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/module.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>
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
;
124 struct gendisk
*rq_disk
;
125 struct hd_struct
*part
;
126 unsigned long start_time
;
127 #ifdef CONFIG_BLK_CGROUP
128 unsigned long long start_time_ns
;
129 unsigned long long io_start_time_ns
; /* when passed to hardware */
131 /* Number of scatter-gather DMA addr+len pairs after
132 * physical address coalescing is performed.
134 unsigned short nr_phys_segments
;
135 #if defined(CONFIG_BLK_DEV_INTEGRITY)
136 unsigned short nr_integrity_segments
;
139 unsigned short ioprio
;
143 void *special
; /* opaque pointer available for LLD use */
144 char *buffer
; /* kaddr of the current segment if available */
150 * when request is used as a packet command carrier
152 unsigned char __cmd
[BLK_MAX_CDB
];
154 unsigned short cmd_len
;
156 unsigned int extra_len
; /* length of alignment and padding */
157 unsigned int sense_len
;
158 unsigned int resid_len
; /* residual count */
161 unsigned long deadline
;
162 struct list_head timeout_list
;
163 unsigned int timeout
;
167 * completion callback.
169 rq_end_io_fn
*end_io
;
173 struct request
*next_rq
;
176 static inline unsigned short req_get_ioprio(struct request
*req
)
182 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
183 * requests. Some step values could eventually be made generic.
185 struct request_pm_state
187 /* PM state machine step value, currently driver specific */
189 /* requested PM state value (S1, S2, S3, S4, ...) */
191 void* data
; /* for driver use */
194 #include <linux/elevator.h>
196 typedef void (request_fn_proc
) (struct request_queue
*q
);
197 typedef int (make_request_fn
) (struct request_queue
*q
, struct bio
*bio
);
198 typedef int (prep_rq_fn
) (struct request_queue
*, struct request
*);
199 typedef void (unprep_rq_fn
) (struct request_queue
*, struct request
*);
202 struct bvec_merge_data
{
203 struct block_device
*bi_bdev
;
208 typedef int (merge_bvec_fn
) (struct request_queue
*, struct bvec_merge_data
*,
210 typedef void (softirq_done_fn
)(struct request
*);
211 typedef int (dma_drain_needed_fn
)(struct request
*);
212 typedef int (lld_busy_fn
) (struct request_queue
*q
);
213 typedef int (bsg_job_fn
) (struct bsg_job
*);
215 enum blk_eh_timer_return
{
221 typedef enum blk_eh_timer_return (rq_timed_out_fn
)(struct request
*);
223 enum blk_queue_state
{
228 struct blk_queue_tag
{
229 struct request
**tag_index
; /* map of busy tags */
230 unsigned long *tag_map
; /* bit map of free/busy tags */
231 int busy
; /* current depth */
232 int max_depth
; /* what we will send to device */
233 int real_max_depth
; /* what the array can hold */
234 atomic_t refcnt
; /* map can be shared */
237 #define BLK_SCSI_MAX_CMDS (256)
238 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
240 struct queue_limits
{
241 unsigned long bounce_pfn
;
242 unsigned long seg_boundary_mask
;
244 unsigned int max_hw_sectors
;
245 unsigned int max_sectors
;
246 unsigned int max_segment_size
;
247 unsigned int physical_block_size
;
248 unsigned int alignment_offset
;
251 unsigned int max_discard_sectors
;
252 unsigned int discard_granularity
;
253 unsigned int discard_alignment
;
255 unsigned short logical_block_size
;
256 unsigned short max_segments
;
257 unsigned short max_integrity_segments
;
259 unsigned char misaligned
;
260 unsigned char discard_misaligned
;
261 unsigned char cluster
;
262 unsigned char discard_zeroes_data
;
265 struct request_queue
{
267 * Together with queue_head for cacheline sharing
269 struct list_head queue_head
;
270 struct request
*last_merge
;
271 struct elevator_queue
*elevator
;
274 * the queue request freelist, one for reads and one for writes
276 struct request_list rq
;
278 request_fn_proc
*request_fn
;
279 make_request_fn
*make_request_fn
;
280 prep_rq_fn
*prep_rq_fn
;
281 unprep_rq_fn
*unprep_rq_fn
;
282 merge_bvec_fn
*merge_bvec_fn
;
283 softirq_done_fn
*softirq_done_fn
;
284 rq_timed_out_fn
*rq_timed_out_fn
;
285 dma_drain_needed_fn
*dma_drain_needed
;
286 lld_busy_fn
*lld_busy_fn
;
289 * Dispatch queue sorting
292 struct request
*boundary_rq
;
295 * Delayed queue handling
297 struct delayed_work delay_work
;
299 struct backing_dev_info backing_dev_info
;
302 * The queue owner gets to use this for whatever they like.
