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 elevator data.
124 struct list_head list
;
125 rq_end_io_fn
*saved_end_io
;
129 struct gendisk
*rq_disk
;
130 struct hd_struct
*part
;
131 unsigned long start_time
;
132 #ifdef CONFIG_BLK_CGROUP
133 unsigned long long start_time_ns
;
134 unsigned long long io_start_time_ns
; /* when passed to hardware */
136 /* Number of scatter-gather DMA addr+len pairs after
137 * physical address coalescing is performed.
139 unsigned short nr_phys_segments
;
140 #if defined(CONFIG_BLK_DEV_INTEGRITY)
141 unsigned short nr_integrity_segments
;
144 unsigned short ioprio
;
148 void *special
; /* opaque pointer available for LLD use */
149 char *buffer
; /* kaddr of the current segment if available */
155 * when request is used as a packet command carrier
157 unsigned char __cmd
[BLK_MAX_CDB
];
159 unsigned short cmd_len
;
161 unsigned int extra_len
; /* length of alignment and padding */
162 unsigned int sense_len
;
163 unsigned int resid_len
; /* residual count */
166 unsigned long deadline
;
167 struct list_head timeout_list
;
168 unsigned int timeout
;
172 * completion callback.
174 rq_end_io_fn
*end_io
;
178 struct request
*next_rq
;
181 static inline unsigned short req_get_ioprio(struct request
*req
)
187 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
188 * requests. Some step values could eventually be made generic.
190 struct request_pm_state
192 /* PM state machine step value, currently driver specific */
194 /* requested PM state value (S1, S2, S3, S4, ...) */
196 void* data
; /* for driver use */
199 #include <linux/elevator.h>
201 typedef void (request_fn_proc
) (struct request_queue
*q
);
202 typedef void (make_request_fn
) (struct request_queue
*q
, struct bio
*bio
);
203 typedef int (prep_rq_fn
) (struct request_queue
*, struct request
*);
204 typedef void (unprep_rq_fn
) (struct request_queue
*, struct request
*);
207 struct bvec_merge_data
{
208 struct block_device
*bi_bdev
;
213 typedef int (merge_bvec_fn
) (struct request_queue
*, struct bvec_merge_data
*,
215 typedef void (softirq_done_fn
)(struct request
*);
216 typedef int (dma_drain_needed_fn
)(struct request
*);
217 typedef int (lld_busy_fn
) (struct request_queue
*q
);
218 typedef int (bsg_job_fn
) (struct bsg_job
*);
220 enum blk_eh_timer_return
{
226 typedef enum blk_eh_timer_return (rq_timed_out_fn
)(struct request
*);
228 enum blk_queue_state
{
233 struct blk_queue_tag
{
234 struct request
**tag_index
; /* map of busy tags */
235 unsigned long *tag_map
; /* bit map of free/busy tags */
236 int busy
; /* current depth */
237 int max_depth
; /* what we will send to device */
238 int real_max_depth
; /* what the array can hold */
239 atomic_t refcnt
; /* map can be shared */
242 #define BLK_SCSI_MAX_CMDS (256)
243 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
245 struct queue_limits
{
246 unsigned long bounce_pfn
;
247 unsigned long seg_boundary_mask
;
249 unsigned int max_hw_sectors
;
250 unsigned int max_sectors
;
251 unsigned int max_segment_size
;
252 unsigned int physical_block_size
;
253 unsigned int alignment_offset
;
256 unsigned int max_discard_sectors
;
257 unsigned int discard_granularity
;
258 unsigned int discard_alignment
;
260 unsigned short logical_block_size
;
261 unsigned short max_segments
;
262 unsigned short max_integrity_segments
;
264 unsigned char misaligned
;
265 unsigned char discard_misaligned
;
266 unsigned char cluster
;
267 unsigned char discard_zeroes_data
;
270 struct request_queue
{
272 * Together with queue_head for cacheline sharing
274 struct list_head queue_head
;
275 struct request
*last_merge
;
276 struct elevator_queue
*elevator
;
279 * the queue request freelist, one for reads and one for writes
281 struct request_list rq
;
283 request_fn_proc
*request_fn
;
284 make_request_fn
*make_request_fn
;
285 prep_rq_fn
*prep_rq_fn
;
286 unprep_rq_fn
*unprep_rq_fn
;
287 merge_bvec_fn
*merge_bvec_fn
;
288 softirq_done_fn
*softirq_done_fn
;
289 rq_timed_out_fn
*rq_timed_out_fn
;
290 dma_drain_needed_fn
*dma_drain_needed
;
291 lld_busy_fn
*lld_busy_fn
;
294 * Dispatch queue sorting
297 struct request
*boundary_rq
;
300 * Delayed queue handling
302 struct delayed_work delay_work
;
304 struct backing_dev_info backing_dev_info
;
307 * The queue owner gets to use this for whatever they like.
308 * ll_rw_blk doesn't touch it.
