue(struct request_queue
*q
);
740 extern void blk_stop_queue(struct request_queue
*q
);
741 extern void blk_sync_queue(struct request_queue
*q
);
742 extern void __blk_stop_queue(struct request_queue
*q
);
743 extern void __blk_run_queue(struct request_queue
*q
);
744 extern void blk_run_queue(struct request_queue
*);
745 extern void blk_run_queue_async(struct request_queue
*q
);
746 extern int blk_rq_map_user(struct request_queue
*, struct request
*,
747 struct rq_map_data
*, void __user
*, unsigned long,
749 extern int blk_rq_unmap_user(struct bio
*);
750 extern int blk_rq_map_kern(struct request_queue
*, struct request
*, void *, unsigned int, gfp_t
);
751 extern int blk_rq_map_user_iov(struct request_queue
*, struct request
*,
752 struct rq_map_data
*, struct sg_iovec
*, int,
753 unsigned int, gfp_t
);
754 extern int blk_execute_rq(struct request_queue
*, struct gendisk
*,
755 struct request
*, int);
756 extern void blk_execute_rq_nowait(struct request_queue
*, struct gendisk
*,
757 struct request
*, int, rq_end_io_fn
*);
759 static inline struct request_queue
*bdev_get_queue(struct block_device
*bdev
)
761 return bdev
->bd_disk
->queue
;
765 * blk_rq_pos() : the current sector
766 * blk_rq_bytes() : bytes left in the entire request
767 * blk_rq_cur_bytes() : bytes left in the current segment
768 * blk_rq_err_bytes() : bytes left till the next error boundary
769 * blk_rq_sectors() : sectors left in the entire request
770 * blk_rq_cur_sectors() : sectors left in the current segment
772 static inline sector_t
blk_rq_pos(const struct request
*rq
)
777 static inline unsigned int blk_rq_bytes(const struct request
*rq
)
779 return rq
->__data_len
;
782 static inline int blk_rq_cur_bytes(const struct request
*rq
)
784 return rq
->bio
? bio_cur_bytes(rq
->bio
) : 0;
787 extern unsigned int blk_rq_err_bytes(const struct request
*rq
);
789 static inline unsigned int blk_rq_sectors(const struct request
*rq
)
791 return blk_rq_bytes(rq
) >> 9;
794 static inline unsigned int blk_rq_cur_sectors(const struct request
*rq
)
796 return blk_rq_cur_bytes(rq
) >> 9;
800 * Request issue related functions.
802 extern struct request
*blk_peek_request(struct request_queue
*q
);
803 extern void blk_start_request(struct request
*rq
);
804 extern struct request
*blk_fetch_request(struct request_queue
*q
);
807 * Request completion related functions.
809 * blk_update_request() completes given number of bytes and updates
810 * the request without completing it.
812 * blk_end_request() and friends. __blk_end_request() must be called
813 * with the request queue spinlock acquired.
815 * Several drivers define their own end_request and call
816 * blk_end_request() for parts of the original function.
817 * This prevents code duplication in drivers.
