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
;
36 #define BLKDEV_MIN_RQ 4
37 #define BLKDEV_MAX_RQ 128 /* Default maximum */
40 typedef void (rq_end_io_fn
)(struct request
*, int);
44 * count[], starved[], and wait[] are indexed by
45 * BLK_RW_SYNC/BLK_RW_ASYNC
51 wait_queue_head_t wait
[2];
55 * request command types
57 enum rq_cmd_type_bits
{
58 REQ_TYPE_FS
= 1, /* fs request */
59 REQ_TYPE_BLOCK_PC
, /* scsi command */
60 REQ_TYPE_SENSE
, /* sense request */
61 REQ_TYPE_PM_SUSPEND
, /* suspend request */
62 REQ_TYPE_PM_RESUME
, /* resume request */
63 REQ_TYPE_PM_SHUTDOWN
, /* shutdown request */
64 REQ_TYPE_SPECIAL
, /* driver defined type */
66 * for ATA/ATAPI devices. this really doesn't belong here, ide should
67 * use REQ_TYPE_SPECIAL and use rq->cmd[0] with the range of driver
68 * private REQ_LB opcodes to differentiate what type of request this is
70 REQ_TYPE_ATA_TASKFILE
,
74 #define BLK_MAX_CDB 16
77 * try to put the fields that are referenced together in the same cacheline.
78 * if you modify this structure, be sure to check block/blk-core.c:blk_rq_init()
82 struct list_head queuelist
;
83 struct call_single_data csd
;
85 struct request_queue
*q
;
87 unsigned int cmd_flags
;
88 enum rq_cmd_type_bits cmd_type
;
89 unsigned long atomic_flags
;
93 /* the following two fields are internal, NEVER access directly */
94 unsigned int __data_len
; /* total data len */
95 sector_t __sector
; /* sector cursor */
100 struct hlist_node hash
; /* merge hash */
102 * The rb_node is only used inside the io scheduler, requests
103 * are pruned when moved to the dispatch queue. So let the
104 * completion_data share space with the rb_node.
107 struct rb_node rb_node
; /* sort/lookup */
108 void *completion_data
;
112 * Three pointers are available for the IO schedulers, if they need
113 * more they have to dynamically allocate it. Flush requests are
114 * never put on the IO scheduler. So let the flush fields share
115 * space with the elevator data.
125 struct list_head list
;
126 rq_end_io_fn
*saved_end_io
;
130 struct gendisk
*rq_disk
;
131 struct hd_struct
*part
;
132 unsigned long start_time
;
133 #ifdef CONFIG_BLK_CGROUP
134 unsigned long long start_time_ns
;
135 unsigned long long io_start_time_ns
; /* when passed to hardware */
137 /* Number of scatter-gather DMA addr+len pairs after
138 * physical address coalescing is performed.
140 unsigned short nr_phys_segments
;
141 #if defined(CONFIG_BLK_DEV_INTEGRITY)
142 unsigned short nr_integrity_segments
;
145 unsigned short ioprio
;
149 void *special
; /* opaque pointer available for LLD use */
150 char *buffer
; /* kaddr of the current segment if available */
156 * when request is used as a packet command carrier
158 unsigned char __cmd
[BLK_MAX_CDB
];
160 unsigned short cmd_len
;
162 unsigned int extra_len
; /* length of alignment and padding */
163 unsigned int sense_len
;
164 unsigned int resid_len
; /* residual count */
167 unsigned long deadline
;
168 struct list_head timeout_list
;
169 unsigned int timeout
;
173 * completion callback.
175 rq_end_io_fn
*end_io
;
179 struct request
*next_rq
;
182 static inline unsigned short req_get_ioprio(struct request
*req
)
188 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
189 * requests. Some step values could eventually be made generic.
191 struct request_pm_state
193 /* PM state machine step value, currently driver specific */
195 /* requested PM state value (S1, S2, S3, S4, ...) */
197 void* data
; /* for driver use */
200 #include <linux/elevator.h>
202 typedef void (request_fn_proc
) (struct request_queue
*q
);
203 typedef void (make_request_fn
) (struct request_queue
*q
, struct bio
*bio
);
204 typedef int (prep_rq_fn
) (struct request_queue
*, struct request
*);
205 typedef void (unprep_rq_fn
) (struct request_queue
*, struct request
*);
208 struct bvec_merge_data
{
209 struct block_device
*bi_bdev
;
214 typedef int (merge_bvec_fn
) (struct request_queue
*, struct bvec_merge_data
*,
216 typedef void (softirq_done_fn
)(struct request
*);
217 typedef int (dma_drain_needed_fn
)(struct request
*);
218 typedef int (lld_busy_fn
) (struct request_queue
*q
);
219 typedef int (bsg_job_fn
) (struct bsg_job
*);
221 enum blk_eh_timer_return
{
227 typedef enum blk_eh_timer_return (rq_timed_out_fn
)(struct request
*);
229 enum blk_queue_state
{
234 struct blk_queue_tag
{
235 struct request
