1 #ifndef _LINUX_BLKDEV_H
2 #define _LINUX_BLKDEV_H
6 #include <linux/sched.h>
7 #include <linux/major.h>
8 #include <linux/genhd.h>
9 #include <linux/list.h>
10 #include <linux/timer.h>
11 #include <linux/workqueue.h>
12 #include <linux/pagemap.h>
13 #include <linux/backing-dev.h>
14 #include <linux/wait.h>
15 #include <linux/mempool.h>
16 #include <linux/bio.h>
17 #include <linux/stringify.h>
18 #include <linux/gfp.h>
19 #include <linux/bsg.h>
20 #include <linux/smp.h>
22 #include <asm/scatterlist.h>
25 struct scsi_ioctl_command
;
28 struct elevator_queue
;
29 struct request_pm_state
;
35 #define BLKDEV_MIN_RQ 4
36 #define BLKDEV_MAX_RQ 128 /* Default maximum */
39 typedef void (rq_end_io_fn
)(struct request
*, int);
43 * count[], starved[], and wait[] are indexed by
44 * BLK_RW_SYNC/BLK_RW_ASYNC
50 wait_queue_head_t wait
[2];
54 * request command types
56 enum rq_cmd_type_bits
{
57 REQ_TYPE_FS
= 1, /* fs request */
58 REQ_TYPE_BLOCK_PC
, /* scsi command */
59 REQ_TYPE_SENSE
, /* sense request */
60 REQ_TYPE_PM_SUSPEND
, /* suspend request */
61 REQ_TYPE_PM_RESUME
, /* resume request */
62 REQ_TYPE_PM_SHUTDOWN
, /* shutdown request */
63 REQ_TYPE_SPECIAL
, /* driver defined type */
65 * for ATA/ATAPI devices. this really doesn't belong here, ide should
66 * use REQ_TYPE_SPECIAL and use rq->cmd[0] with the range of driver
67 * private REQ_LB opcodes to differentiate what type of request this is
69 REQ_TYPE_ATA_TASKFILE
,
73 #define BLK_MAX_CDB 16
76 * try to put the fields that are referenced together in the same cacheline.
77 * if you modify this structure, be sure to check block/blk-core.c:blk_rq_init()
81 struct list_head queuelist
;
82 struct call_single_data csd
;
84 struct request_queue
*q
;
86 unsigned int cmd_flags
;
87 enum rq_cmd_type_bits cmd_type
;
88 unsigned long atomic_flags
;
92 /* the following two fields are internal, NEVER access directly */
93 unsigned int __data_len
; /* total data len */
94 sector_t __sector
; /* sector cursor */
99 struct hlist_node hash
; /* merge hash */
101 * The rb_node is only used inside the io scheduler, requests
102 * are pruned when moved to the dispatch queue. So let the
103 * completion_data share space with the rb_node.
106 struct rb_node rb_node
; /* sort/lookup */
107 void *completion_data
;
111 * Three pointers are available for the IO schedulers, if they need
112 * more they have to dynamically allocate it. Flush requests are
113 * never put on the IO scheduler. So let the flush fields share
114 * space with the three elevator_private pointers.
117 void *elevator_private
[3];
120 struct list_head list
;
121 rq_end_io_fn
*saved_end_io
;
125 struct gendisk
*rq_disk
;
126 struct hd_struct
*part
;
127 unsigned long start_time
;
128 #ifdef CONFIG_BLK_CGROUP
129 unsigned long long start_time_ns
;
130 unsigned long long io_start_time_ns
; /* when passed to hardware */
132 /* Number of scatter-gather DMA addr+len pairs after
133 * physical address coalescing is performed.
135 unsigned short nr_phys_segments
;
136 #if defined(CONFIG_BLK_DEV_INTEGRITY)
137 unsigned short nr_integrity_segments
;
140 unsigned short ioprio
;
144 void *special
; /* opaque pointer available for LLD use */
145 char *buffer
; /* kaddr of the current segment if available */
151 * when request is used as a packet command carrier
153 unsigned char __cmd
[BLK_MAX_CDB
];
155 unsigned short cmd_len
;
157 unsigned int extra_len
; /* length of alignment and padding */
158 unsigned int sense_len
;
159 unsigned int resid_len
; /* residual count */
162 unsigned long deadline
;
163 struct list_head timeout_list
;
164 unsigned int timeout
;
168 * completion callback.
170 rq_end_io_fn
*end_io
;
174 struct request
*next_rq
;
177 static inline unsigned short req_get_ioprio(struct request
*req
)
183 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
184 * requests. Some step values could eventually be made generic.