313 * various queue flags, see QUEUE_* below
315 unsigned long queue_flags
;
318 * ida allocated id for this queue. Used to index queues from
324 * queue needs bounce pages for pages above this limit
329 * protects queue structures from reentrancy. ->__queue_lock should
330 * _never_ be used directly, it is queue private. always use
333 spinlock_t __queue_lock
;
334 spinlock_t
*queue_lock
;
344 unsigned long nr_requests
; /* Max # of requests */
345 unsigned int nr_congestion_on
;
346 unsigned int nr_congestion_off
;
347 unsigned int nr_batching
;
349 unsigned int dma_drain_size
;
350 void *dma_drain_buffer
;
351 unsigned int dma_pad_mask
;
352 unsigned int dma_alignment
;
354 struct blk_queue_tag
*queue_tags
;
355 struct list_head tag_busy_list
;
357 unsigned int nr_sorted
;
358 unsigned int in_flight
[2];
360 unsigned int rq_timeout
;
361 struct timer_list timeout
;
362 struct list_head timeout_list
;
364 struct list_head icq_list
;
365 #ifdef CONFIG_BLK_CGROUP
366 /* XXX: array size hardcoded to avoid include dependency (temporary) */
367 struct list_head blkg_list
[2];
371 struct queue_limits limits
;
376 unsigned int sg_timeout
;
377 unsigned int sg_reserved_size
;
379 #ifdef CONFIG_BLK_DEV_IO_TRACE
380 struct blk_trace
*blk_trace
;
383 * for flush operations
385 unsigned int flush_flags
;
386 unsigned int flush_not_queueable
:1;
387 unsigned int flush_queue_delayed
:1;
388 unsigned int flush_pending_idx
:1;
389 unsigned int flush_running_idx
:1;
390 unsigned long flush_pending_since
;
391 struct list_head flush_queue
[2];
392 struct list_head flush_data_in_flight
;
393 struct request flush_rq
;
395 struct mutex sysfs_lock
;
399 #if defined(CONFIG_BLK_DEV_BSG)
400 bsg_job_fn
*bsg_job_fn
;
402 struct bsg_class_device bsg_dev
;
405 #ifdef CONFIG_BLK_CGROUP
406 struct list_head all_q_node
;
408 #ifdef CONFIG_BLK_DEV_THROTTLING
410 struct throtl_data
*td
;
414 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
415 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
416 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
417 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
418 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
419 #define QUEUE_FLAG_BYPASS 6 /* act as dumb FIFO queue */
420 #define QUEUE_FLAG_BIDI 7 /* queue supports bidi requests */
421 #define QUEUE_FLAG_NOMERGES 8 /* disable merge attempts */
422 #define QUEUE_FLAG_SAME_COMP 9 /* complete on same CPU-group */
423 #define QUEUE_FLAG_FAIL_IO 10 /* fake timeout */
424 #define QUEUE_FLAG_STACKABLE 11 /* supports request stacking */
425 #define QUEUE_FLAG_NONROT 12 /* non-rotational device (SSD) */
426 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
427 #define QUEUE_FLAG_IO_STAT 13 /* do IO stats */
428 #define QUEUE_FLAG_DISCARD 14 /* supports DISCARD */
429 #define QUEUE_FLAG_NOXMERGES 15 /* No extended merges */
430 #define QUEUE_FLAG_ADD_RANDOM 16 /* Contributes to random pool */
431 #define QUEUE_FLAG_SECDISCARD 17 /* supports SECDISCARD */
432 #define QUEUE_FLAG_SAME_FORCE 18 /* force complete on same CPU */
434 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
435 (1 << QUEUE_FLAG_STACKABLE) | \
436 (1 << QUEUE_FLAG_SAME_COMP) | \
437 (1 << QUEUE_FLAG_ADD_RANDOM))
439 static inline int queue_is_locked(struct request_queue
*q
)
442 spinlock_t
*lock
= q
->queue_lock
;
443 return lock
&& spin_is_locked(lock
);
449 static inline void queue_flag_set_unlocked(unsigned int flag
,
450 struct request_queue
*q
)
452 __set_bit(flag
, &q
->queue_flags
);
455 static inline int queue_flag_test_and_clear(unsigned int flag
,
456 struct request_queue
*q
)
458 WARN_ON_ONCE(!queue_is_locked(q
));
460 if (test_bit(flag
, &q
->queue_flags
)) {
461 __clear_bit(flag
, &q
->queue_flags
);
468 static inline int queue_flag_test_and_set(unsigned int flag
,
469 struct request_queue
*q
)
471 WARN_ON_ONCE(!queue_is_locked(q
));
473 if (!test_bit(flag
, &q
->queue_flags
)) {
474 __set_bit(flag
, &q
->queue_flags
);
481 static inline void queue_flag_set(unsigned int flag
, struct request_queue
*q
)
483 WARN_ON_ONCE(!queue_is_locked(q
));
484 __set_bit(flag
, &q
->queue_flags
);
487 static inline void queue_flag_clear_unlocked(unsigned int flag
,
488 struct request_queue
*q
)
490 __clear_bit(flag
, &q
->queue_flags
);
493 static inline int queue_in_flight(struct request_queue
*q
)
495 return q
->in_flight
[0] + q
->in_flight
[1];
498 static inline void queue_flag_clear(unsigned int flag
, struct request_queue
*q
)
500 WARN_ON_ONCE(!queue_is_locked(q
));
501 __clear_bit(flag
, &q
->queue_flags
);
504 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
505 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
506 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
507 #define blk_queue_bypass(q) test_bit(QUEUE_FLAG_BYPASS, &(q)->queue_flags)
508 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