**tag_index
; /* map of busy tags */
236 unsigned long *tag_map
; /* bit map of free/busy tags */
237 int busy
; /* current depth */
238 int max_depth
; /* what we will send to device */
239 int real_max_depth
; /* what the array can hold */
240 atomic_t refcnt
; /* map can be shared */
243 #define BLK_SCSI_MAX_CMDS (256)
244 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
246 struct queue_limits
{
247 unsigned long bounce_pfn
;
248 unsigned long seg_boundary_mask
;
250 unsigned int max_hw_sectors
;
251 unsigned int max_sectors
;
252 unsigned int max_segment_size
;
253 unsigned int physical_block_size
;
254 unsigned int alignment_offset
;
257 unsigned int max_discard_sectors
;
258 unsigned int discard_granularity
;
259 unsigned int discard_alignment
;
261 unsigned short logical_block_size
;
262 unsigned short max_segments
;
263 unsigned short max_integrity_segments
;
265 unsigned char misaligned
;
266 unsigned char discard_misaligned
;
267 unsigned char cluster
;
268 unsigned char discard_zeroes_data
;
271 struct request_queue
{
273 * Together with queue_head for cacheline sharing
275 struct list_head queue_head
;
276 struct request
*last_merge
;
277 struct elevator_queue
*elevator
;
280 * the queue request freelist, one for reads and one for writes
282 struct request_list rq
;
284 request_fn_proc
*request_fn
;
285 make_request_fn
*make_request_fn
;
286 prep_rq_fn
*prep_rq_fn
;
287 unprep_rq_fn
*unprep_rq_fn
;
288 merge_bvec_fn
*merge_bvec_fn
;
289 softirq_done_fn
*softirq_done_fn
;
290 rq_timed_out_fn
*rq_timed_out_fn
;
291 dma_drain_needed_fn
*dma_drain_needed
;
292 lld_busy_fn
*lld_busy_fn
;
295 * Dispatch queue sorting
298 struct request
*boundary_rq
;
301 * Delayed queue handling
303 struct delayed_work delay_work
;
305 struct backing_dev_info backing_dev_info
;
308 * The queue owner gets to use this for whatever they like.
309 * ll_rw_blk doesn't touch it.
314 * various queue flags, see QUEUE_* below
316 unsigned long queue_flags
;
319 * ida allocated id for this queue. Used to index queues from
325 * queue needs bounce pages for pages above this limit
330 * protects queue structures from reentrancy. ->__queue_lock should
331 * _never_ be used directly, it is queue private. always use
334 spinlock_t __queue_lock
;
335 spinlock_t
*queue_lock
;
345 unsigned long nr_requests
; /* Max # of requests */
346 unsigned int nr_congestion_on
;
347 unsigned int nr_congestion_off
;
348 unsigned int nr_batching
;
350 unsigned int dma_drain_size
;
351 void *dma_drain_buffer
;
352 unsigned int dma_pad_mask
;
353 unsigned int dma_alignment
;
355 struct blk_queue_tag
*queue_tags
;
356 struct list_head tag_busy_list
;
358 unsigned int nr_sorted
;
359 unsigned int in_flight
[2];
361 unsigned int rq_timeout
;
362 struct timer_list timeout
;
363 struct list_head timeout_list
;
365 struct list_head icq_list
;
367 struct queue_limits limits
;
372 unsigned int sg_timeout
;
373 unsigned int sg_reserved_size
;
375 #ifdef CONFIG_BLK_DEV_IO_TRACE
376 struct blk_trace
*blk_trace
;
379 * for flush operations
381 unsigned int flush_flags
;
382 unsigned int flush_not_queueable
:1;
383 unsigned int flush_queue_delayed
:1;
384 unsigned int flush_pending_idx
:1;
385 unsigned int flush_running_idx
:1;
386 unsigned long flush_pending_since
;
387 struct list_head flush_queue
[2];
388 struct list_head flush_data_in_flight
;
389 struct request flush_rq
;
391 struct mutex sysfs_lock
;
393 #if defined(CONFIG_BLK_DEV_BSG)
394 bsg_job_fn
*bsg_job_fn
;
396 struct bsg_class_device bsg_dev
;
399 #ifdef CONFIG_BLK_DEV_THROTTLING
401 struct throtl_data
*td
;
405 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
406 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
407 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
408 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
409 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
410 #define QUEUE_FLAG_ELVSWITCH 6 /* don't use elevator, just do FIFO */
411 #define QUEUE_FLAG_BIDI 7 /* queue supports bidi requests */
412 #define QUEUE_FLAG_NOMERGES 8 /* disable merge attempts */
413 #define QUEUE_FLAG_SAME_COMP 9 /* complete on same CPU-group */
414 #define QUEUE_FLAG_FAIL_IO 10 /* fake timeout */
415 #define QUEUE_FLAG_STACKABLE 11 /* supports request stacking */
416 #define QUEUE_FLAG_NONROT 12 /* non-rotational device (SSD) */
417 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