819 extern bool blk_update_request(struct request
*rq
, int error
,
820 unsigned int nr_bytes
);
821 extern bool blk_end_request(struct request
*rq
, int error
,
822 unsigned int nr_bytes
);
823 extern void blk_end_request_all(struct request
*rq
, int error
);
824 extern bool blk_end_request_cur(struct request
*rq
, int error
);
825 extern bool blk_end_request_err(struct request
*rq
, int error
);
826 extern bool __blk_end_request(struct request
*rq
, int error
,
827 unsigned int nr_bytes
);
828 extern void __blk_end_request_all(struct request
*rq
, int error
);
829 extern bool __blk_end_request_cur(struct request
*rq
, int error
);
830 extern bool __blk_end_request_err(struct request
*rq
, int error
);
832 extern void blk_complete_request(struct request
*);
833 extern void __blk_complete_request(struct request
*);
834 extern void blk_abort_request(struct request
*);
835 extern void blk_unprep_request(struct request
*);
838 * Access functions for manipulating queue properties
840 extern struct request_queue
*blk_init_queue_node(request_fn_proc
*rfn
,
841 spinlock_t
*lock
, int node_id
);
842 extern struct request_queue
*blk_init_queue(request_fn_proc
*, spinlock_t
*);
843 extern struct request_queue
*blk_init_allocated_queue(struct request_queue
*,
844 request_fn_proc
*, spinlock_t
*);
845 extern void blk_cleanup_queue(struct request_queue
*);
846 extern void blk_queue_make_request(struct request_queue
*, make_request_fn
*);
847 extern void blk_queue_bounce_limit(struct request_queue
*, u64
);
848 extern void blk_limits_max_hw_sectors(struct queue_limits
*, unsigned int);
849 extern void blk_queue_max_hw_sectors(struct request_queue
*, unsigned int);
850 extern void blk_queue_max_segments(struct request_queue
*, unsigned short);
851 extern void blk_queue_max_segment_size(struct request_queue
*, unsigned int);
852 extern void blk_queue_max_discard_sectors(struct request_queue
*q
,
853 unsigned int max_discard_sectors
);
854 extern void blk_queue_logical_block_size(struct request_queue
*, unsigned short);
855 extern void blk_queue_physical_block_size(struct request_queue
*, unsigned int);
856 extern void blk_queue_alignment_offset(struct request_queue
*q
,
857 unsigned int alignment
);
858 extern void blk_limits_io_min(struct queue_limits
*limits
, unsigned int min
);
859 extern void blk_queue_io_min(struct request_queue
*q
, unsigned int min
);
860 extern void blk_limits_io_opt(struct queue_limits
*limits
, unsigned int opt
);
861 extern void blk_queue_io_opt(struct request_queue
*q
, unsigned int opt
);
862 extern void blk_set_default_limits(struct queue_limits
*lim
);
863 extern void blk_set_stacking_limits(struct queue_limits
*lim
);
864 extern int blk_stack_limits(struct queue_limits
*t
, struct queue_limits
*b
,
866 extern int bdev_stack_limits(struct queue_limits
*t
, struct block_device
*bdev
,
868 extern void disk_stack_limits(struct gendisk
*disk
, struct block_device
*bdev
,
870 extern void blk_queue_stack_limits(struct request_queue
*t
, struct request_queue
*b
);
871 extern void blk_queue_dma_pad(struct request_queue
*, unsigned int);
872 extern void blk_queue_update_dma_pad(struct request_queue
*, unsigned int);
873 extern int blk_queue_dma_drain(struct request_queue
*q
,
874 dma_drain_needed_fn
*dma_drain_needed
,
875 void *buf
, unsigned int size
);
876 extern void blk_queue_lld_busy(struct request_queue
*q
, lld_busy_fn
*fn
);
877 extern void blk_queue_segment_boundary(struct request_queue
*, unsigned long);
878 extern void blk_queue_prep_rq(struct request_queue
*, prep_rq_fn
*pfn
);
879 extern void blk_queue_unprep_rq(struct request_queue
*, unprep_rq_fn
*ufn
);
880 extern void blk_queue_merge_bvec(struct request_queue
*, merge_bvec_fn
*);
881 extern void blk_queue_dma_alignment(struct request_queue
*, int);
882 extern void blk_queue_update_dma_alignment(struct request_queue
*, int);
883 extern void blk_queue_softirq_done(struct request_queue
*, softirq_done_fn
*);
884 extern void blk_queue_rq_timed_out(struct request_queue
*, rq_timed_out_fn
*);
885 extern void blk_queue_rq_timeout(struct request_queue
*, unsigned int);
886 extern void blk_queue_flush(struct request_queue
*q
, unsigned int flush
);
887 extern void blk_queue_flush_queueable(struct request_queue
*q
, bool queueable
);
888 extern struct backing_dev_info
*blk_get_backing_dev_info(struct block_device
*bdev
);
890 extern int blk_rq_map_sg(struct request_queue
*, struct request
*, struct scatterlist
*);
891 extern void blk_dump_rq_flags(struct request
*, char *);
892 extern long nr_blockdev_pages(void);
894 bool __must_check
blk_get_queue(struct request_queue
*);
895 struct request_queue
*blk_alloc_queue(gfp_t
);
896 struct request_queue
*blk_alloc_queue_node(gfp_t
, int);
897 extern void blk_put_queue(struct request_queue
*);
900 * blk_plug permits building a queue of related requests by holding the I/O
901 * fragments for a short period. This allows merging of sequential requests
902 * into single larger request. As the requests are moved from a per-task list to
903 * the device's request_queue in a batch, this results in improved scalability
904 * as the lock contention for request_queue lock is reduced.