509 #define blk_queue_noxmerges(q) \
510 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
511 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
512 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
513 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
514 #define blk_queue_stackable(q) \
515 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
516 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
517 #define blk_queue_secdiscard(q) (blk_queue_discard(q) && \
518 test_bit(QUEUE_FLAG_SECDISCARD, &(q)->queue_flags))
520 #define blk_noretry_request(rq) \
521 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
522 REQ_FAILFAST_DRIVER))
524 #define blk_account_rq(rq) \
525 (((rq)->cmd_flags & REQ_STARTED) && \
526 ((rq)->cmd_type == REQ_TYPE_FS || \
527 ((rq)->cmd_flags & REQ_DISCARD)))
529 #define blk_pm_request(rq) \
530 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
531 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
533 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
534 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
535 /* rq->queuelist of dequeued request must be list_empty() */
536 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
538 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
540 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
542 static inline unsigned int blk_queue_cluster(struct request_queue
*q
)
544 return q
->limits
.cluster
;
548 * We regard a request as sync, if either a read or a sync write
550 static inline bool rw_is_sync(unsigned int rw_flags
)
552 return !(rw_flags
& REQ_WRITE
) || (rw_flags
& REQ_SYNC
);
555 static inline bool rq_is_sync(struct request
*rq
)
557 return rw_is_sync(rq
->cmd_flags
);
560 static inline int blk_queue_full(struct request_queue
*q
, int sync
)
563 return test_bit(QUEUE_FLAG_SYNCFULL
, &q
->queue_flags
);
564 return test_bit(QUEUE_FLAG_ASYNCFULL
, &q
->queue_flags
);
567 static inline void blk_set_queue_full(struct request_queue
*q
, int sync
)
570 queue_flag_set(QUEUE_FLAG_SYNCFULL
, q
);
572 queue_flag_set(QUEUE_FLAG_ASYNCFULL
, q
);
575 static inline void blk_clear_queue_full(struct request_queue
*q
, int sync
)
578 queue_flag_clear(QUEUE_FLAG_SYNCFULL
, q
);
580 queue_flag_clear(QUEUE_FLAG_ASYNCFULL
, q
);
585 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
586 * it already be started by driver.
588 #define RQ_NOMERGE_FLAGS \
589 (REQ_NOMERGE | REQ_STARTED | REQ_SOFTBARRIER | REQ_FLUSH | REQ_FUA)
590 #define rq_mergeable(rq) \
591 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
592 (((rq)->cmd_flags & REQ_DISCARD) || \
593 (rq)->cmd_type == REQ_TYPE_FS))
596 * q->prep_rq_fn return values
598 #define BLKPREP_OK 0 /* serve it */
599 #define BLKPREP_KILL 1 /* fatal error, kill */
600 #define BLKPREP_DEFER 2 /* leave on queue */
602 extern unsigned long blk_max_low_pfn
, blk_max_pfn
;
605 * standard bounce addresses:
607 * BLK_BOUNCE_HIGH : bounce all highmem pages
608 * BLK_BOUNCE_ANY : don't bounce anything
609 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
612 #if BITS_PER_LONG == 32
613 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
615 #define BLK_BOUNCE_HIGH -1ULL
617 #define BLK_BOUNCE_ANY (-1ULL)
618 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
621 * default timeout for SG_IO if none specified
623 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
624 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
627 extern int init_emergency_isa_pool(void);
628 extern void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
);
630 static inline int init_emergency_isa_pool(void)
634 static inline void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
)
637 #endif /* CONFIG_MMU */
643 unsigned long offset
;
648 struct req_iterator
{
653 /* This should not be used directly - use rq_for_each_segment */
654 #define for_each_bio(_bio) \
655 for (; _bio; _bio = _bio->bi_next)
656 #define __rq_for_each_bio(_bio, rq) \
658 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
660 #define rq_for_each_segment(bvl, _rq, _iter) \
661 __rq_for_each_bio(_iter.bio, _rq) \
662 bio_for_each_segment(bvl, _iter.bio, _iter.i)
664 #define rq_iter_last(rq, _iter) \
665 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
667 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
668 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
670 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
671 extern void rq_flush_dcache_pages(struct request
*rq
);
673 static inline void rq_flush_dcache_pages(struct request
*rq
)
678 extern int blk_register_queue(struct gendisk
*disk
);
679 extern void blk_unregister_queue(struct gendisk
*disk
);
680 extern void generic_make_request(struct bio
*bio
);
681 extern void blk_rq_init(struct request_queue
*q
, struct request
*rq
);
682 extern void blk_put_request(struct request
*);