418 #define QUEUE_FLAG_IO_STAT 13 /* do IO stats */
419 #define QUEUE_FLAG_DISCARD 14 /* supports DISCARD */
420 #define QUEUE_FLAG_NOXMERGES 15 /* No extended merges */
421 #define QUEUE_FLAG_ADD_RANDOM 16 /* Contributes to random pool */
422 #define QUEUE_FLAG_SECDISCARD 17 /* supports SECDISCARD */
423 #define QUEUE_FLAG_SAME_FORCE 18 /* force complete on same CPU */
425 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
426 (1 << QUEUE_FLAG_STACKABLE) | \
427 (1 << QUEUE_FLAG_SAME_COMP) | \
428 (1 << QUEUE_FLAG_ADD_RANDOM))
430 static inline void queue_lockdep_assert_held(struct request_queue
*q
)
433 lockdep_assert_held(q
->queue_lock
);
436 static inline void queue_flag_set_unlocked(unsigned int flag
,
437 struct request_queue
*q
)
439 __set_bit(flag
, &q
->queue_flags
);
442 static inline int queue_flag_test_and_clear(unsigned int flag
,
443 struct request_queue
*q
)
445 queue_lockdep_assert_held(q
);
447 if (test_bit(flag
, &q
->queue_flags
)) {
448 __clear_bit(flag
, &q
->queue_flags
);
455 static inline int queue_flag_test_and_set(unsigned int flag
,
456 struct request_queue
*q
)
458 queue_lockdep_assert_held(q
);
460 if (!test_bit(flag
, &q
->queue_flags
)) {
461 __set_bit(flag
, &q
->queue_flags
);
468 static inline void queue_flag_set(unsigned int flag
, struct request_queue
*q
)
470 queue_lockdep_assert_held(q
);
471 __set_bit(flag
, &q
->queue_flags
);
474 static inline void queue_flag_clear_unlocked(unsigned int flag
,
475 struct request_queue
*q
)
477 __clear_bit(flag
, &q
->queue_flags
);
480 static inline int queue_in_flight(struct request_queue
*q
)
482 return q
->in_flight
[0] + q
->in_flight
[1];
485 static inline void queue_flag_clear(unsigned int flag
, struct request_queue
*q
)
487 queue_lockdep_assert_held(q
);
488 __clear_bit(flag
, &q
->queue_flags
);
491 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
492 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
493 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
494 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
495 #define blk_queue_noxmerges(q) \
496 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
497 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
498 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
499 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
500 #define blk_queue_stackable(q) \
501 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
502 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
503 #define blk_queue_secdiscard(q) (blk_queue_discard(q) && \
504 test_bit(QUEUE_FLAG_SECDISCARD, &(q)->queue_flags))
506 #define blk_noretry_request(rq) \
507 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
508 REQ_FAILFAST_DRIVER))
510 #define blk_account_rq(rq) \
511 (((rq)->cmd_flags & REQ_STARTED) && \
512 ((rq)->cmd_type == REQ_TYPE_FS || \
513 ((rq)->cmd_flags & REQ_DISCARD)))
515 #define blk_pm_request(rq) \
516 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
517 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
519 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
520 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
521 /* rq->queuelist of dequeued request must be list_empty() */
522 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
524 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
526 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
528 static inline unsigned int blk_queue_cluster(struct request_queue
*q
)
530 return q
->limits
.cluster
;
534 * We regard a request as sync, if either a read or a sync write
536 static inline bool rw_is_sync(unsigned int rw_flags
)
538 return !(rw_flags
& REQ_WRITE
) || (rw_flags
& REQ_SYNC
);
541 static inline bool rq_is_sync(struct request
*rq
)
543 return rw_is_sync(rq
->cmd_flags
);
546 static inline int blk_queue_full(struct request_queue
*q
, int sync
)
549 return test_bit(QUEUE_FLAG_SYNCFULL
, &q
->queue_flags
);
550 return test_bit(QUEUE_FLAG_ASYNCFULL
, &q
->queue_flags
);
553 static inline void blk_set_queue_full(struct request_queue
*q
, int sync
)
556 queue_flag_set(QUEUE_FLAG_SYNCFULL
, q
);
558 queue_flag_set(QUEUE_FLAG_ASYNCFULL
, q
);
561 static inline void blk_clear_queue_full(struct request_queue
*q
, int sync
)
564 queue_flag_clear(QUEUE_FLAG_SYNCFULL
, q
);
566 queue_flag_clear(QUEUE_FLAG_ASYNCFULL
, q
);
571 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
572 * it already be started by driver.