186 struct request_pm_state
188 /* PM state machine step value, currently driver specific */
190 /* requested PM state value (S1, S2, S3, S4, ...) */
192 void* data
; /* for driver use */
195 #include <linux/elevator.h>
197 typedef void (request_fn_proc
) (struct request_queue
*q
);
198 typedef void (make_request_fn
) (struct request_queue
*q
, struct bio
*bio
);
199 typedef int (prep_rq_fn
) (struct request_queue
*, struct request
*);
200 typedef void (unprep_rq_fn
) (struct request_queue
*, struct request
*);
203 struct bvec_merge_data
{
204 struct block_device
*bi_bdev
;
209 typedef int (merge_bvec_fn
) (struct request_queue
*, struct bvec_merge_data
*,
211 typedef void (softirq_done_fn
)(struct request
*);
212 typedef int (dma_drain_needed_fn
)(struct request
*);
213 typedef int (lld_busy_fn
) (struct request_queue
*q
);
214 typedef int (bsg_job_fn
) (struct bsg_job
*);
216 enum blk_eh_timer_return
{
222 typedef enum blk_eh_timer_return (rq_timed_out_fn
)(struct request
*);
224 enum blk_queue_state
{
229 struct blk_queue_tag
{
230 struct request
**tag_index
; /* map of busy tags */
231 unsigned long *tag_map
; /* bit map of free/busy tags */
232 int busy
; /* current depth */
233 int max_depth
; /* what we will send to device */
234 int real_max_depth
; /* what the array can hold */
235 atomic_t refcnt
; /* map can be shared */
238 #define BLK_SCSI_MAX_CMDS (256)
239 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
241 struct queue_limits
{
242 unsigned long bounce_pfn
;
243 unsigned long seg_boundary_mask
;
245 unsigned int max_hw_sectors
;
246 unsigned int max_sectors
;
247 unsigned int max_segment_size
;
248 unsigned int physical_block_size
;
249 unsigned int alignment_offset
;
252 unsigned int max_discard_sectors
;
253 unsigned int discard_granularity
;
254 unsigned int discard_alignment
;
256 unsigned short logical_block_size
;
257 unsigned short max_segments
;
258 unsigned short max_integrity_segments
;
260 unsigned char misaligned
;
261 unsigned char discard_misaligned
;
262 unsigned char cluster
;
263 unsigned char discard_zeroes_data
;
266 struct request_queue
{
268 * Together with queue_head for cacheline sharing
270 struct list_head queue_head
;
271 struct request
*last_merge
;
272 struct elevator_queue
*elevator
;
275 * the queue request freelist, one for reads and one for writes
277 struct request_list rq
;
279 request_fn_proc
*request_fn
;
280 make_request_fn
*make_request_fn
;
281 prep_rq_fn
*prep_rq_fn
;
282 unprep_rq_fn
*unprep_rq_fn
;
283 merge_bvec_fn
*merge_bvec_fn
;
284 softirq_done_fn
*softirq_done_fn
;
285 rq_timed_out_fn
*rq_timed_out_fn
;
286 dma_drain_needed_fn
*dma_drain_needed
;
287 lld_busy_fn
*lld_busy_fn
;
290 * Dispatch queue sorting
293 struct request
*boundary_rq
;
296 * Delayed queue handling
298 struct delayed_work delay_work
;
300 struct backing_dev_info backing_dev_info
;
303 * The queue owner gets to use this for whatever they like.
304 * ll_rw_blk doesn't touch it.
309 * various queue flags, see QUEUE_* below
311 unsigned long queue_flags
;
314 * queue needs bounce pages for pages above this limit
319 * protects queue structures from reentrancy. ->__queue_lock should
320 * _never_ be used directly, it is queue private. always use
323 spinlock_t __queue_lock
;
324 spinlock_t
*queue_lock
;
334 unsigned long nr_requests
; /* Max # of requests */
335 unsigned int nr_congestion_on
;
336 unsigned int nr_congestion_off
;
337 unsigned int nr_batching
;
339 unsigned int dma_drain_size
;
340 void *dma_drain_buffer
;
341 unsigned int dma_pad_mask
;
342 unsigned int dma_alignment
;
344 struct blk_queue_tag
*queue_tags
;
345 struct list_head tag_busy_list
;
347 unsigned int nr_sorted
;
348 unsigned int in_flight
[2];
350 unsigned int rq_timeout
;
351 struct timer_list timeout
;
352 struct list_head timeout_list
;
354 struct queue_limits limits
;
359 unsigned int sg_timeout
;
360 unsigned int sg_reserved_size
;
362 #ifdef CONFIG_BLK_DEV_IO_TRACE
363 struct blk_trace
*blk_trace
;
366 * for flush operations
368 unsigned int flush_flags
;
369 unsigned int flush_not_queueable
:1;
370 unsigned int flush_queue_delayed
:1;
371 unsigned int flush_pending_idx
:1;
372 unsigned int flush_running_idx
:1;
373 unsigned long flush_pending_since
;
374 struct list_head flush_queue
[2];
375 struct list_head flush_data_in_flight
;
376 struct request flush_rq
;
378 struct mutex sysfs_lock
;
380 #if defined(CONFIG_BLK_DEV_BSG)
381 bsg_job_fn
*bsg_job_fn
;
383 struct bsg_class_device bsg_dev
;
386 #ifdef CONFIG_BLK_DEV_THROTTLING
388 struct throtl_data
*td
;
392 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
393 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
394 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
395 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
396 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