683 extern void __blk_put_request(struct request_queue
*, struct request
*);
684 extern struct request
*blk_get_request(struct request_queue
*, int, gfp_t
);
685 extern struct request
*blk_make_request(struct request_queue
*, struct bio
*,
687 extern void blk_requeue_request(struct request_queue
*, struct request
*);
688 extern void blk_add_request_payload(struct request
*rq
, struct page
*page
,
690 extern int blk_rq_check_limits(struct request_queue
*q
, struct request
*rq
);
691 extern int blk_lld_busy(struct request_queue
*q
);
692 extern int blk_rq_prep_clone(struct request
*rq
, struct request
*rq_src
,
693 struct bio_set
*bs
, gfp_t gfp_mask
,
694 int (*bio_ctr
)(struct bio
*, struct bio
*, void *),
696 extern void blk_rq_unprep_clone(struct request
*rq
);
697 extern int blk_insert_cloned_request(struct request_queue
*q
,
699 extern void blk_delay_queue(struct request_queue
*, unsigned long);
700 extern void blk_recount_segments(struct request_queue
*, struct bio
*);
701 extern int scsi_verify_blk_ioctl(struct block_device
*, unsigned int);
702 extern int scsi_cmd_blk_ioctl(struct block_device
*, fmode_t
,
703 unsigned int, void __user
*);
704 extern int scsi_cmd_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
705 unsigned int, void __user
*);
706 extern int sg_scsi_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
707 struct scsi_ioctl_command __user
*);
709 extern void blk_queue_bio(struct request_queue
*q
, struct bio
*bio
);
712 * A queue has just exitted congestion. Note this in the global counter of
713 * congested queues, and wake up anyone who was waiting for requests to be
716 static inline void blk_clear_queue_congested(struct request_queue
*q
, int sync
)
718 clear_bdi_congested(&q
->backing_dev_info
, sync
);
722 * A queue has just entered congestion. Flag that in the queue's VM-visible
723 * state flags and increment the global gounter of congested queues.
725 static inline void blk_set_queue_congested(struct request_queue
*q
, int sync
)
727 set_bdi_congested(&q
->backing_dev_info
, sync
);
730 extern void blk_start_queue(struct request_queue
*q
);
731 extern void blk_stop_queue(struct request_queue
*q
);
732 extern void blk_sync_queue(struct request_queue
*q
);
733 extern void __blk_stop_queue(struct request_queue
*q
);
734 extern void __blk_run_queue(struct request_queue
*q
);
735 extern void blk_run_queue(struct request_queue
*);
736 extern void blk_run_queue_async(struct request_queue
*q
);
737 extern int blk_rq_map_user(struct request_queue
*, struct request
*,
738 struct rq_map_data
*, void __user
*, unsigned long,
740 extern int blk_rq_unmap_user(struct bio
*);
741 extern int blk_rq_map_kern(struct request_queue
*, struct request
*, void *, unsigned int, gfp_t
);
742 extern int blk_rq_map_user_iov(struct request_queue
*, struct request
*,
743 struct rq_map_data
*, struct sg_iovec
*, int,
744 unsigned int, gfp_t
);
745 extern int blk_execute_rq(struct request_queue
*, struct gendisk
*,
746 struct request
*, int);
747 extern void blk_execute_rq_nowait(struct request_queue
*, struct gendisk
*,
748 struct request
*, int, rq_end_io_fn
*);
750 static inline struct request_queue
*bdev_get_queue(struct block_device
*bdev
)
752 return bdev
->bd_disk
->queue
;
756 * blk_rq_pos() : the current sector
757 * blk_rq_bytes() : bytes left in the entire request
758 * blk_rq_cur_bytes() : bytes left in the current segment
759 * blk_rq_err_bytes() : bytes left till the next error boundary
760 * blk_rq_sectors() : sectors left in the entire request
761 * blk_rq_cur_sectors() : sectors left in the current segment
763 static inline sector_t
blk_rq_pos(const struct request
*rq
)
768 static inline unsigned int blk_rq_bytes(const struct request
*rq
)
770 return rq
->__data_len
;
773 static inline int blk_rq_cur_bytes(const struct request
*rq
)
775 return rq
->bio
? bio_cur_bytes(rq
->bio
) : 0;
778 extern unsigned int blk_rq_err_bytes(const struct request
*rq
);
780 static inline unsigned int blk_rq_sectors(const struct request
*rq
)
782 return blk_rq_bytes(rq
) >> 9;
785 static inline unsigned int blk_rq_cur_sectors(const struct request
*rq
)
787 return blk_rq_cur_bytes(rq
) >> 9;
791 * Request issue related functions.
793 extern struct request
*blk_peek_request(struct request_queue
*q
);
794 extern void blk_start_request(struct request
*rq
);
795 extern struct request
*blk_fetch_request(struct request_queue
*q
);
798 * Request completion related functions.
800 * blk_update_request() completes given number of bytes and updates
801 * the request without completing it.
803 * blk_end_request() and friends. __blk_end_request() must be called
804 * with the request queue spinlock acquired.
806 * Several drivers define their own end_request and call
807 * blk_end_request() for parts of the original function.
808 * This prevents code duplication in drivers.