574 #define RQ_NOMERGE_FLAGS \
575 (REQ_NOMERGE | REQ_STARTED | REQ_SOFTBARRIER | REQ_FLUSH | REQ_FUA)
576 #define rq_mergeable(rq) \
577 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
578 (((rq)->cmd_flags & REQ_DISCARD) || \
579 (rq)->cmd_type == REQ_TYPE_FS))
582 * q->prep_rq_fn return values
584 #define BLKPREP_OK 0 /* serve it */
585 #define BLKPREP_KILL 1 /* fatal error, kill */
586 #define BLKPREP_DEFER 2 /* leave on queue */
588 extern unsigned long blk_max_low_pfn
, blk_max_pfn
;
591 * standard bounce addresses:
593 * BLK_BOUNCE_HIGH : bounce all highmem pages
594 * BLK_BOUNCE_ANY : don't bounce anything
595 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
598 #if BITS_PER_LONG == 32
599 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
601 #define BLK_BOUNCE_HIGH -1ULL
603 #define BLK_BOUNCE_ANY (-1ULL)
604 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
607 * default timeout for SG_IO if none specified
609 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
610 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
613 extern int init_emergency_isa_pool(void);
614 extern void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
);
616 static inline int init_emergency_isa_pool(void)
620 static inline void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
)
623 #endif /* CONFIG_MMU */
629 unsigned long offset
;
634 struct req_iterator
{
639 /* This should not be used directly - use rq_for_each_segment */
640 #define for_each_bio(_bio) \
641 for (; _bio; _bio = _bio->bi_next)
642 #define __rq_for_each_bio(_bio, rq) \
644 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
646 #define rq_for_each_segment(bvl, _rq, _iter) \
647 __rq_for_each_bio(_iter.bio, _rq) \
648 bio_for_each_segment(bvl, _iter.bio, _iter.i)
650 #define rq_iter_last(rq, _iter) \
651 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
653 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
654 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
656 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
657 extern void rq_flush_dcache_pages(struct request
*rq
);
659 static inline void rq_flush_dcache_pages(struct request
*rq
)
664 extern int blk_register_queue(struct gendisk
*disk
);
665 extern void blk_unregister_queue(struct gendisk
*disk
);
666 extern void generic_make_request(struct bio
*bio
);
667 extern void blk_rq_init(struct request_queue
*q
, struct request
*rq
);
668 extern void blk_put_request(struct request
*);
669 extern void __blk_put_request(struct request_queue
*, struct request
*);
670 extern struct request
*blk_get_request(struct request_queue
*, int, gfp_t
);
671 extern struct request
*blk_make_request(struct request_queue
*, struct bio
*,
673 extern void blk_requeue_request(struct request_queue
*, struct request
*);
674 extern void blk_add_request_payload(struct request
*rq
, struct page
*page
,
676 extern int blk_rq_check_limits(struct request_queue
*q
, struct request
*rq
);
677 extern int blk_lld_busy(struct request_queue
*q
);
678 extern int blk_rq_prep_clone(struct request
*rq
, struct request
*rq_src
,
679 struct bio_set
*bs
, gfp_t gfp_mask
,
680 int (*bio_ctr
)(struct bio
*, struct bio
*, void *),
682 extern void blk_rq_unprep_clone(struct request
*rq
);
683 extern int blk_insert_cloned_request(struct request_queue
*q
,
685 extern void blk_delay_queue(struct request_queue
*, unsigned long);
686 extern void blk_recount_segments(struct request_queue
*, struct bio
*);
687 extern int scsi_verify_blk_ioctl(struct block_device
*, unsigned int);
688 extern int scsi_cmd_blk_ioctl(struct block_device
*, fmode_t
,
689 unsigned int, void __user
*);
690 extern int scsi_cmd_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
691 unsigned int, void __user
*);
692 extern int sg_scsi_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
693 struct scsi_ioctl_command __user
*);
695 extern void blk_queue_bio(struct request_queue
*q
, struct bio
*bio
);
698 * A queue has just exitted congestion. Note this in the global counter of
699 * congested queues, and wake up anyone who was waiting for requests to be
702 static inline void blk_clear_queue_congested(struct request_queue
*q
, int sync
)
704 clear_bdi_congested(&q
->backing_dev_info
, sync
);
708 * A queue has just entered congestion. Flag that in the queue's VM-visible
709 * state flags and increment the global gounter of congested queues.
711 static inline void blk_set_queue_congested(struct request_queue
*q
, int sync
)
713 set_bdi_congested(&q
->backing_dev_info
, sync
);
716 extern void blk_start_queue(struct request_queue
*q
);
717 extern void blk_stop_queue(struct request_queue
*q
);
718 extern void blk_sync_queue(struct request_queue
*q
);
719 extern void __blk_stop_queue(struct request_queue
*q
);
720 extern void __blk_run_queue(struct request_queue
*q
);
721 extern void blk_run_queue(struct request_queue
*);
722 extern void blk_run_queue_async(struct request_queue
*q
);
723 extern int blk_rq_map_user(struct request_queue
*, struct request
*,
724 struct rq_map_data
*, void __user
*, unsigned long,
726 extern int blk_rq_unmap_user(struct bio
*);
727 extern int blk_rq_map_kern(struct request_queue
*, struct request
*, void *, unsigned int, gfp_t
);
728 extern int blk_rq_map_user_iov(struct request_queue
*, struct request
*,
729 struct rq_map_data
*, struct sg_iovec
*, int,
730 unsigned int, gfp_t
);
731 extern int blk_execute_rq(struct request_queue
*, struct gendisk
*,
732 struct request
*, int);
733 extern void blk_execute_rq_nowait(struct request_queue
*, struct gendisk
*,
734 struct request
*, int, rq_end_io_fn
*);
736 static inline struct request_queue
*bdev_get_queue(struct block_device
*bdev
)
738 return bdev
->bd_disk
->queue
;
742 * blk_rq_pos() : the current sector
743 * blk_rq_bytes() : bytes left in the entire request
744 * blk_rq_cur_bytes() : bytes left in the current segment
745 * blk_rq_err_bytes() : bytes left till the next error boundary
746 * blk_rq_sectors() : sectors left in the entire request
747 * blk_rq_cur_sectors() : sectors left in the current segment
749 static inline sector_t
blk_rq_pos(const struct request
*rq
)
754 static inline unsigned int blk_rq_bytes(const struct request
*rq
)
756 return rq
->__data_len
;
759 static inline int blk_rq_cur_bytes(const struct request
*rq
)
761 return rq
->bio
? bio_cur_bytes(rq
->bio
) : 0;
764 extern unsigned int blk_rq_err_bytes(const struct request
*rq
);
766 static inline unsigned int blk_rq_sectors(const struct request
*rq
)
768 return blk_rq_bytes(rq
) >> 9;
771 static inline unsigned int blk_rq_cur_sectors(const struct request
*rq
)
773 return blk_rq_cur_bytes(rq
) >> 9;
777 * Request issue related functions.