906 * It is ok not to disable preemption when adding the request to the plug list
907 * or when attempting a merge, because blk_schedule_flush_list() will only flush
908 * the plug list when the task sleeps by itself. For details, please see
909 * schedule() where blk_schedule_flush_plug() is called.
912 unsigned long magic
; /* detect uninitialized use-cases */
913 struct list_head list
; /* requests */
914 struct list_head cb_list
; /* md requires an unplug callback */
915 unsigned int should_sort
; /* list to be sorted before flushing? */
917 #define BLK_MAX_REQUEST_COUNT 16
920 struct list_head list
;
921 void (*callback
)(struct blk_plug_cb
*);
924 extern void blk_start_plug(struct blk_plug
*);
925 extern void blk_finish_plug(struct blk_plug
*);
926 extern void blk_flush_plug_list(struct blk_plug
*, bool);
928 static inline void blk_flush_plug(struct task_struct
*tsk
)
930 struct blk_plug
*plug
= tsk
->plug
;
933 blk_flush_plug_list(plug
, false);
936 static inline void blk_schedule_flush_plug(struct task_struct
*tsk
)
938 struct blk_plug
*plug
= tsk
->plug
;
941 blk_flush_plug_list(plug
, true);
944 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
946 struct blk_plug
*plug
= tsk
->plug
;
948 return plug
&& (!list_empty(&plug
->list
) || !list_empty(&plug
->cb_list
));
954 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
955 extern int blk_queue_start_tag(struct request_queue
*, struct request
*);
956 extern struct request
*blk_queue_find_tag(struct request_queue
*, int);
957 extern void blk_queue_end_tag(struct request_queue
*, struct request
*);
958 extern int blk_queue_init_tags(struct request_queue
*, int, struct blk_queue_tag
*);
959 extern void blk_queue_free_tags(struct request_queue
*);
960 extern int blk_queue_resize_tags(struct request_queue
*, int);
961 extern void blk_queue_invalidate_tags(struct request_queue
*);
962 extern struct blk_queue_tag
*blk_init_tags(int);
963 extern void blk_free_tags(struct blk_queue_tag
*);
965 static inline struct request
*blk_map_queue_find_tag(struct blk_queue_tag
*bqt
,
968 if (unlikely(bqt
== NULL
|| tag
>= bqt
->real_max_depth
))
970 return bqt
->tag_index
[tag
];
973 #define BLKDEV_DISCARD_SECURE 0x01 /* secure discard */
975 extern int blkdev_issue_flush(struct block_device
*, gfp_t
, sector_t
*);
976 extern int blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
977 sector_t nr_sects
, gfp_t gfp_mask
, unsigned long flags
);
978 extern int blkdev_issue_zeroout(struct block_device
*bdev
, sector_t sector
,
979 sector_t nr_sects
, gfp_t gfp_mask
);
980 static inline int sb_issue_discard(struct super_block
*sb
, sector_t block
,
981 sector_t nr_blocks
, gfp_t gfp_mask
, unsigned long flags
)
983 return blkdev_issue_discard(sb
->s_bdev
, block
<< (sb
->s_blocksize_bits
- 9),
984 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
987 static inline int sb_issue_zeroout(struct super_block
*sb
, sector_t block
,
988 sector_t nr_blocks
, gfp_t gfp_mask
)
990 return blkdev_issue_zeroout(sb
->s_bdev
,
991 block
<< (sb
->s_blocksize_bits
- 9),
992 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
996 extern int blk_verify_command(unsigned char *cmd
, fmode_t has_write_perm
);
998 enum blk_default_limits
{
999 BLK_MAX_SEGMENTS
= 128,
1000 BLK_SAFE_MAX_SECTORS
= 255,
1001 BLK_DEF_MAX_SECTORS
= 1024,
1002 BLK_MAX_SEGMENT_SIZE
= 65536,
1003 BLK_SEG_BOUNDARY_MASK
= 0xFFFFFFFFUL
,
1006 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