303 * ll_rw_blk doesn't touch it.
308 * various queue flags, see QUEUE_* below
310 unsigned long queue_flags
;
313 * queue needs bounce pages for pages above this limit
318 * protects queue structures from reentrancy. ->__queue_lock should
319 * _never_ be used directly, it is queue private. always use
322 spinlock_t __queue_lock
;
323 spinlock_t
*queue_lock
;
333 unsigned long nr_requests
; /* Max # of requests */
334 unsigned int nr_congestion_on
;
335 unsigned int nr_congestion_off
;
336 unsigned int nr_batching
;
338 unsigned int dma_drain_size
;
339 void *dma_drain_buffer
;
340 unsigned int dma_pad_mask
;
341 unsigned int dma_alignment
;
343 struct blk_queue_tag
*queue_tags
;
344 struct list_head tag_busy_list
;
346 unsigned int nr_sorted
;
347 unsigned int in_flight
[2];
349 unsigned int rq_timeout
;
350 struct timer_list timeout
;
351 struct list_head timeout_list
;
353 struct queue_limits limits
;
358 unsigned int sg_timeout
;
359 unsigned int sg_reserved_size
;
361 #ifdef CONFIG_BLK_DEV_IO_TRACE
362 struct blk_trace
*blk_trace
;
365 * for flush operations
367 unsigned int flush_flags
;
368 unsigned int flush_not_queueable
:1;
369 unsigned int flush_queue_delayed
:1;
370 unsigned int flush_pending_idx
:1;
371 unsigned int flush_running_idx
:1;
372 unsigned long flush_pending_since
;
373 struct list_head flush_queue
[2];
374 struct list_head flush_data_in_flight
;
375 struct request flush_rq
;
377 struct mutex sysfs_lock
;
379 #if defined(CONFIG_BLK_DEV_BSG)
380 bsg_job_fn
*bsg_job_fn
;
382 struct bsg_class_device bsg_dev
;
385 #ifdef CONFIG_BLK_DEV_THROTTLING
387 struct throtl_data
*td
;
391 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
392 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
393 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
394 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
395 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
396 #define QUEUE_FLAG_ELVSWITCH 6 /* don't use elevator, just do FIFO */
397 #define QUEUE_FLAG_BIDI 7 /* queue supports bidi requests */
398 #define QUEUE_FLAG_NOMERGES 8 /* disable merge attempts */
399 #define QUEUE_FLAG_SAME_COMP 9 /* complete on same CPU-group */
400 #define QUEUE_FLAG_FAIL_IO 10 /* fake timeout */
401 #define QUEUE_FLAG_STACKABLE 11 /* supports request stacking */
402 #define QUEUE_FLAG_NONROT 12 /* non-rotational device (SSD) */
403 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
404 #define QUEUE_FLAG_IO_STAT 13 /* do IO stats */
405 #define QUEUE_FLAG_DISCARD 14 /* supports DISCARD */
406 #define QUEUE_FLAG_NOXMERGES 15 /* No extended merges */
407 #define QUEUE_FLAG_ADD_RANDOM 16 /* Contributes to random pool */
408 #define QUEUE_FLAG_SECDISCARD 17 /* supports SECDISCARD */
409 #define QUEUE_FLAG_SAME_FORCE 18 /* force complete on same CPU */
411 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
412 (1 << QUEUE_FLAG_STACKABLE) | \
413 (1 << QUEUE_FLAG_SAME_COMP) | \
414 (1 << QUEUE_FLAG_ADD_RANDOM))
416 static inline int queue_is_locked(struct request_queue
*q
)
419 spinlock_t
*lock
= q
->queue_lock
;
420 return lock
&& spin_is_locked(lock
);
426 static inline void queue_flag_set_unlocked(unsigned int flag
,
427 struct request_queue
*q
)
429 __set_bit(flag
, &q
->queue_flags
);
432 static inline int queue_flag_test_and_clear(unsigned int flag
,
433 struct request_queue
*q
)
435 WARN_ON_ONCE(!queue_is_locked(q
));
437 if (test_bit(flag