779 extern struct request
*blk_peek_request(struct request_queue
*q
);
780 extern void blk_start_request(struct request
*rq
);
781 extern struct request
*blk_fetch_request(struct request_queue
*q
);
784 * Request completion related functions.
786 * blk_update_request() completes given number of bytes and updates
787 * the request without completing it.
789 * blk_end_request() and friends. __blk_end_request() must be called
790 * with the request queue spinlock acquired.
792 * Several drivers define their own end_request and call
793 * blk_end_request() for parts of the original function.
794 * This prevents code duplication in drivers.
796 extern bool blk_update_request(struct request
*rq
, int error
,
797 unsigned int nr_bytes
);
798 extern bool blk_end_request(struct request
*rq
, int error
,
799 unsigned int nr_bytes
);
800 extern void blk_end_request_all(struct request
*rq
, int error
);
801 extern bool blk_end_request_cur(struct request
*rq
, int error
);
802 extern bool blk_end_request_err(struct request
*rq
, int error
);
803 extern bool __blk_end_request(struct request
*rq
, int error
,
804 unsigned int nr_bytes
);
805 extern void __blk_end_request_all(struct request
*rq
, int error
);
806 extern bool __blk_end_request_cur(struct request
*rq
, int error
);
807 extern bool __blk_end_request_err(struct request
*rq
, int error
);
809 extern void blk_complete_request(struct request
*);
810 extern void __blk_complete_request(struct request
*);
811 extern void blk_abort_request(struct request
*);
812 extern void blk_abort_queue(struct request_queue
*);
813 extern void blk_unprep_request(struct request
*);
816 * Access functions for manipulating queue properties
818 extern struct request_queue
*blk_init_queue_node(request_fn_proc
*rfn
,
819 spinlock_t
*lock
, int node_id
);
820 extern struct request_queue
*blk_init_queue(request_fn_proc
*, spinlock_t
*);
821 extern struct request_queue
*blk_init_allocated_queue(struct request_queue
*,
822 request_fn_proc
*, spinlock_t
*);
823 extern void blk_cleanup_queue(struct request_queue
*);
824 extern void blk_queue_make_request(struct request_queue
*, make_request_fn
*);
825 extern void blk_queue_bounce_limit(struct request_queue
*, u64
);
826 extern void blk_limits_max_hw_sectors(struct queue_limits
*, unsigned int);
827 extern void blk_queue_max_hw_sectors(struct request_queue
*, unsigned int);
828 extern void blk_queue_max_segments(struct request_queue
*, unsigned short);
829 extern void blk_queue_max_segment_size(struct request_queue
*, unsigned int);
830 extern void blk_queue_max_discard_sectors(struct request_queue
*q
,
831 unsigned int max_discard_sectors
);
832 extern void blk_queue_logical_block_size(struct request_queue
*, unsigned short);
833 extern void blk_queue_physical_block_size(struct request_queue
*, unsigned int);
834 extern void blk_queue_alignment_offset(struct request_queue
*q
,
835 unsigned int alignment
);
836 extern void blk_limits_io_min(struct queue_limits
*limits
, unsigned int min
);
837 extern void blk_queue_io_min(struct request_queue
*q
, unsigned int min
);
838 extern void blk_limits_io_opt(struct queue_limits
*limits
, unsigned int opt
);
839 extern void blk_queue_io_opt(struct request_queue
*q
, unsigned int opt
);
840 extern void blk_set_default_limits(struct queue_limits
*lim
);
841 extern void blk_set_stacking_limits(struct queue_limits
*lim
);
842 extern int blk_stack_limits(struct queue_limits
*t
, struct queue_limits
*b
,
844 extern int bdev_stack_limits(struct queue_limits
*t
, struct block_device
*bdev
,
846 extern void disk_stack_limits(struct gendisk
*disk
, struct block_device
*bdev
,
848 extern void blk_queue_stack_limits(struct request_queue
*t
, struct request_queue
*b
);
849 extern void blk_queue_dma_pad(struct request_queue
*, unsigned int);
850 extern void blk_queue_update_dma_pad(struct request_queue
*, unsigned int);
851 extern int blk_queue_dma_drain(struct request_queue
*q
,
852 dma_drain_needed_fn
*dma_drain_needed
,
853 void *buf
, unsigned int size
);
854 extern void blk_queue_lld_busy(struct request_queue
*q
, lld_busy_fn
*fn
);
855 extern void blk_queue_segment_boundary(struct request_queue
*, unsigned long);
856 extern void blk_queue_prep_rq(struct request_queue
*, prep_rq_fn
*pfn
);
857 extern void blk_queue_unprep_rq(struct request_queue
*, unprep_rq_fn
*ufn
);
858 extern void blk_queue_merge_bvec(struct request_queue
*, merge_bvec_fn
*);
859 extern void blk_queue_dma_alignment(struct request_queue
*, int);
860 extern void blk_queue_update_dma_alignment(struct request_queue
*, int);
861 extern void blk_queue_softirq_done(struct request_queue
*, softirq_done_fn
*);
862 extern void blk_queue_rq_timed_out(struct request_queue
*, rq_timed_out_fn
*);
863 extern void blk_queue_rq_timeout(struct request_queue
*, unsigned int);
864 extern void blk_queue_flush(struct request_queue
*q
, unsigned int flush
);
865 extern void blk_queue_flush_queueable(struct request_queue
*q
, bool queueable
);
866 extern struct backing_dev_info
*blk_get_backing_dev_info(struct block_device
*bdev
);
868 extern int blk_rq_map_sg(struct request_queue
*, struct request
*, struct scatterlist
*);
869 extern void blk_dump_rq_flags(struct request
*, char *);
870 extern long nr_blockdev_pages(void);
872 bool __must_check
blk_get_queue(struct request_queue
*);
873 struct request_queue
*blk_alloc_queue(gfp_t
);
874 struct request_queue
*blk_alloc_queue_node(gfp_t
, int);
875 extern void blk_put_queue(struct request_queue
*);
878 * blk_plug permits building a queue of related requests by holding the I/O
879 * fragments for a short period. This allows merging of sequential requests
880 * into single larger request. As the requests are moved from a per-task list to
881 * the device's request_queue in a batch, this results in improved scalability
882 * as the lock contention for request_queue lock is reduced.