1008 static inline unsigned long queue_bounce_pfn(struct request_queue
*q
)
1010 return q
->limits
.bounce_pfn
;
1013 static inline unsigned long queue_segment_boundary(struct request_queue
*q
)
1015 return q
->limits
.seg_boundary_mask
;
1018 static inline unsigned int queue_max_sectors(struct request_queue
*q
)
1020 return q
->limits
.max_sectors
;
1023 static inline unsigned int queue_max_hw_sectors(struct request_queue
*q
)
1025 return q
->limits
.max_hw_sectors
;
1028 static inline unsigned short queue_max_segments(struct request_queue
*q
)
1030 return q
->limits
.max_segments
;
1033 static inline unsigned int queue_max_segment_size(struct request_queue
*q
)
1035 return q
->limits
.max_segment_size
;
1038 static inline unsigned short queue_logical_block_size(struct request_queue
*q
)
1042 if (q
&& q
->limits
.logical_block_size
)
1043 retval
= q
->limits
.logical_block_size
;
1048 static inline unsigned short bdev_logical_block_size(struct block_device
*bdev
)
1050 return queue_logical_block_size(bdev_get_queue(bdev
));
1053 static inline unsigned int queue_physical_block_size(struct request_queue
*q
)
1055 return q
->limits
.physical_block_size
;
1058 static inline unsigned int bdev_physical_block_size(struct block_device
*bdev
)
1060 return queue_physical_block_size(bdev_get_queue(bdev
));
1063 static inline unsigned int queue_io_min(struct request_queue
*q
)
1065 return q
->limits
.io_min
;
1068 static inline int bdev_io_min(struct block_device
*bdev
)
1070 return queue_io_min(bdev_get_queue(bdev
));
1073 static inline unsigned int queue_io_opt(struct request_queue
*q
)
1075 return q
->limits
.io_opt
;
1078 static inline int bdev_io_opt(struct block_device
*bdev
)
1080 return queue_io_opt(bdev_get_queue(bdev
));
1083 static inline int queue_alignment_offset(struct request_queue
*q
)
1085 if (q
->limits
.misaligned
)
1088 return q
->limits
.alignment_offset
;
1091 static inline int queue_limit_alignment_offset(struct queue_limits
*lim
, sector_t sector
)
1093 unsigned int granularity
= max(lim
->physical_block_size
, lim
->io_min
);
1094 unsigned int alignment
= (sector
<< 9) & (granularity
- 1);
1096 return (granularity
+ lim
->alignment_offset
- alignment
)
1097 & (granularity
- 1);
1100 static inline int bdev_alignment_offset(struct block_device
*bdev
)
1102 struct request_queue
*q
= bdev_get_queue(bdev
);
1104 if (q
->limits
.misaligned
)
1107 if (bdev
!= bdev
->bd_contains
)
1108 return bdev
->bd_part
->alignment_offset
;
1110 return q
->limits
.alignment_offset
;
1113 static inline int queue_discard_alignment(struct request_queue
*q
)
1115 if (q
->limits
.discard_misaligned
)
1118 return q
->limits
.discard_alignment
;
1121 static inline int queue_limit_discard_alignment(struct queue_limits
*lim
, sector_t sector
)
1123 unsigned int alignment
= (sector
<< 9) & (lim
->discard_granularity
- 1);
1125 if (!lim
->max_discard_sectors
)
1128 return (lim
->discard_granularity
+ lim
->discard_alignment
- alignment
)
1129 & (lim
->discard_granularity
- 1);
1132 static inline unsigned int queue_discard_zeroes_data(struct request_queue
*q
)
1134 if (q
->limits
.max_discard_sectors
&& q
->limits
.discard_zeroes_data
== 1)
1140 static inline unsigned int bdev_discard_zeroes_data(struct block_device
*bdev
)
1142 return queue_discard_zeroes_data(bdev_get_queue(bdev
));
1145 static inline int queue_dma_alignment(struct request_queue
*q
)
1147 return q
? q