810 extern bool blk_update_request(struct request
*rq
, int error
,
811 unsigned int nr_bytes
);
812 extern bool blk_end_request(struct request
*rq
, int error
,
813 unsigned int nr_bytes
);
814 extern void blk_end_request_all(struct request
*rq
, int error
);
815 extern bool blk_end_request_cur(struct request
*rq
, int error
);
816 extern bool blk_end_request_err(struct request
*rq
, int error
);
817 extern bool __blk_end_request(struct request
*rq
, int error
,
818 unsigned int nr_bytes
);
819 extern void __blk_end_request_all(struct request
*rq
, int error
);
820 extern bool __blk_end_request_cur(struct request
*rq
, int error
);
821 extern bool __blk_end_request_err(struct request
*rq
, int error
);
823 extern void blk_complete_request(struct request
*);
824 extern void __blk_complete_request(struct request
*);
825 extern void blk_abort_request(struct request
*);
826 extern void blk_abort_queue(struct request_queue
*);
827 extern void blk_unprep_request(struct request
*);
830 * Access functions for manipulating queue properties
832 extern struct request_queue
*blk_init_queue_node(request_fn_proc
*rfn
,
833 spinlock_t
*lock
, int node_id
);
834 extern struct request_queue
*blk_init_queue(request_fn_proc
*, spinlock_t
*);
835 extern struct request_queue
*blk_init_allocated_queue(struct request_queue
*,
836 request_fn_proc
*, spinlock_t
*);
837 extern void blk_cleanup_queue(struct request_queue
*);
838 extern void blk_queue_make_request(struct request_queue
*, make_request_fn
*);
839 extern void blk_queue_bounce_limit(struct request_queue
*, u64
);
840 extern void blk_limits_max_hw_sectors(struct queue_limits
*, unsigned int);
841 extern void blk_queue_max_hw_sectors(struct request_queue
*, unsigned int);
842 extern void blk_queue_max_segments(struct request_queue
*, unsigned short);
843 extern void blk_queue_max_segment_size(struct request_queue
*, unsigned int);
844 extern void blk_queue_max_discard_sectors(struct request_queue
*q
,
845 unsigned int max_discard_sectors
);
846 extern void blk_queue_logical_block_size(struct request_queue
*, unsigned short);
847 extern void blk_queue_physical_block_size(struct request_queue
*, unsigned int);
848 extern void blk_queue_alignment_offset(struct request_queue
*q
,
849 unsigned int alignment
);
850 extern void blk_limits_io_min(struct queue_limits
*limits
, unsigned int min
);
851 extern void blk_queue_io_min(struct request_queue
*q
, unsigned int min
);
852 extern void blk_limits_io_opt(struct queue_limits
*limits
, unsigned int opt
);
853 extern void blk_queue_io_opt(struct request_queue
*q
, unsigned int opt
);
854 extern void blk_set_default_limits(struct queue_limits
*lim
);
855 extern void blk_set_stacking_limits(struct queue_limits
*lim
);
856 extern int blk_stack_limits(struct queue_limits
*t
, struct queue_limits
*b
,
858 extern int bdev_stack_limits(struct queue_limits
*t
, struct block_device
*bdev
,
860 extern void disk_stack_limits(struct gendisk
*disk
, struct block_device
*bdev
,
862 extern void blk_queue_stack_limits(struct request_queue
*t
, struct request_queue
*b
);
863 extern void blk_queue_dma_pad(struct request_queue
*, unsigned int);
864 extern void blk_queue_update_dma_pad(struct request_queue
*, unsigned int);
865 extern int blk_queue_dma_drain(struct request_queue
*q
,
866 dma_drain_needed_fn
*dma_drain_needed
,
867 void *buf
, unsigned int size
);
868 extern void blk_queue_lld_busy(struct request_queue
*q
, lld_busy_fn
*fn
);
869 extern void blk_queue_segment_boundary(struct request_queue
*, unsigned long);
870 extern void blk_queue_prep_rq(struct request_queue
*, prep_rq_fn
*pfn
);
871 extern void blk_queue_unprep_rq(struct request_queue
*, unprep_rq_fn
*ufn
);
872 extern void blk_queue_merge_bvec(struct request_queue
*, merge_bvec_fn
*);
873 extern void blk_queue_dma_alignment(struct request_queue
*, int);
874 extern void blk_queue_update_dma_alignment(struct request_queue
*, int);
875 extern void blk_queue_softirq_done(struct request_queue
*, softirq_done_fn
*);
876 extern void blk_queue_rq_timed_out(struct request_queue
*, rq_timed_out_fn
*);
877 extern void blk_queue_rq_timeout(struct request_queue
*, unsigned int);
878 extern void blk_queue_flush(struct request_queue
*q
, unsigned int flush
);
879 extern void blk_queue_flush_queueable(struct request_queue
*q
, bool queueable
);
880 extern struct backing_dev_info
*blk_get_backing_dev_info(struct block_device
*bdev
);
882 extern int blk_rq_map_sg(struct request_queue
*, struct request
*, struct scatterlist
*);
883 extern void blk_dump_rq_flags(struct request
*, char *);
884 extern long nr_blockdev_pages(void);
886 bool __must_check
blk_get_queue(struct request_queue
*);
887 struct request_queue
*blk_alloc_queue(gfp_t
);
888 struct request_queue
*blk_alloc_queue_node(gfp_t
, int);
889 extern void blk_put_queue(struct request_queue
*);
892 * blk_plug permits building a queue of related requests by holding the I/O
893 * fragments for a short period. This allows merging of sequential requests
894 * into single larger request. As the requests are moved from a per-task list to
895 * the device's request_queue in a batch, this results in improved scalability
896 * as the lock contention for request_queue lock is reduced.