884 * It is ok not to disable preemption when adding the request to the plug list
885 * or when attempting a merge, because blk_schedule_flush_list() will only flush
886 * the plug list when the task sleeps by itself. For details, please see
887 * schedule() where blk_schedule_flush_plug() is called.
890 unsigned long magic
; /* detect uninitialized use-cases */
891 struct list_head list
; /* requests */
892 struct list_head cb_list
; /* md requires an unplug callback */
893 unsigned int should_sort
; /* list to be sorted before flushing? */
895 #define BLK_MAX_REQUEST_COUNT 16
898 struct list_head list
;
899 void (*callback
)(struct blk_plug_cb
*);
902 extern void blk_start_plug(struct blk_plug
*);
903 extern void blk_finish_plug(struct blk_plug
*);
904 extern void blk_flush_plug_list(struct blk_plug
*, bool);
906 static inline void blk_flush_plug(struct task_struct
*tsk
)
908 struct blk_plug
*plug
= tsk
->plug
;
911 blk_flush_plug_list(plug
, false);
914 static inline void blk_schedule_flush_plug(struct task_struct
*tsk
)
916 struct blk_plug
*plug
= tsk
->plug
;
919 blk_flush_plug_list(plug
, true);
922 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
924 struct blk_plug
*plug
= tsk
->plug
;
926 return plug
&& (!list_empty(&plug
->list
) || !list_empty(&plug
->cb_list
));
932 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
933 extern int blk_queue_start_tag(struct request_queue
*, struct request
*);
934 extern struct request
*blk_queue_find_tag(struct request_queue
*, int);
935 extern void blk_queue_end_tag(struct request_queue
*, struct request
*);
936 extern int blk_queue_init_tags(struct request_queue
*, int, struct blk_queue_tag
*);
937 extern void blk_queue_free_tags(struct request_queue
*);
938 extern int blk_queue_resize_tags(struct request_queue
*, int);
939 extern void blk_queue_invalidate_tags(struct request_queue
*);
940 extern struct blk_queue_tag
*blk_init_tags(int);
941 extern void blk_free_tags(struct blk_queue_tag
*);
943 static inline struct request
*blk_map_queue_find_tag(struct blk_queue_tag
*bqt
,
946 if (unlikely(bqt
== NULL
|| tag
>= bqt
->real_max_depth
))
948 return bqt
->tag_index
[tag
];
951 #define BLKDEV_DISCARD_SECURE 0x01 /* secure discard */
953 extern int blkdev_issue_flush(struct block_device
*, gfp_t
, sector_t
*);
954 extern int blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
955 sector_t nr_sects
, gfp_t gfp_mask
, unsigned long flags
);
956 extern int blkdev_issue_zeroout(struct block_device
*bdev
, sector_t sector
,
957 sector_t nr_sects
, gfp_t gfp_mask
);
958 static inline int sb_issue_discard(struct super_block
*sb
, sector_t block
,
959 sector_t nr_blocks
, gfp_t gfp_mask
, unsigned long flags
)
961 return blkdev_issue_discard(sb
->s_bdev
, block
<< (sb
->s_blocksize_bits
- 9),
962 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
965 static inline int sb_issue_zeroout(struct super_block
*sb
, sector_t block
,
966 sector_t nr_blocks
, gfp_t gfp_mask
)
968 return blkdev_issue_zeroout(sb
->s_bdev
,
969 block
<< (sb
->s_blocksize_bits
- 9),
970 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
974 extern int blk_verify_command(unsigned char *cmd
, fmode_t has_write_perm
);
976 enum blk_default_limits
{
977 BLK_MAX_SEGMENTS
= 128,
978 BLK_SAFE_MAX_SECTORS
= 255,
979 BLK_DEF_MAX_SECTORS
= 1024,
980 BLK_MAX_SEGMENT_SIZE
= 65536,
981 BLK_SEG_BOUNDARY_MASK
= 0xFFFFFFFFUL
,
984 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
986 static inline unsigned long queue_bounce_pfn(struct request_queue
*q
)
988 return q
->limits
.bounce_pfn
;
991 static inline unsigned long queue_segment_boundary(struct request_queue
*q
)
993 return q
->limits
.seg_boundary_mask
;
996 static inline unsigned int queue_max_sectors(struct request_queue
*q
)
998 return q
->limits
.max_sectors
;
1001 static inline unsigned int queue_max_hw_sectors(struct request_queue
*q
)
1003 return q
->limits
.max_hw_sectors
;
1006 static inline unsigned short queue_max_segments(struct request_queue
*q
)
1008 return q