->dma_alignment
: 511;
1150 static inline int blk_rq_aligned(struct request_queue
*q
, unsigned long addr
,
1153 unsigned int alignment
= queue_dma_alignment(q
) | q
->dma_pad_mask
;
1154 return !(addr
& alignment
) && !(len
& alignment
);
1157 /* assumes size > 256 */
1158 static inline unsigned int blksize_bits(unsigned int size
)
1160 unsigned int bits
= 8;
1164 } while (size
> 256);
1168 static inline unsigned int block_size(struct block_device
*bdev
)
1170 return bdev
->bd_block_size
;
1173 static inline bool queue_flush_queueable(struct request_queue
*q
)
1175 return !q
->flush_not_queueable
;
1178 typedef struct {struct page
*v
;} Sector
;
1180 unsigned char *read_dev_sector(struct block_device
*, sector_t
, Sector
*);
1182 static inline void put_dev_sector(Sector p
)
1184 page_cache_release(p
.v
);
1188 int kblockd_schedule_work(struct request_queue
*q
, struct work_struct
*work
);
1190 #ifdef CONFIG_BLK_CGROUP
1192 * This should not be using sched_clock(). A real patch is in progress
1193 * to fix this up, until that is in place we need to disable preemption
1194 * around sched_clock() in this function and set_io_start_time_ns().
1196 static inline void set_start_time_ns(struct request
*req
)
1199 req
->start_time_ns
= sched_clock();
1203 static inline void set_io_start_time_ns(struct request
*req
)
1206 req
->io_start_time_ns
= sched_clock();
1210 static inline uint64_t rq_start_time_ns(struct request
*req
)
1212 return req
->start_time_ns
;
1215 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1217 return req
->io_start_time_ns
;
1220 static inline void set_start_time_ns(struct request
*req
) {}
1221 static inline void set_io_start_time_ns(struct request
*req
) {}
1222 static inline uint64_t rq_start_time_ns(struct request
*req
)
1226 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1232 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1233 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1234 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1235 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1237 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1239 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1240 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1242 struct blk_integrity_exchg
{
1246 unsigned int data_size
;
1247 unsigned short sector_size
;
1248 const char *disk_name
;
1251 typedef void (integrity_gen_fn
) (struct blk_integrity_exchg
*);
1252 typedef int (integrity_vrfy_fn
) (struct blk_integrity_exchg
*);
1253 typedef void (integrity_set_tag_fn
) (void *, void *, unsigned int);
1254 typedef void (integrity_get_tag_fn
) (void *, void *, unsigned int);
1256 struct blk_integrity
{
1257 integrity_gen_fn
*generate_fn
;
1258 integrity_vrfy_fn
*verify_fn
;
1259 integrity_set_tag_fn
*set_tag_fn
;
1260 integrity_get_tag_fn
*get_tag_fn
;
1262 unsigned short flags
;
1263 unsigned short tuple_size
;
1264 unsigned short sector_size
;
1265 unsigned short tag_size
;
1269 struct kobject kobj
;
1272 extern bool blk_integrity_is_initialized(struct gendisk
*);
1273 extern int blk_integrity_register(struct gendisk
*, struct blk_integrity
*);
1274 extern void blk_integrity_unregister(struct gendisk
*);
1275 extern int blk_integrity_compare(struct gendisk
*, struct gendisk
*);
1276 extern int blk_rq_map_integrity_sg(struct request_queue
*, struct bio
*,
1277 struct scatterlist
*);
1278 extern int blk_rq_count_integrity_sg(struct request_queue
*, struct bio