397 #define QUEUE_FLAG_ELVSWITCH 6 /* don't use elevator, just do FIFO */
398 #define QUEUE_FLAG_BIDI 7 /* queue supports bidi requests */
399 #define QUEUE_FLAG_NOMERGES 8 /* disable merge attempts */
400 #define QUEUE_FLAG_SAME_COMP 9 /* complete on same CPU-group */
401 #define QUEUE_FLAG_FAIL_IO 10 /* fake timeout */
402 #define QUEUE_FLAG_STACKABLE 11 /* supports request stacking */
403 #define QUEUE_FLAG_NONROT 12 /* non-rotational device (SSD) */
404 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
405 #define QUEUE_FLAG_IO_STAT 13 /* do IO stats */
406 #define QUEUE_FLAG_DISCARD 14 /* supports DISCARD */
407 #define QUEUE_FLAG_NOXMERGES 15 /* No extended merges */
408 #define QUEUE_FLAG_ADD_RANDOM 16 /* Contributes to random pool */
409 #define QUEUE_FLAG_SECDISCARD 17 /* supports SECDISCARD */
410 #define QUEUE_FLAG_SAME_FORCE 18 /* force complete on same CPU */
412 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
413 (1 << QUEUE_FLAG_STACKABLE) | \
414 (1 << QUEUE_FLAG_SAME_COMP) | \
415 (1 << QUEUE_FLAG_ADD_RANDOM))
417 static inline int queue_is_locked(struct request_queue
*q
)
420 spinlock_t
*lock
= q
->queue_lock
;
421 return lock
&& spin_is_locked(lock
);
427 static inline void queue_flag_set_unlocked(unsigned int flag
,
428 struct request_queue
*q
)
430 __set_bit(flag
, &q
->queue_flags
);
433 static inline int queue_flag_test_and_clear(unsigned int flag
,
434 struct request_queue
*q
)
436 WARN_ON_ONCE(!queue_is_locked(q
));
438 if (test_bit(flag
, &q
->queue_flags
)) {
439 __clear_bit(flag
, &q
->queue_flags
);
446 static inline int queue_flag_test_and_set(unsigned int flag
,
447 struct request_queue
*q
)
449 WARN_ON_ONCE(!queue_is_locked(q
));
451 if (!test_bit(flag
, &q
->queue_flags
)) {
452 __set_bit(flag
, &q
->queue_flags
);
459 static inline void queue_flag_set(unsigned int flag
, struct request_queue
*q
)
461 WARN_ON_ONCE(!queue_is_locked(q
));
462 __set_bit(flag
, &q
->queue_flags
);
465 static inline void queue_flag_clear_unlocked(unsigned int flag
,
466 struct request_queue
*q
)
468 __clear_bit(flag
, &q
->queue_flags
);
471 static inline int queue_in_flight(struct request_queue
*q
)
473 return q
->in_flight
[0] + q
->in_flight
[1];
476 static inline void queue_flag_clear(unsigned int flag
, struct request_queue
*q
)
478 WARN_ON_ONCE(!queue_is_locked(q
));
479 __clear_bit(flag
, &q
->queue_flags
);
482 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
483 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
484 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
485 #define blk_queue_noxmerges(q) \
486 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
487 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
488 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
489 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
490 #define blk_queue_stackable(q) \
491 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
492 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
493 #define blk_queue_secdiscard(q) (blk_queue_discard(q) && \
494 test_bit(QUEUE_FLAG_SECDISCARD, &(q)->queue_flags))
496 #define blk_noretry_request(rq) \
497 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
498 REQ_FAILFAST_DRIVER))
500 #define blk_account_rq(rq) \
501 (((rq)->cmd_flags & REQ_STARTED) && \
502 ((rq)->cmd_type == REQ_TYPE_FS || \
503 ((rq)->cmd_flags & REQ_DISCARD)))
505 #define blk_pm_request(rq) \
506 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
507 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
509 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
510 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
511 /* rq->queuelist of dequeued request must be list_empty() */
512 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
514 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
516 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
518 static inline unsigned int blk_queue_cluster(struct request_queue
*q
)
520 return q
->limits
.cluster
;
524 * We regard a request as sync, if either a read or a sync write
526 static inline bool rw_is_sync(unsigned int rw_flags
)
528 return !(rw_flags
& REQ_WRITE
) || (rw_flags
& REQ_SYNC
);
531 static inline bool rq_is_sync(struct request
*rq
)
533 return rw_is_sync(rq
->cmd_flags
);
536 static inline int blk_queue_full(struct request_queue
*q
, int sync
)
539 return test_bit(QUEUE_FLAG_SYNCFULL
, &q
->queue_flags
);
540 return test_bit(QUEUE_FLAG_ASYNCFULL
, &q
->queue_flags
);
543 static inline void blk_set_queue_full(struct request_queue
*q
, int sync
)
546 queue_flag_set(QUEUE_FLAG_SYNCFULL
, q
);
548 queue_flag_set(QUEUE_FLAG_ASYNCFULL
, q
);
551 static inline void blk_clear_queue_full(struct request_queue
*q
, int sync
)
554 queue_flag_clear(QUEUE_FLAG_SYNCFULL
, q
);
556 queue_flag_clear(QUEUE_FLAG_ASYNCFULL
, q
);
561 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
562 * it already be started by driver.