->limits
.max_segments
;
1011 static inline unsigned int queue_max_segment_size(struct request_queue
*q
)
1013 return q
->limits
.max_segment_size
;
1016 static inline unsigned short queue_logical_block_size(struct request_queue
*q
)
1020 if (q
&& q
->limits
.logical_block_size
)
1021 retval
= q
->limits
.logical_block_size
;
1026 static inline unsigned short bdev_logical_block_size(struct block_device
*bdev
)
1028 return queue_logical_block_size(bdev_get_queue(bdev
));
1031 static inline unsigned int queue_physical_block_size(struct request_queue
*q
)
1033 return q
->limits
.physical_block_size
;
1036 static inline unsigned int bdev_physical_block_size(struct block_device
*bdev
)
1038 return queue_physical_block_size(bdev_get_queue(bdev
));
1041 static inline unsigned int queue_io_min(struct request_queue
*q
)
1043 return q
->limits
.io_min
;
1046 static inline int bdev_io_min(struct block_device
*bdev
)
1048 return queue_io_min(bdev_get_queue(bdev
));
1051 static inline unsigned int queue_io_opt(struct request_queue
*q
)
1053 return q
->limits
.io_opt
;
1056 static inline int bdev_io_opt(struct block_device
*bdev
)
1058 return queue_io_opt(bdev_get_queue(bdev
));
1061 static inline int queue_alignment_offset(struct request_queue
*q
)
1063 if (q
->limits
.misaligned
)
1066 return q
->limits
.alignment_offset
;
1069 static inline int queue_limit_alignment_offset(struct queue_limits
*lim
, sector_t sector
)
1071 unsigned int granularity
= max(lim
->physical_block_size
, lim
->io_min
);
1072 unsigned int alignment
= (sector
<< 9) & (granularity
- 1);
1074 return (granularity
+ lim
->alignment_offset
- alignment
)
1075 & (granularity
- 1);
1078 static inline int bdev_alignment_offset(struct block_device
*bdev
)
1080 struct request_queue
*q
= bdev_get_queue(bdev
);
1082 if (q
->limits
.misaligned
)
1085 if (bdev
!= bdev
->bd_contains
)
1086 return bdev
->bd_part
->alignment_offset
;
1088 return q
->limits
.alignment_offset
;
1091 static inline int queue_discard_alignment(struct request_queue
*q
)
1093 if (q
->limits
.discard_misaligned
)
1096 return q
->limits
.discard_alignment
;
1099 static inline int queue_limit_discard_alignment(struct queue_limits
*lim
, sector_t sector
)
1101 unsigned int alignment
= (sector
<< 9) & (lim
->discard_granularity
- 1);
1103 if (!lim
->max_discard_sectors
)
1106 return (lim
->discard_granularity
+ lim
->discard_alignment
- alignment
)
1107 & (lim
->discard_granularity
- 1);
1110 static inline unsigned int queue_discard_zeroes_data(struct request_queue
*q
)
1112 if (q
->limits
.max_discard_sectors
&& q
->limits
.discard_zeroes_data
== 1)
1118 static inline unsigned int bdev_discard_zeroes_data(struct block_device
*bdev
)
1120 return queue_discard_zeroes_data(bdev_get_queue(bdev
));
1123 static inline int queue_dma_alignment(struct request_queue
*q
)
1125 return q
? q
->dma_alignment
: 511;
1128 static inline int blk_rq_aligned(struct request_queue
*q
, unsigned long addr
,
1131 unsigned int alignment
= queue_dma_alignment(q
) | q
->dma_pad_mask
;
1132 return !(addr
& alignment
) && !(len
& alignment
);
1135 /* assumes size > 256 */
1136 static inline unsigned int blksize_bits(unsigned int size
)
1138 unsigned int bits
= 8;
1142 } while (size
> 256);
1146 static inline unsigned int block_size(struct block_device
*bdev
)
1148 return bdev
->bd_block_size
;
1151 static inline bool queue_flush_queueable(struct request_queue
*q
)
1153 return !q
->flush_not_queueable
;
1156 typedef struct {struct page
*v
;} Sector
;
1158 unsigned char *read_dev_sector(struct block_device
*, sector_t
, Sector
*);
1160 static inline void put_dev_sector(Sector p
)
1162 page_cache_release(p
.v
);
1166 int kblockd_schedule_work(struct request_queue
*q
, struct work_struct
*work
);
1168 #ifdef CONFIG_BLK_CGROUP
1170 * This should not be using sched_clock(). A real patch is in progress
1171 * to fix this up, until that is in place we need to disable preemption
1172 * around sched_clock() in this function and set_io_start_time_ns().