*);
1279 extern int blk_integrity_merge_rq(struct request_queue
*, struct request
*,
1281 extern int blk_integrity_merge_bio(struct request_queue
*, struct request
*,
1285 struct blk_integrity
*bdev_get_integrity(struct block_device
*bdev
)
1287 return bdev
->bd_disk
->integrity
;
1290 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1292 return disk
->integrity
;
1295 static inline int blk_integrity_rq(struct request
*rq
)
1297 if (rq
->bio
== NULL
)
1300 return bio_integrity(rq
->bio
);
1303 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1306 q
->limits
.max_integrity_segments
= segs
;
1309 static inline unsigned short
1310 queue_max_integrity_segments(struct request_queue
*q
)
1312 return q
->limits
.max_integrity_segments
;
1315 #else /* CONFIG_BLK_DEV_INTEGRITY */
1318 struct block_device
;
1320 struct blk_integrity
;
1322 static inline int blk_integrity_rq(struct request
*rq
)
1326 static inline int blk_rq_count_integrity_sg(struct request_queue
*q
,
1331 static inline int blk_rq_map_integrity_sg(struct request_queue
*q
,
1333 struct scatterlist
*s
)
1337 static inline struct blk_integrity
*bdev_get_integrity(struct block_device
*b
)
1341 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1345 static inline int blk_integrity_compare(struct gendisk
*a
, struct gendisk
*b
)
1349 static inline int blk_integrity_register(struct gendisk
*d
,
1350 struct blk_integrity
*b
)
1354 static inline void blk_integrity_unregister(struct gendisk
*d
)
1357 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1361 static inline unsigned short queue_max_integrity_segments(struct request_queue
*q
)
1365 static inline int blk_integrity_merge_rq(struct request_queue
*rq
,
1371 static inline int blk_integrity_merge_bio(struct request_queue
*rq
,
1377 static inline bool blk_integrity_is_initialized(struct gendisk
*g
)
1382 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1384 struct block_device_operations
{
1385 int (*open
) (struct block_device
*, fmode_t
);
1386 int (*release
) (struct gendisk
*, fmode_t
);
1387 int (*ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1388 int (*compat_ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1389 int (*direct_access
) (struct block_device
*, sector_t
,
1390 void **, unsigned long *);
1391 unsigned int (*check_events
) (struct gendisk
*disk
,
1392 unsigned int clearing
);
1393 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1394 int (*media_changed
) (struct gendisk
*);
1395 void (*unlock_native_capacity
) (struct gendisk
*);
1396 int (*revalidate_disk
) (struct gendisk
*);
1397 int (*getgeo
)(struct block_device
*, struct hd_geometry
*);
1398 /* this callback is with swap_lock and sometimes page table lock held */
1399 void (*swap_slot_free_notify
) (struct block_device
*, unsigned long);
1400 struct module
*owner
;
1403 extern int __blkdev_driver_ioctl(struct block_device
*, fmode_t
, unsigned int,
1405 #else /* CONFIG_BLOCK */
1407 * stubs for when the block layer is configured out
1409 #define buffer_heads_over_limit 0
1411 static inline long nr_blockdev_pages(void)
1419 static inline void blk_start_plug(struct blk_plug
*plug
)
1423 static inline void blk_finish_plug(struct blk_plug
*plug
)
1427 static inline void blk_flush_plug(struct task_struct
*task
)
1431 static inline void blk_schedule_flush_plug(struct task_struct
*task
)
1436 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
1441 #endif /* CONFIG_BLOCK */