564 #define RQ_NOMERGE_FLAGS \
565 (REQ_NOMERGE | REQ_STARTED | REQ_SOFTBARRIER | REQ_FLUSH | REQ_FUA)
566 #define rq_mergeable(rq) \
567 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
568 (((rq)->cmd_flags & REQ_DISCARD) || \
569 (rq)->cmd_type == REQ_TYPE_FS))
572 * q->prep_rq_fn return values
574 #define BLKPREP_OK 0 /* serve it */
575 #define BLKPREP_KILL 1 /* fatal error, kill */
576 #define BLKPREP_DEFER 2 /* leave on queue */
578 extern unsigned long blk_max_low_pfn
, blk_max_pfn
;
581 * standard bounce addresses:
583 * BLK_BOUNCE_HIGH : bounce all highmem pages
584 * BLK_BOUNCE_ANY : don't bounce anything
585 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
588 #if BITS_PER_LONG == 32
589 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
591 #define BLK_BOUNCE_HIGH -1ULL
593 #define BLK_BOUNCE_ANY (-1ULL)
594 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
597 * default timeout for SG_IO if none specified
599 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
600 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
603 extern int init_emergency_isa_pool(void);
604 extern void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
);
606 static inline int init_emergency_isa_pool(void)
610 static inline void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
)
613 #endif /* CONFIG_MMU */
619 unsigned long offset
;
624 struct req_iterator
{
629 /* This should not be used directly - use rq_for_each_segment */
630 #define for_each_bio(_bio) \
631 for (; _bio; _bio = _bio->bi_next)
632 #define __rq_for_each_bio(_bio, rq) \
634 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
636 #define rq_for_each_segment(bvl, _rq, _iter) \
637 __rq_for_each_bio(_iter.bio, _rq) \
638 bio_for_each_segment(bvl, _iter.bio, _iter.i)
640 #define rq_iter_last(rq, _iter) \
641 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
643 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
644 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
646 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
647 extern void rq_flush_dcache_pages(struct request
*rq
);
649 static inline void rq_flush_dcache_pages(struct request
*rq
)
654 extern int blk_register_queue(struct gendisk
*disk
);
655 extern void blk_unregister_queue(struct gendisk
*disk
);
656 extern void generic_make_request(struct bio
*bio
);
657 extern void blk_rq_init(struct request_queue
*q
, struct request
*rq
);
658 extern void blk_put_request(struct request
*);
659 extern void __blk_put_request(struct request_queue
*, struct request
*);
660 extern struct request
*blk_get_request(struct request_queue
*, int, gfp_t
);
661 extern struct request
*blk_make_request(struct request_queue
*, struct bio
*,
663 extern void blk_insert_request(struct request_queue
*, struct request
*, int, void *);
664 extern void blk_requeue_request(struct request_queue
*, struct request
*);
665 extern void blk_add_request_payload(struct request
*rq
, struct page
*page
,
667 extern int blk_rq_check_limits(struct request_queue
*q
, struct request
*rq
);
668 extern int blk_lld_busy(struct request_queue
*q
);
669 extern int blk_rq_prep_clone(struct request
*rq
, struct request
*rq_src
,
670 struct bio_set
*bs
, gfp_t gfp_mask
,
671 int (*bio_ctr
)(struct bio
*, struct bio
*, void *),
673 extern void blk_rq_unprep_clone(struct request
*rq
);
674 extern int blk_insert_cloned_request(struct request_queue
*q
,
676 extern void blk_delay_queue(struct request_queue
*, unsigned long);
677 extern void blk_recount_segments(struct request_queue
*, struct bio
*);
678 extern int scsi_cmd_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
679 unsigned int, void __user
*);
680 extern int sg_scsi_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
681 struct scsi_ioctl_command __user
*);
683 extern void blk_queue_bio(struct request_queue
*q
, struct bio
*bio
);
686 * A queue has just exitted congestion. Note this in the global counter of
687 * congested queues, and wake up anyone who was waiting for requests to be
690 static inline void blk_clear_queue_congested(struct request_queue
*q
, int sync
)
692 clear_bdi_congested(&q
->backing_dev_info
, sync
);
696 * A queue has just entered congestion. Flag that in the queue's VM-visible
697 * state flags and increment the global gounter of congested queues.
699 static inline void blk_set_queue_congested(struct request_queue
*q
, int sync
)
701 set_bdi_congested(&q
->backing_dev_info
, sync
);
704 extern void blk_start_queue(struct request_queue
*q
);
705 extern void blk_stop_queue(struct request_queue
*q
);
706 extern void blk_sync_queue(struct request_queue
*q
);
707 extern void __blk_stop_queue(struct request_queue
*q
);
708 extern void __blk_run_queue(struct request_queue
*q
);
709 extern void blk_run_queue(struct request_queue
*);
710 extern void blk_run_queue_async(struct request_queue
*q
);
711 extern int blk_rq_map_user(struct request_queue
*, struct request
*,
712 struct rq_map_data
*, void __user
*, unsigned long,
714 extern int blk_rq_unmap_user(struct bio
*);
715 extern int blk_rq_map_kern(struct request_queue
*, struct request
*, void *, unsigned int, gfp_t
);
716 extern int blk_rq_map_user_iov(struct request_queue
*, struct request
*,
717 struct rq_map_data
*, struct sg_iovec
*, int,
718 unsigned int, gfp_t
);
719 extern int blk_execute_rq(struct request_queue
*, struct gendisk
*,
720 struct request
*, int);
721 extern void blk_execute_rq_nowait(struct request_queue
*, struct gendisk
*,
722 struct request
*, int, rq_end_io_fn
*);
724 static inline struct request_queue
*bdev_get_queue(struct block_device
*bdev
)
726 return bdev
->bd_disk
->queue
;
730 * blk_rq_pos() : the current sector
731 * blk_rq_bytes() : bytes left in the entire request
732 * blk_rq_cur_bytes() : bytes left in the current segment
733 * blk_rq_err_bytes() : bytes left till the next error boundary
734 * blk_rq_sectors() : sectors left in the entire request
735 * blk_rq_cur_sectors() : sectors left in the current segment
737 static inline sector_t
blk_rq_pos(const struct request
*rq
)
742 static inline unsigned int blk_rq_bytes(const struct request
*rq
)
744 return rq
->__data_len
;
747 static inline int blk_rq_cur_bytes(const struct request
*rq
)
749 return rq
->bio
? bio_cur_bytes(rq
->bio
) : 0;
752 extern unsigned int blk_rq_err_bytes(const struct request
*rq
);
754 static inline unsigned int blk_rq_sectors(const struct request
*rq
)
756 return blk_rq_bytes(rq
) >> 9;
759 static inline unsigned int blk_rq_cur_sectors(const struct request
*rq
)
761 return blk_rq_cur_bytes(rq
) >> 9;
765 * Request issue related functions.