898 * It is ok not to disable preemption when adding the request to the plug list
899 * or when attempting a merge, because blk_schedule_flush_list() will only flush
900 * the plug list when the task sleeps by itself. For details, please see
901 * schedule() where blk_schedule_flush_plug() is called.
904 unsigned long magic
; /* detect uninitialized use-cases */
905 struct list_head list
; /* requests */
906 struct list_head cb_list
; /* md requires an unplug callback */
907 unsigned int should_sort
; /* list to be sorted before flushing? */
909 #define BLK_MAX_REQUEST_COUNT 16
912 struct list_head list
;
913 void (*callback
)(struct blk_plug_cb
*);
916 extern void blk_start_plug(struct blk_plug
*);
917 extern void blk_finish_plug(struct blk_plug
*);
918 extern void blk_flush_plug_list(struct blk_plug
*, bool);
920 static inline void blk_flush_plug(struct task_struct
*tsk
)
922 struct blk_plug
*plug
= tsk
->plug
;
925 blk_flush_plug_list(plug
, false);
928 static inline void blk_schedule_flush_plug(struct task_struct
*tsk
)
930 struct blk_plug
*plug
= tsk
->plug
;
933 blk_flush_plug_list(plug
, true);
936 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
938 struct blk_plug
*plug
= tsk
->plug
;
940 return plug
&& (!list_empty(&plug
->list
) || !list_empty(&plug
->cb_list
));
946 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
947 extern int blk_queue_start_tag(struct request_queue
*, struct request
*);
948 extern struct request
*blk_queue_find_tag(struct request_queue
*, int);
949 extern void blk_queue_end_tag(struct request_queue
*, struct request
*);
950 extern int blk_queue_init_tags(struct request_queue
*, int, struct blk_queue_tag
*);
951 extern void blk_queue_free_tags(struct request_queue
*);
952 extern int blk_queue_resize_tags(struct request_queue
*, int);
953 extern void blk_queue_invalidate_tags(struct request_queue
*);
954 extern struct blk_queue_tag
*blk_init_tags(int);
955 extern void blk_free_tags(struct blk_queue_tag
*);
957 static inline struct request
*blk_map_queue_find_tag(struct blk_queue_tag
*bqt
,
960 if (unlikely(bqt
== NULL
|| tag
>= bqt
->real_max_depth
))
962 return bqt
->tag_index
[tag
];
965 #define BLKDEV_DISCARD_SECURE 0x01 /* secure discard */
967 extern int blkdev_issue_flush(struct block_device
*, gfp_t
, sector_t
*);
968 extern int blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
969 sector_t nr_sects
, gfp_t gfp_mask
, unsigned long flags
);
970 extern int blkdev_issue_zeroout(struct block_device
*bdev
, sector_t sector
,
971 sector_t nr_sects
, gfp_t gfp_mask
);
972 static inline int sb_issue_discard(struct super_block
*sb
, sector_t block
,
973 sector_t nr_blocks
, gfp_t gfp_mask
, unsigned long flags
)
975 return blkdev_issue_discard(sb
->s_bdev
, block
<< (sb
->s_blocksize_bits
- 9),
976 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
979 static inline int sb_issue_zeroout(struct super_block
*sb
, sector_t block
,
980 sector_t nr_blocks
, gfp_t gfp_mask
)
982 return blkdev_issue_zeroout(sb
->s_bdev
,
983 block
<< (sb
->s_blocksize_bits
- 9),
984 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
988 extern int blk_verify_command(unsigned char *cmd
, fmode_t has_write_perm
);
990 enum blk_default_limits
{
991 BLK_MAX_SEGMENTS
= 128,
992 BLK_SAFE_MAX_SECTORS
= 255,
993 BLK_DEF_MAX_SECTORS
= 1024,
994 BLK_MAX_SEGMENT_SIZE
= 65536,
995 BLK_SEG_BOUNDARY_MASK
= 0xFFFFFFFFUL
,
998 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
1000 static inline unsigned long queue_bounce_pfn(struct request_queue
*q
)
1002 return q
->limits
.bounce_pfn
;
1005 static inline unsigned long queue_segment_boundary(struct request_queue
*q
)
1007 return q
->limits
.seg_boundary_mask
;
1010 static inline unsigned int queue_max_sectors(struct request_queue
*q
)
1012 return q
->limits
.max_sectors
;
1015 static inline unsigned int queue_max_hw_sectors(struct request_queue
*q
)
1017 return q
->limits
.max_hw_sectors
;
1020 static inline unsigned short queue_max_segments(struct request_queue
*q
)