, &q
->queue_flags
)) {
438 __clear_bit(flag
, &q
->queue_flags
);
445 static inline int queue_flag_test_and_set(unsigned int flag
,
446 struct request_queue
*q
)
448 WARN_ON_ONCE(!queue_is_locked(q
));
450 if (!test_bit(flag
, &q
->queue_flags
)) {
451 __set_bit(flag
, &q
->queue_flags
);
458 static inline void queue_flag_set(unsigned int flag
, struct request_queue
*q
)
460 WARN_ON_ONCE(!queue_is_locked(q
));
461 __set_bit(flag
, &q
->queue_flags
);
464 static inline void queue_flag_clear_unlocked(unsigned int flag
,
465 struct request_queue
*q
)
467 __clear_bit(flag
, &q
->queue_flags
);
470 static inline int queue_in_flight(struct request_queue
*q
)
472 return q
->in_flight
[0] + q
->in_flight
[1];
475 static inline void queue_flag_clear(unsigned int flag
, struct request_queue
*q
)
477 WARN_ON_ONCE(!queue_is_locked(q
));
478 __clear_bit(flag
, &q
->queue_flags
);
481 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
482 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
483 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
484 #define blk_queue_noxmerges(q) \
485 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
486 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
487 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
488 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
489 #define blk_queue_stackable(q) \
490 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
491 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
492 #define blk_queue_secdiscard(q) (blk_queue_discard(q) && \
493 test_bit(QUEUE_FLAG_SECDISCARD, &(q)->queue_flags))
495 #define blk_noretry_request(rq) \
496 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
497 REQ_FAILFAST_DRIVER))
499 #define blk_account_rq(rq) \
500 (((rq)->cmd_flags & REQ_STARTED) && \
501 ((rq)->cmd_type == REQ_TYPE_FS || \
502 ((rq)->cmd_flags & REQ_DISCARD)))
504 #define blk_pm_request(rq) \
505 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
506 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
508 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
509 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
510 /* rq->queuelist of dequeued request must be list_empty() */
511 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
513 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
515 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
517 static inline unsigned int blk_queue_cluster(struct request_queue
*q
)
519 return q
->limits
.cluster
;
523 * We regard a request as sync, if either a read or a sync write
525 static inline bool rw_is_sync(unsigned int rw_flags
)
527 return !(rw_flags
& REQ_WRITE
) || (rw_flags
& REQ_SYNC
);
530 static inline bool rq_is_sync(struct request
*rq
)
532 return rw_is_sync(rq
->cmd_flags
);
535 static inline int blk_queue_full(struct request_queue
*q
, int sync
)
538 return test_bit(QUEUE_FLAG_SYNCFULL
, &q
->queue_flags
);
539 return test_bit(QUEUE_FLAG_ASYNCFULL
, &q
->queue_flags
);
542 static inline void blk_set_queue_full(struct request_queue
*q
, int sync
)
545 queue_flag_set(QUEUE_FLAG_SYNCFULL
, q
);
547 queue_flag_set(QUEUE_FLAG_ASYNCFULL
, q
);
550 static inline void blk_clear_queue_full(struct request_queue
*q
, int sync
)
553 queue_flag_clear(QUEUE_FLAG_SYNCFULL
, q
);
555 queue_flag_clear(QUEUE_FLAG_ASYNCFULL
, q
);
560 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
561 * it already be started by driver.