767 extern struct request
*blk_peek_request(struct request_queue
*q
);
768 extern void blk_start_request(struct request
*rq
);
769 extern struct request
*blk_fetch_request(struct request_queue
*q
);
772 * Request completion related functions.
774 * blk_update_request() completes given number of bytes and updates
775 * the request without completing it.
777 * blk_end_request() and friends. __blk_end_request() must be called
778 * with the request queue spinlock acquired.
780 * Several drivers define their own end_request and call
781 * blk_end_request() for parts of the original function.
782 * This prevents code duplication in drivers.
784 extern bool blk_update_request(struct request
*rq
, int error
,
785 unsigned int nr_bytes
);
786 extern bool blk_end_request(struct request
*rq
, int error
,
787 unsigned int nr_bytes
);
788 extern void blk_end_request_all(struct request
*rq
, int error
);
789 extern bool blk_end_request_cur(struct request
*rq
, int error
);
790 extern bool blk_end_request_err(struct request
*rq
, int error
);
791 extern bool __blk_end_request(struct request
*rq
, int error
,
792 unsigned int nr_bytes
);
793 extern void __blk_end_request_all(struct request
*rq
, int error
);
794 extern bool __blk_end_request_cur(struct request
*rq
, int error
);
795 extern bool __blk_end_request_err(struct request
*rq
, int error
);
797 extern void blk_complete_request(struct request
*);
798 extern void __blk_complete_request(struct request
*);
799 extern void blk_abort_request(struct request
*);
800 extern void blk_abort_queue(struct request_queue
*);
801 extern void blk_unprep_request(struct request
*);
804 * Access functions for manipulating queue properties
806 extern struct request_queue
*blk_init_queue_node(request_fn_proc
*rfn
,
807 spinlock_t
*lock
, int node_id
);
808 extern struct request_queue
*blk_init_allocated_queue_node(struct request_queue
*,
810 spinlock_t
*, int node_id
);
811 extern struct request_queue
*blk_init_queue(request_fn_proc
*, spinlock_t
*);
812 extern struct request_queue
*blk_init_allocated_queue(struct request_queue
*,
813 request_fn_proc
*, spinlock_t
*);
814 extern void blk_cleanup_queue(struct request_queue
*);
815 extern void blk_queue_make_request(struct request_queue
*, make_request_fn
*);
816 extern void blk_queue_bounce_limit(struct request_queue
*, u64
);
817 extern void blk_limits_max_hw_sectors(struct queue_limits
*, unsigned int);
818 extern void blk_queue_max_hw_sectors(struct request_queue
*, unsigned int);
819 extern void blk_queue_max_segments(struct request_queue
*, unsigned short);
820 extern void blk_queue_max_segment_size(struct request_queue
*, unsigned int);
821 extern void blk_queue_max_discard_sectors(struct request_queue
*q
,
822 unsigned int max_discard_sectors
);
823 extern void blk_queue_logical_block_size(struct request_queue
*, unsigned short);
824 extern void blk_queue_physical_block_size(struct request_queue
*, unsigned int);
825 extern void blk_queue_alignment_offset(struct request_queue
*q
,
826 unsigned int alignment
);
827 extern void blk_limits_io_min(struct queue_limits
*limits
, unsigned int min
);
828 extern void blk_queue_io_min(struct request_queue
*q
, unsigned int min
);
829 extern void blk_limits_io_opt(struct queue_limits
*limits
, unsigned int opt
);
830 extern void blk_queue_io_opt(struct request_queue
*q
, unsigned int opt
);
831 extern void blk_set_default_limits(struct queue_limits
*lim
);
832 extern int blk_stack_limits(struct queue_limits
*t
, struct queue_limits
*b
,
834 extern int bdev_stack_limits(struct queue_limits
*t
, struct block_device
*bdev
,
836 extern void disk_stack_limits(struct gendisk
*disk
, struct block_device
*bdev
,
838 extern void blk_queue_stack_limits(struct request_queue
*t
, struct request_queue
*b
);
839 extern void blk_queue_dma_pad(struct request_queue
*, unsigned int);
840 extern void blk_queue_update_dma_pad(struct request_queue
*, unsigned int);
841 extern int blk_queue_dma_drain(struct request_queue
*q
,
842 dma_drain_needed_fn
*dma_drain_needed
,
843 void *buf
, unsigned int size
);
844 extern void blk_queue_lld_busy(struct request_queue
*q
, lld_busy_fn
*fn
);
845 extern void blk_queue_segment_boundary(struct request_queue
*, unsigned long);
846 extern void blk_queue_prep_rq(struct request_queue
*, prep_rq_fn
*pfn
);
847 extern void blk_queue_unprep_rq(struct request_queue
*, unprep_rq_fn
*ufn
);
848 extern void blk_queue_merge_bvec(struct request_queue
*, merge_bvec_fn
*);
849 extern void blk_queue_dma_alignment(struct request_queue
*, int);
850 extern void blk_queue_update_dma_alignment(struct request_queue
*, int);
851 extern void blk_queue_softirq_done(struct request_queue
*, softirq_done_fn
*);
852 extern void blk_queue_rq_timed_out(struct request_queue
*, rq_timed_out_fn
*);
853 extern void blk_queue_rq_timeout(struct request_queue
*, unsigned int);
854 extern void blk_queue_flush(struct request_queue
*q
, unsigned int flush
);
855 extern void blk_queue_flush_queueable(struct request_queue
*q
, bool queueable
);
856 extern struct backing_dev_info
*blk_get_backing_dev_info(struct block_device
*bdev
);
858 extern int blk_rq_map_sg(struct request_queue
*, struct request
*, struct scatterlist
*);
859 extern void blk_dump_rq_flags(struct request
*, char *);
860 extern long nr_blockdev_pages(void);
862 int blk_get_queue(struct request_queue
*);
863 struct request_queue
*blk_alloc_queue(gfp_t
);
864 struct request_queue
*blk_alloc_queue_node(gfp_t
, int);
865 extern void blk_put_queue(struct request_queue
*);
868 * blk_plug permits building a queue of related requests by holding the I/O
869 * fragments for a short period. This allows merging of sequential requests
870 * into single larger request. As the requests are moved from a per-task list to
871 * the device's request_queue in a batch, this results in improved scalability
872 * as the lock contention for request_queue lock is reduced.