1022 return q
->limits
.max_segments
;
1025 static inline unsigned int queue_max_segment_size(struct request_queue
*q
)
1027 return q
->limits
.max_segment_size
;
1030 static inline unsigned short queue_logical_block_size(struct request_queue
*q
)
1034 if (q
&& q
->limits
.logical_block_size
)
1035 retval
= q
->limits
.logical_block_size
;
1040 static inline unsigned short bdev_logical_block_size(struct block_device
*bdev
)
1042 return queue_logical_block_size(bdev_get_queue(bdev
));
1045 static inline unsigned int queue_physical_block_size(struct request_queue
*q
)
1047 return q
->limits
.physical_block_size
;
1050 static inline unsigned int bdev_physical_block_size(struct block_device
*bdev
)
1052 return queue_physical_block_size(bdev_get_queue(bdev
));
1055 static inline unsigned int queue_io_min(struct request_queue
*q
)
1057 return q
->limits
.io_min
;
1060 static inline int bdev_io_min(struct block_device
*bdev
)
1062 return queue_io_min(bdev_get_queue(bdev
));
1065 static inline unsigned int queue_io_opt(struct request_queue
*q
)
1067 return q
->limits
.io_opt
;
1070 static inline int bdev_io_opt(struct block_device
*bdev
)
1072 return queue_io_opt(bdev_get_queue(bdev
));
1075 static inline int queue_alignment_offset(struct request_queue
*q
)
1077 if (q
->limits
.misaligned
)
1080 return q
->limits
.alignment_offset
;
1083 static inline int queue_limit_alignment_offset(struct queue_limits
*lim
, sector_t sector
)
1085 unsigned int granularity
= max(lim
->physical_block_size
, lim
->io_min
);
1086 unsigned int alignment
= (sector
<< 9) & (granularity
- 1);
1088 return (granularity
+ lim
->alignment_offset
- alignment
)
1089 & (granularity
- 1);
1092 static inline int bdev_alignment_offset(struct block_device
*bdev
)
1094 struct request_queue
*q
= bdev_get_queue(bdev
);
1096 if (q
->limits
.misaligned
)
1099 if (bdev
!= bdev
->bd_contains
)
1100 return bdev
->bd_part
->alignment_offset
;
1102 return q
->limits
.alignment_offset
;
1105 static inline int queue_discard_alignment(struct request_queue
*q
)
1107 if (q
->limits
.discard_misaligned
)
1110 return q
->limits
.discard_alignment
;
1113 static inline int queue_limit_discard_alignment(struct queue_limits
*lim
, sector_t sector
)
1115 unsigned int alignment
= (sector
<< 9) & (lim
->discard_granularity
- 1);
1117 if (!lim
->max_discard_sectors
)
1120 return (lim
->discard_granularity
+ lim
->discard_alignment
- alignment
)
1121 & (lim
->discard_granularity
- 1);
1124 static inline unsigned int queue_discard_zeroes_data(struct request_queue
*q
)
1126 if (q
->limits
.max_discard_sectors
&& q
->limits
.discard_zeroes_data
== 1)
1132 static inline unsigned int bdev_discard_zeroes_data(struct block_device
*bdev
)
1134 return queue_discard_zeroes_data(bdev_get_queue(bdev
));
1137 static inline int queue_dma_alignment(struct request_queue
*q
)
1139 return q
? q
->dma_alignment
: 511;
1142 static inline int blk_rq_aligned(struct request_queue
*q
, unsigned long addr
,
1145 unsigned int alignment
= queue_dma_alignment(q
) | q
->dma_pad_mask
;
1146 return !(addr
& alignment
) && !(len
& alignment
);
1149 /* assumes size > 256 */
1150 static inline unsigned int blksize_bits(unsigned int size
)
1152 unsigned int bits
= 8;
1156 } while (size
> 256);
1160 static inline unsigned int block_size(struct block_device
*bdev
)
1162 return bdev
->bd_block_size
;
1165 static inline bool queue_flush_queueable(struct request_queue
*q
)
1167 return !q
->flush_not_queueable
;
1170 typedef struct {struct page
*v
;} Sector
;
1172 unsigned char *read_dev_sector(struct block_device
*, sector_t
, Sector
*);
1174 static inline void put_dev_sector(Sector p
)
1176 page_cache_release(p
.v
);
1180 int kblockd_schedule_work(struct request_queue
*q
, struct work_struct
*work
);
1182 #ifdef CONFIG_BLK_CGROUP
1184 * This should not be using sched_clock(). A real patch is in progress
1185 * to fix this up, until that is in place we need to disable preemption
1186 * around sched_clock() in this function and set_io_start_time_ns().