563 #define RQ_NOMERGE_FLAGS \
564 (REQ_NOMERGE | REQ_STARTED | REQ_SOFTBARRIER | REQ_FLUSH | REQ_FUA)
565 #define rq_mergeable(rq) \
566 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
567 (((rq)->cmd_flags & REQ_DISCARD) || \
568 (rq)->cmd_type == REQ_TYPE_FS))
571 * q->prep_rq_fn return values
573 #define BLKPREP_OK 0 /* serve it */
574 #define BLKPREP_KILL 1 /* fatal error, kill */
575 #define BLKPREP_DEFER 2 /* leave on queue */
577 extern unsigned long blk_max_low_pfn
, blk_max_pfn
;
580 * standard bounce addresses:
582 * BLK_BOUNCE_HIGH : bounce all highmem pages
583 * BLK_BOUNCE_ANY : don't bounce anything
584 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
587 #if BITS_PER_LONG == 32
588 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
590 #define BLK_BOUNCE_HIGH -1ULL
592 #define BLK_BOUNCE_ANY (-1ULL)
593 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
596 * default timeout for SG_IO if none specified
598 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
599 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
602 extern int init_emergency_isa_pool(void);
603 extern void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
);
605 static inline int init_emergency_isa_pool(void)
609 static inline void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
)
612 #endif /* CONFIG_MMU */
618 unsigned long offset
;
623 struct req_iterator
{
628 /* This should not be used directly - use rq_for_each_segment */
629 #define for_each_bio(_bio) \
630 for (; _bio; _bio = _bio->bi_next)
631 #define __rq_for_each_bio(_bio, rq) \
633 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
635 #define rq_for_each_segment(bvl, _rq, _iter) \
636 __rq_for_each_bio(_iter.bio, _rq) \
637 bio_for_each_segment(bvl, _iter.bio, _iter.i)
639 #define rq_iter_last(rq, _iter) \
640 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
642 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
643 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
645 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
646 extern void rq_flush_dcache_pages(struct request
*rq
);
648 static inline void rq_flush_dcache_pages(struct request
*rq
)
653 extern int blk_register_queue(struct gendisk
*disk
);
654 extern void blk_unregister_queue(struct gendisk
*disk
);
655 extern void generic_make_request(struct bio
*bio
);
656 extern void blk_rq_init(struct request_queue
*q
, struct request
*rq
);
657 extern void blk_put_request(struct request
*);
658 extern void __blk_put_request(struct request_queue
*, struct request
*);
659 extern struct request
*blk_get_request(struct request_queue
*, int, gfp_t
);
660 extern struct request
*blk_make_request(struct request_queue
*, struct bio
*,
662 extern void blk_insert_request(struct request_queue
*, struct request
*, int, void *);
663 extern void blk_requeue_request(struct request_queue
*, struct request
*);
664 extern void blk_add_request_payload(struct request
*rq
, struct page
*page
,
666 extern int blk_rq_check_limits(struct request_queue
*q
, struct request
*rq
);
667 extern int blk_lld_busy(struct request_queue
*q
);
668 extern int blk_rq_prep_clone(struct request
*rq
, struct request
*rq_src
,
669 struct bio_set
*bs
, gfp_t gfp_mask
,
670 int (*bio_ctr
)(struct bio
*, struct bio
*, void *),
672 extern void blk_rq_unprep_clone(struct request
*rq
);
673 extern int blk_insert_cloned_request(struct request_queue
*q
,
675 extern void blk_delay_queue(struct request_queue
*, unsigned long);
676 extern void blk_recount_segments(struct request_queue
*, struct bio
*);
677 extern int scsi_cmd_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
678 unsigned int, void __user
*);
679 extern int sg_scsi_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
680 struct scsi_ioctl_command __user
*);
683 * A queue has just exitted congestion. Note this in the global counter of
684 * congested queues, and wake up anyone who was waiting for requests to be
687 static inline void blk_clear_queue_congested(struct request_queue
*q
, int sync
)
689 clear_bdi_congested(&q
->backing_dev_info
, sync
);
693 * A queue has just entered congestion. Flag that in the queue's VM-visible
694 * state flags and increment the global gounter of congested queues.
696 static inline void blk_set_queue_congested(struct request_queue
*q
, int sync
)
698 set_bdi_congested(&q
->backing_dev_info
, sync
);
701 extern void blk_start_queue(struct request_queue
*q
);
702 extern void blk_stop_queue(struct request_queue
*q
);
703 extern void blk_sync_queue(struct request_queue
*q
);
704 extern void __blk_stop_queue(struct request_queue
*q
);
705 extern void __blk_run_queue(struct request_queue
*q
);
706 extern void blk_run_queue(struct request_queue
*);
707 extern void blk_run_queue_async(struct request_queue
*q
);
708 extern int blk_rq_map_user(struct request_queue
*, struct request
*,
709 struct rq_map_data
*, void __user
*, unsigned long,
711 extern int blk_rq_unmap_user(struct bio
*);
712 extern int blk_rq_map_kern(struct request_queue
*, struct request
*, void *, unsigned int, gfp_t
);
713 extern int blk_rq_map_user_iov(struct request_queue
*, struct request
*,
714 struct rq_map_data
*, struct sg_iovec
*, int,
715 unsigned int, gfp_t
);
716 extern int blk_execute_rq(struct request_queue
*, struct gendisk
*,
717 struct request
*, int);
718 extern void blk_execute_rq_nowait(struct request_queue
*, struct gendisk
*,
719 struct request
*, int, rq_end_io_fn
*);
721 static inline struct request_queue
*bdev_get_queue(struct block_device
*bdev
)
723 return bdev
->bd_disk
->queue
;
727 * blk_rq_pos() : the current sector
728 * blk_rq_bytes() : bytes left in the entire request
729 * blk_rq_cur_bytes() : bytes left in the current segment
730 * blk_rq_err_bytes() : bytes left till the next error boundary
731 * blk_rq_sectors() : sectors left in the entire request
732 * blk_rq_cur_sectors() : sectors left in the current segment
734 static inline sector_t
blk_rq_pos(const struct request
*rq
)
739 static inline unsigned int blk_rq_bytes(const struct request
*rq
)
741 return rq
->__data_len
;
744 static inline int blk_rq_cur_bytes(const struct request
*rq
)
746 return rq
->bio
? bio_cur_bytes(rq
->bio
) : 0;
749 extern unsigned int blk_rq_err_bytes(const struct request
*rq
);
751 static inline unsigned int blk_rq_sectors(const struct request
*rq
)
753 return blk_rq_bytes(rq
) >> 9;
756 static inline unsigned int blk_rq_cur_sectors(const struct request
*rq
)
758 return blk_rq_cur_bytes(rq
) >> 9;
762 * Request issue related functions.