874 * It is ok not to disable preemption when adding the request to the plug list
875 * or when attempting a merge, because blk_schedule_flush_list() will only flush
876 * the plug list when the task sleeps by itself. For details, please see
877 * schedule() where blk_schedule_flush_plug() is called.
880 unsigned long magic
; /* detect uninitialized use-cases */
881 struct list_head list
; /* requests */
882 struct list_head cb_list
; /* md requires an unplug callback */
883 unsigned int should_sort
; /* list to be sorted before flushing? */
885 #define BLK_MAX_REQUEST_COUNT 16
888 struct list_head list
;
889 void (*callback
)(struct blk_plug_cb
*);
892 extern void blk_start_plug(struct blk_plug
*);
893 extern void blk_finish_plug(struct blk_plug
*);
894 extern void blk_flush_plug_list(struct blk_plug
*, bool);
896 static inline void blk_flush_plug(struct task_struct
*tsk
)
898 struct blk_plug
*plug
= tsk
->plug
;
901 blk_flush_plug_list(plug
, false);
904 static inline void blk_schedule_flush_plug(struct task_struct
*tsk
)
906 struct blk_plug
*plug
= tsk
->plug
;
909 blk_flush_plug_list(plug
, true);
912 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
914 struct blk_plug
*plug
= tsk
->plug
;
916 return plug
&& (!list_empty(&plug
->list
) || !list_empty(&plug
->cb_list
));
922 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
923 extern int blk_queue_start_tag(struct request_queue
*, struct request
*);
924 extern struct request
*blk_queue_find_tag(struct request_queue
*, int);
925 extern void blk_queue_end_tag(struct request_queue
*, struct request
*);
926 extern int blk_queue_init_tags(struct request_queue
*, int, struct blk_queue_tag
*);
927 extern void blk_queue_free_tags(struct request_queue
*);
928 extern int blk_queue_resize_tags(struct request_queue
*, int);
929 extern void blk_queue_invalidate_tags(struct request_queue
*);
930 extern struct blk_queue_tag
*blk_init_tags(int);
931 extern void blk_free_tags(struct blk_queue_tag
*);
933 static inline struct request
*blk_map_queue_find_tag(struct blk_queue_tag
*bqt
,
936 if (unlikely(bqt
== NULL
|| tag
>= bqt
->real_max_depth
))
938 return bqt
->tag_index
[tag
];
941 #define BLKDEV_DISCARD_SECURE 0x01 /* secure discard */
943 extern int blkdev_issue_flush(struct block_device
*, gfp_t
, sector_t
*);
944 extern int blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
945 sector_t nr_sects
, gfp_t gfp_mask
, unsigned long flags
);
946 extern int blkdev_issue_zeroout(struct block_device
*bdev
, sector_t sector
,
947 sector_t nr_sects
, gfp_t gfp_mask
);
948 static inline int sb_issue_discard(struct super_block
*sb
, sector_t block
,
949 sector_t nr_blocks
, gfp_t gfp_mask
, unsigned long flags
)
951 return blkdev_issue_discard(sb
->s_bdev
, block
<< (sb
->s_blocksize_bits
- 9),
952 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
955 static inline int sb_issue_zeroout(struct super_block
*sb
, sector_t block
,
956 sector_t nr_blocks
, gfp_t gfp_mask
)
958 return blkdev_issue_zeroout(sb
->s_bdev
,
959 block
<< (sb
->s_blocksize_bits
- 9),
960 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
964 extern int blk_verify_command(unsigned char *cmd
, fmode_t has_write_perm
);
966 enum blk_default_limits
{
967 BLK_MAX_SEGMENTS
= 128,
968 BLK_SAFE_MAX_SECTORS
= 255,
969 BLK_DEF_MAX_SECTORS
= 1024,
970 BLK_MAX_SEGMENT_SIZE
= 65536,
971 BLK_SEG_BOUNDARY_MASK
= 0xFFFFFFFFUL
,
974 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
976 static inline unsigned long queue_bounce_pfn(struct request_queue
*q
)
978 return q
->limits
.bounce_pfn
;
981 static inline unsigned long queue_segment_boundary(struct request_queue
*q
)
983 return q
->limits
.seg_boundary_mask
;
986 static inline unsigned int queue_max_sectors(struct request_queue
*q
)
988 return q
->limits
.max_sectors
;
991 static inline unsigned int queue_max_hw_sectors(struct request_queue
*q
)
993 return q
->limits
.max_hw_sectors
;
996 static inline unsigned short queue_max_segments(struct request_queue
*q
)
998 return q