1174 static inline void set_start_time_ns(struct request
*req
)
1177 req
->start_time_ns
= sched_clock();
1181 static inline void set_io_start_time_ns(struct request
*req
)
1184 req
->io_start_time_ns
= sched_clock();
1188 static inline uint64_t rq_start_time_ns(struct request
*req
)
1190 return req
->start_time_ns
;
1193 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1195 return req
->io_start_time_ns
;
1198 static inline void set_start_time_ns(struct request
*req
) {}
1199 static inline void set_io_start_time_ns(struct request
*req
) {}
1200 static inline uint64_t rq_start_time_ns(struct request
*req
)
1204 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1210 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1211 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1212 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1213 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1215 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1217 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1218 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1220 struct blk_integrity_exchg
{
1224 unsigned int data_size
;
1225 unsigned short sector_size
;
1226 const char *disk_name
;
1229 typedef void (integrity_gen_fn
) (struct blk_integrity_exchg
*);
1230 typedef int (integrity_vrfy_fn
) (struct blk_integrity_exchg
*);
1231 typedef void (integrity_set_tag_fn
) (void *, void *, unsigned int);
1232 typedef void (integrity_get_tag_fn
) (void *, void *, unsigned int);
1234 struct blk_integrity
{
1235 integrity_gen_fn
*generate_fn
;
1236 integrity_vrfy_fn
*verify_fn
;
1237 integrity_set_tag_fn
*set_tag_fn
;
1238 integrity_get_tag_fn
*get_tag_fn
;
1240 unsigned short flags
;
1241 unsigned short tuple_size
;
1242 unsigned short sector_size
;
1243 unsigned short tag_size
;
1247 struct kobject kobj
;
1250 extern bool blk_integrity_is_initialized(struct gendisk
*);
1251 extern int blk_integrity_register(struct gendisk
*, struct blk_integrity
*);
1252 extern void blk_integrity_unregister(struct gendisk
*);
1253 extern int blk_integrity_compare(struct gendisk
*, struct gendisk
*);
1254 extern int blk_rq_map_integrity_sg(struct request_queue
*, struct bio
*,
1255 struct scatterlist
*);
1256 extern int blk_rq_count_integrity_sg(struct request_queue
*, struct bio
*);
1257 extern int blk_integrity_merge_rq(struct request_queue
*, struct request
*,
1259 extern int blk_integrity_merge_bio(struct request_queue
*, struct request
*,
1263 struct blk_integrity
*bdev_get_integrity(struct block_device
*bdev
)
1265 return bdev
->bd_disk
->integrity
;
1268 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1270 return disk
->integrity
;
1273 static inline int blk_integrity_rq(struct request
*rq
)
1275 if (rq
->bio
== NULL
)
1278 return bio_integrity(rq
->bio
);
1281 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1284 q
->limits
.max_integrity_segments
= segs
;
1287 static inline unsigned short
1288 queue_max_integrity_segments(struct request_queue
*q
)
1290 return q
->limits
.max_integrity_segments
;
1293 #else /* CONFIG_BLK_DEV_INTEGRITY */
1296 struct block_device
;
1298 struct blk_integrity
;
1300 static inline int blk_integrity_rq(struct request
*rq
)
1304 static inline int blk_rq_count_integrity_sg(struct request_queue
*q
,
1309 static inline int blk_rq_map_integrity_sg(struct request_queue
*q
,
1311 struct scatterlist
*s
)
1315 static inline struct blk_integrity
*bdev_get_integrity(struct block_device
*b
)
1319 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1323 static inline int blk_integrity_compare(struct gendisk
*a
, struct gendisk
*b
)
1327 static inline int blk_integrity_register(struct gendisk
*d
,
1328 struct blk_integrity
*b
)
1332 static inline void blk_integrity_unregister(struct gendisk
*d
)
1335 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1339 static inline unsigned short queue_max_integrity_segments(struct request_queue
*q
)
1343 static inline int blk_integrity_merge_rq(struct request_queue
*rq
,
1349 static inline int blk_integrity_merge_bio(struct request_queue
*rq
,
1355 static inline bool blk_integrity_is_initialized(struct gendisk
*g
)
1360 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1362 struct block_device_operations
{
1363 int (*open
) (struct block_device
*, fmode_t
);
1364 int (*release
) (struct gendisk
*, fmode_t
);
1365 int (*ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1366 int (*compat_ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1367 int (*direct_access
) (struct block_device
*, sector_t
,
1368 void **, unsigned long *);
1369 unsigned int (*check_events
) (struct gendisk
*disk
,
1370 unsigned int clearing
);
1371 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1372 int (*media_changed
) (struct gendisk
*);
1373 void (*unlock_native_capacity
) (struct gendisk
*);
1374 int (*revalidate_disk
) (struct gendisk
*);
1375 int (*getgeo
)(struct block_device
*, struct hd_geometry
*);
1376 /* this callback is with swap_lock and sometimes page table lock held */
1377 void (*swap_slot_free_notify
) (struct block_device
*, unsigned long);
1378 struct module
*owner
;
1381 extern int __blkdev_driver_ioctl(struct block_device
*, fmode_t
, unsigned int,
1383 #else /* CONFIG_BLOCK */
1385 * stubs for when the block layer is configured out
1387 #define buffer_heads_over_limit 0
1389 static inline long nr_blockdev_pages(void)
1397 static inline void blk_start_plug(struct blk_plug
*plug
)
1401 static inline void blk_finish_plug(struct blk_plug
*plug
)
1405 static inline void blk_flush_plug(struct task_struct
*task
)
1409 static inline void blk_schedule_flush_plug(struct task_struct
*task
)
1414 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
1419 #endif /* CONFIG_BLOCK */