1188 static inline void set_start_time_ns(struct request
*req
)
1191 req
->start_time_ns
= sched_clock();
1195 static inline void set_io_start_time_ns(struct request
*req
)
1198 req
->io_start_time_ns
= sched_clock();
1202 static inline uint64_t rq_start_time_ns(struct request
*req
)
1204 return req
->start_time_ns
;
1207 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1209 return req
->io_start_time_ns
;
1212 static inline void set_start_time_ns(struct request
*req
) {}
1213 static inline void set_io_start_time_ns(struct request
*req
) {}
1214 static inline uint64_t rq_start_time_ns(struct request
*req
)
1218 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1224 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1225 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1226 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1227 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1229 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1231 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1232 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1234 struct blk_integrity_exchg
{
1238 unsigned int data_size
;
1239 unsigned short sector_size
;
1240 const char *disk_name
;
1243 typedef void (integrity_gen_fn
) (struct blk_integrity_exchg
*);
1244 typedef int (integrity_vrfy_fn
) (struct blk_integrity_exchg
*);
1245 typedef void (integrity_set_tag_fn
) (void *, void *, unsigned int);
1246 typedef void (integrity_get_tag_fn
) (void *, void *, unsigned int);
1248 struct blk_integrity
{
1249 integrity_gen_fn
*generate_fn
;
1250 integrity_vrfy_fn
*verify_fn
;
1251 integrity_set_tag_fn
*set_tag_fn
;
1252 integrity_get_tag_fn
*get_tag_fn
;
1254 unsigned short flags
;
1255 unsigned short tuple_size
;
1256 unsigned short sector_size
;
1257 unsigned short tag_size
;
1261 struct kobject kobj
;
1264 extern bool blk_integrity_is_initialized(struct gendisk
*);
1265 extern int blk_integrity_register(struct gendisk
*, struct blk_integrity
*);
1266 extern void blk_integrity_unregister(struct gendisk
*);
1267 extern int blk_integrity_compare(struct gendisk
*, struct gendisk
*);
1268 extern int blk_rq_map_integrity_sg(struct request_queue
*, struct bio
*,
1269 struct scatterlist
*);
1270 extern int blk_rq_count_integrity_sg(struct request_queue
*, struct bio
*);
1271 extern int blk_integrity_merge_rq(struct request_queue
*, struct request
*,
1273 extern int blk_integrity_merge_bio(struct request_queue
*, struct request
*,
1277 struct blk_integrity
*bdev_get_integrity(struct block_device
*bdev
)
1279 return bdev
->bd_disk
->integrity
;
1282 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1284 return disk
->integrity
;
1287 static inline int blk_integrity_rq(struct request
*rq
)
1289 if (rq
->bio
== NULL
)
1292 return bio_integrity(rq
->bio
);
1295 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1298 q
->limits
.max_integrity_segments
= segs
;
1301 static inline unsigned short
1302 queue_max_integrity_segments(struct request_queue
*q
)
1304 return q
->limits
.max_integrity_segments
;
1307 #else /* CONFIG_BLK_DEV_INTEGRITY */
1310 struct block_device
;
1312 struct blk_integrity
;
1314 static inline int blk_integrity_rq(struct request
*rq
)
1318 static inline int blk_rq_count_integrity_sg(struct request_queue
*q
,
1323 static inline int blk_rq_map_integrity_sg(struct request_queue
*q
,
1325 struct scatterlist
*s
)
1329 static inline struct blk_integrity
*bdev_get_integrity(struct block_device
*b
)
1333 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1337 static inline int blk_integrity_compare(struct gendisk
*a
, struct gendisk
*b
)
1341 static inline int blk_integrity_register(struct gendisk
*d
,
1342 struct blk_integrity
*b
)
1346 static inline void blk_integrity_unregister(struct gendisk
*d
)
1349 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1353 static inline unsigned short queue_max_integrity_segments(struct request_queue
*q
)
1357 static inline int blk_integrity_merge_rq(struct request_queue
*rq
,
1363 static inline int blk_integrity_merge_bio(struct request_queue
*rq
,
1369 static inline bool blk_integrity_is_initialized(struct gendisk
*g
)
1374 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1376 struct block_device_operations
{
1377 int (*open
) (struct block_device
*, fmode_t
);
1378 int (*release
) (struct gendisk
*, fmode_t
);
1379 int (*ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1380 int (*compat_ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1381 int (*direct_access
) (struct block_device
*, sector_t
,
1382 void **, unsigned long *);
1383 unsigned int (*check_events
) (struct gendisk
*disk
,
1384 unsigned int clearing
);
1385 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1386 int (*media_changed
) (struct gendisk
*);
1387 void (*unlock_native_capacity
) (struct gendisk
*);
1388 int (*revalidate_disk
) (struct gendisk
*);
1389 int (*getgeo
)(struct block_device
*, struct hd_geometry
*);
1390 /* this callback is with swap_lock and sometimes page table lock held */
1391 void (*swap_slot_free_notify
) (struct block_device
*, unsigned long);
1392 struct module
*owner
;
1395 extern int __blkdev_driver_ioctl(struct block_device
*, fmode_t
, unsigned int,
1397 #else /* CONFIG_BLOCK */
1399 * stubs for when the block layer is configured out
1401 #define buffer_heads_over_limit 0
1403 static inline long nr_blockdev_pages(void)
1411 static inline void blk_start_plug(struct blk_plug
*plug
)
1415 static inline void blk_finish_plug(struct blk_plug
*plug
)
1419 static inline void blk_flush_plug(struct task_struct
*task
)
1423 static inline void blk_schedule_flush_plug(struct task_struct
*task
)
1428 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
1433 #endif /* CONFIG_BLOCK */