->limits
.max_segments
;
1001 static inline unsigned int queue_max_segment_size(struct request_queue
*q
)
1003 return q
->limits
.max_segment_size
;
1006 static inline unsigned short queue_logical_block_size(struct request_queue
*q
)
1010 if (q
&& q
->limits
.logical_block_size
)
1011 retval
= q
->limits
.logical_block_size
;
1016 static inline unsigned short bdev_logical_block_size(struct block_device
*bdev
)
1018 return queue_logical_block_size(bdev_get_queue(bdev
));
1021 static inline unsigned int queue_physical_block_size(struct request_queue
*q
)
1023 return q
->limits
.physical_block_size
;
1026 static inline unsigned int bdev_physical_block_size(struct block_device
*bdev
)
1028 return queue_physical_block_size(bdev_get_queue(bdev
));
1031 static inline unsigned int queue_io_min(struct request_queue
*q
)
1033 return q
->limits
.io_min
;
1036 static inline int bdev_io_min(struct block_device
*bdev
)
1038 return queue_io_min(bdev_get_queue(bdev
));
1041 static inline unsigned int queue_io_opt(struct request_queue
*q
)
1043 return q
->limits
.io_opt
;
1046 static inline int bdev_io_opt(struct block_device
*bdev
)
1048 return queue_io_opt(bdev_get_queue(bdev
));
1051 static inline int queue_alignment_offset(struct request_queue
*q
)
1053 if (q
->limits
.misaligned
)
1056 return q
->limits
.alignment_offset
;
1059 static inline int queue_limit_alignment_offset(struct queue_limits
*lim
, sector_t sector
)
1061 unsigned int granularity
= max(lim
->physical_block_size
, lim
->io_min
);
1062 unsigned int alignment
= (sector
<< 9) & (granularity
- 1);
1064 return (granularity
+ lim
->alignment_offset
- alignment
)
1065 & (granularity
- 1);
1068 static inline int bdev_alignment_offset(struct block_device
*bdev
)
1070 struct request_queue
*q
= bdev_get_queue(bdev
);
1072 if (q
->limits
.misaligned
)
1075 if (bdev
!= bdev
->bd_contains
)
1076 return bdev
->bd_part
->alignment_offset
;
1078 return q
->limits
.alignment_offset
;
1081 static inline int queue_discard_alignment(struct request_queue
*q
)
1083 if (q
->limits
.discard_misaligned
)
1086 return q
->limits
.discard_alignment
;
1089 static inline int queue_limit_discard_alignment(struct queue_limits
*lim
, sector_t sector
)
1091 unsigned int alignment
= (sector
<< 9) & (lim
->discard_granularity
- 1);
1093 if (!lim
->max_discard_sectors
)
1096 return (lim
->discard_granularity
+ lim
->discard_alignment
- alignment
)
1097 & (lim
->discard_granularity
- 1);
1100 static inline unsigned int queue_discard_zeroes_data(struct request_queue
*q
)
1102 if (q
->limits
.max_discard_sectors
&& q
->limits
.discard_zeroes_data
== 1)
1108 static inline unsigned int bdev_discard_zeroes_data(struct block_device
*bdev
)
1110 return queue_discard_zeroes_data(bdev_get_queue(bdev
));
1113 static inline int queue_dma_alignment(struct request_queue
*q
)
1115 return q
? q
->dma_alignment
: 511;
1118 static inline int blk_rq_aligned(struct request_queue
*q
, unsigned long addr
,
1121 unsigned int alignment
= queue_dma_alignment(q
) | q
->dma_pad_mask
;
1122 return !(addr
& alignment
) && !(len
& alignment
);
1125 /* assumes size > 256 */
1126 static inline unsigned int blksize_bits(unsigned int size
)
1128 unsigned int bits
= 8;
1132 } while (size
> 256);
1136 static inline unsigned int block_size(struct block_device
*bdev
)
1138 return bdev
->bd_block_size
;
1141 static inline bool queue_flush_queueable(struct request_queue
*q
)
1143 return !q
->flush_not_queueable
;
1146 typedef struct {struct page
*v
;} Sector
;
1148 unsigned char *read_dev_sector(struct block_device
*, sector_t
, Sector
*);
1150 static inline void put_dev_sector(Sector p
)
1152 page_cache_release(p
.v
);
1156 int kblockd_schedule_work(struct request_queue
*q
, struct work_struct
*work
);
1158 #ifdef CONFIG_BLK_CGROUP
1160 * This should not be using sched_clock(). A real patch is in progress
1161 * to fix this up, until that is in place we need to disable preemption
1162 * around sched_clock() in this function and set_io_start_time_ns().