764 extern struct request
*blk_peek_request(struct request_queue
*q
);
765 extern void blk_start_request(struct request
*rq
);
766 extern struct request
*blk_fetch_request(struct request_queue
*q
);
769 * Request completion related functions.
771 * blk_update_request() completes given number of bytes and updates
772 * the request without completing it.
774 * blk_end_request() and friends. __blk_end_request() must be called
775 * with the request queue spinlock acquired.
777 * Several drivers define their own end_request and call
778 * blk_end_request() for parts of the original function.
779 * This prevents code duplication in drivers.
781 extern bool blk_update_request(struct request
*rq
, int error
,
782 unsigned int nr_bytes
);
783 extern bool blk_end_request(struct request
*rq
, int error
,
784 unsigned int nr_bytes
);
785 extern void blk_end_request_all(struct request
*rq
, int error
);
786 extern bool blk_end_request_cur(struct request
*rq
, int error
);
787 extern bool blk_end_request_err(struct request
*rq
, int error
);
788 extern bool __blk_end_request(struct request
*rq
, int error
,
789 unsigned int nr_bytes
);
790 extern void __blk_end_request_all(struct request
*rq
, int error
);
791 extern bool __blk_end_request_cur(struct request
*rq
, int error
);
792 extern bool __blk_end_request_err(struct request
*rq
, int error
);
794 extern void blk_complete_request(struct request
*);
795 extern void __blk_complete_request(struct request
*);
796 extern void blk_abort_request(struct request
*);
797 extern void blk_abort_queue(struct request_queue
*);
798 extern void blk_unprep_request(struct request
*);
801 * Access functions for manipulating queue properties
803 extern struct request_queue
*blk_init_queue_node(request_fn_proc
*rfn
,
804 spinlock_t
*lock
, int node_id
);
805 extern struct request_queue
*blk_init_allocated_queue_node(struct request_queue
*,
807 spinlock_t
*, int node_id
);
808 extern struct request_queue
*blk_init_queue(request_fn_proc
*, spinlock_t
*);
809 extern struct request_queue
*blk_init_allocated_queue(struct request_queue
*,
810 request_fn_proc
*, spinlock_t
*);
811 extern void blk_cleanup_queue(struct request_queue
*);
812 extern void blk_queue_make_request(struct request_queue
*, make_request_fn
*);
813 extern void blk_queue_bounce_limit(struct request_queue
*, u64
);
814 extern void blk_limits_max_hw_sectors(struct queue_limits
*, unsigned int);
815 extern void blk_queue_max_hw_sectors(struct request_queue
*, unsigned int);
816 extern void blk_queue_max_segments(struct request_queue
*, unsigned short);
817 extern void blk_queue_max_segment_size(struct request_queue
*, unsigned int);
818 extern void blk_queue_max_discard_sectors(struct request_queue
*q
,
819 unsigned int max_discard_sectors
);
820 extern void blk_queue_logical_block_size(struct request_queue
*, unsigned short);
821 extern void blk_queue_physical_block_size(struct request_queue
*, unsigned int);
822 extern void blk_queue_alignment_offset(struct request_queue
*q
,
823 unsigned int alignment
);
824 extern void blk_limits_io_min(struct queue_limits
*limits
, unsigned int min
);
825 extern void blk_queue_io_min(struct request_queue
*q
, unsigned int min
);
826 extern void blk_limits_io_opt(struct queue_limits
*limits
, unsigned int opt
);
827 extern void blk_queue_io_opt(struct request_queue
*q
, unsigned int opt
);
828 extern void blk_set_default_limits(struct queue_limits
*lim
);
829 extern int blk_stack_limits(struct queue_limits
*t
, struct queue_limits
*b
,
831 extern int bdev_stack_limits(struct queue_limits