1164 static inline void set_start_time_ns(struct request
*req
)
1167 req
->start_time_ns
= sched_clock();
1171 static inline void set_io_start_time_ns(struct request
*req
)
1174 req
->io_start_time_ns
= sched_clock();
1178 static inline uint64_t rq_start_time_ns(struct request
*req
)
1180 return req
->start_time_ns
;
1183 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1185 return req
->io_start_time_ns
;
1188 static inline void set_start_time_ns(struct request
*req
) {}
1189 static inline void set_io_start_time_ns(struct request
*req
) {}
1190 static inline uint64_t rq_start_time_ns(struct request
*req
)
1194 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1200 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1201 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1202 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1203 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1205 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1207 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1208 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1210 struct blk_integrity_exchg
{
1214 unsigned int data_size
;
1215 unsigned short sector_size
;
1216 const char *disk_name
;
1219 typedef void (integrity_gen_fn
) (struct blk_integrity_exchg
*);
1220 typedef int (integrity_vrfy_fn
) (struct blk_integrity_exchg
*);
1221 typedef void (integrity_set_tag_fn
) (void *, void *, unsigned int);
1222 typedef void (integrity_get_tag_fn
) (void *, void *, unsigned int);
1224 struct blk_integrity
{
1225 integrity_gen_fn
*generate_fn
;
1226 integrity_vrfy_fn
*verify_fn
;
1227 integrity_set_tag_fn
*set_tag_fn
;
1228 integrity_get_tag_fn
*get_tag_fn
;
1230 unsigned short flags
;
1231 unsigned short tuple_size
;
1232 unsigned short sector_size
;
1233 unsigned short tag_size
;
1237 struct kobject kobj
;
1240 extern bool blk_integrity_is_initialized(struct gendisk
*);
1241 extern int blk_integrity_register(struct gendisk
*, struct blk_integrity
*);
1242 extern void blk_integrity_unregister(struct gendisk
*);
1243 extern int blk_integrity_compare(struct gendisk
*, struct gendisk
*);
1244 extern int blk_rq_map_integrity_sg(struct request_queue
*, struct bio
*,
1245 struct scatterlist
*);
1246 extern int blk_rq_count_integrity_sg(struct request_queue
*, struct bio
*);
1247 extern int blk_integrity_merge_rq(struct request_queue
*, struct request
*,
1249 extern int blk_integrity_merge_bio(struct request_queue
*, struct request
*,
1253 struct blk_integrity
*bdev_get_integrity(struct block_device
*bdev
)
1255 return bdev
->bd_disk
->integrity
;
1258 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1260 return disk
->integrity
;
1263 static inline int blk_integrity_rq(struct request
*rq
)
1265 if (rq
->bio
== NULL
)
1268 return bio_integrity(rq
->bio
);
1271 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1274 q
->limits
.max_integrity_segments
= segs
;
1277 static inline unsigned short
1278 queue_max_integrity_segments(struct request_queue
*q
)
1280 return q
->limits
.max_integrity_segments
;
1283 #else /* CONFIG_BLK_DEV_INTEGRITY */
1285 #define blk_integrity_rq(rq) (0)
1286 #define blk_rq_count_integrity_sg(a, b) (0)
1287 #define blk_rq_map_integrity_sg(a, b, c) (0)
1288 #define bdev_get_integrity(a) (0)
1289 #define blk_get_integrity(a) (0)
1290 #define blk_integrity_compare(a, b) (0)
1291 #define blk_integrity_register(a, b) (0)
1292 #define blk_integrity_unregister(a) do { } while (0)
1293 #define blk_queue_max_integrity_segments(a, b) do { } while (0)
1294 #define queue_max_integrity_segments(a) (0)
1295 #define blk_integrity_merge_rq(a, b, c) (0)
1296 #define blk_integrity_merge_bio(a, b, c) (0)
1297 #define blk_integrity_is_initialized(a) (0)
1299 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1301 struct block_device_operations
{
1302 int (*open
) (struct block_device
*, fmode_t
);
1303 int (*release
) (struct gendisk
*, fmode_t
);
1304 int (*ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1305 int (*compat_ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1306 int (*direct_access
) (struct block_device
*, sector_t
,
1307 void **, unsigned long *);
1308 unsigned int (*check_events
) (struct gendisk
*disk
,
1309 unsigned int clearing
);
1310 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1311 int (*media_changed
) (struct gendisk
*);
1312 void (*unlock_native_capacity
) (struct gendisk
*);
1313 int (*revalidate_disk
) (struct gendisk
*);
1314 int (*getgeo
)(struct block_device
*, struct hd_geometry
*);
1315 /* this callback is with swap_lock and sometimes page table lock held */
1316 void (*swap_slot_free_notify
) (struct block_device
*, unsigned long);
1317 struct module
*owner
;
1320 extern int __blkdev_driver_ioctl(struct block_device
*, fmode_t
, unsigned int,
1322 #else /* CONFIG_BLOCK */
1324 * stubs for when the block layer is configured out
1326 #define buffer_heads_over_limit 0
1328 static inline long nr_blockdev_pages(void)
1336 static inline void blk_start_plug(struct blk_plug
*plug
)
1340 static inline void blk_finish_plug(struct blk_plug
*plug
)
1344 static inline void blk_flush_plug(struct task_struct
*task
)
1348 static inline void blk_schedule_flush_plug(struct task_struct
*task
)
1353 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
1358 #endif /* CONFIG_BLOCK */