*t
, struct block_device
*bdev
,
833 extern void disk_stack_limits(struct gendisk
*disk
, struct block_device
*bdev
,
835 extern void blk_queue_stack_limits(struct request_queue
*t
, struct request_queue
*b
);
836 extern void blk_queue_dma_pad(struct request_queue
*, unsigned int);
837 extern void blk_queue_update_dma_pad(struct request_queue
*, unsigned int);
838 extern int blk_queue_dma_drain(struct request_queue
*q
,
839 dma_drain_needed_fn
*dma_drain_needed
,
840 void *buf
, unsigned int size
);
841 extern void blk_queue_lld_busy(struct request_queue
*q
, lld_busy_fn
*fn
);
842 extern void blk_queue_segment_boundary(struct request_queue
*, unsigned long);
843 extern void blk_queue_prep_rq(struct request_queue
*, prep_rq_fn
*pfn
);
844 extern void blk_queue_unprep_rq(struct request_queue
*, unprep_rq_fn
*ufn
);
845 extern void blk_queue_merge_bvec(struct request_queue
*, merge_bvec_fn
*);
846 extern void blk_queue_dma_alignment(struct request_queue
*, int);
847 extern void blk_queue_update_dma_alignment(struct request_queue
*, int);
848 extern void blk_queue_softirq_done(struct request_queue
*, softirq_done_fn
*);
849 extern void blk_queue_rq_timed_out(struct request_queue
*, rq_timed_out_fn
*);
850 extern void blk_queue_rq_timeout(struct request_queue
*, unsigned int);
851 extern void blk_queue_flush(struct request_queue
*q
, unsigned int flush
);
852 extern void blk_queue_flush_queueable(struct request_queue
*q
, bool queueable
);
853 extern struct backing_dev_info
*blk_get_backing_dev_info(struct block_device
*bdev
);
855 extern int blk_rq_map_sg(struct request_queue
*, struct request
*, struct scatterlist
*);
856 extern void blk_dump_rq_flags(struct request
*, char *);
857 extern long nr_blockdev_pages(void);
859 int blk_get_queue(struct request_queue
*);
860 struct request_queue
*blk_alloc_queue(gfp_t
);
861 struct request_queue
*blk_alloc_queue_node(gfp_t
, int);
862 extern void blk_put_queue(struct request_queue
*);
865 * Note: Code in between changing the blk_plug list/cb_list or element of such
866 * lists is preemptable, but such code can't do sleep (or be very careful),
867 * otherwise data is corrupted. For details, please check schedule() where
868 * blk_schedule_flush_plug() is called.
872 struct list_head list
;
873 struct list_head cb_list
;
874 unsigned int should_sort
;
877 #define BLK_MAX_REQUEST_COUNT 16
880 struct list_head list
;
881 void (*callback
)(struct blk_plug_cb
*);
884 extern void blk_start_plug(struct blk_plug
*);
885 extern void blk_finish_plug(struct blk_plug
*);
886 extern void blk_flush_plug_list(struct blk_plug
*, bool);
888 static inline void blk_flush_plug(struct task_struct
*tsk
)
890 struct blk_plug
*plug
= tsk
->plug
;
893 blk_flush_plug_list(plug
, false);
896 static inline void blk_schedule_flush_plug(struct task_struct
*tsk
)
898 struct blk_plug
*plug
= tsk
->plug
;
901 blk_flush_plug_list(plug
, true);
904 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
906 struct blk_plug
*plug
= tsk
->plug
;
908 return plug
&& (!list_empty(&plug
->list
) || !list_empty(&plug
->cb_list
));
914 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
915 extern int blk_queue_start_tag(struct request_queue
*, struct request
*);
916 extern struct request
*blk_queue_find_tag(struct request_queue
*, int);
917 extern void blk_queue_end_tag(struct request_queue
*, struct request
*);
918 extern int blk_queue_init_tags(struct request_queue
*, int, struct blk_queue_tag
*);
919 extern void blk_queue_free_tags(struct request_queue
*);
920 extern int blk_queue_resize_tags(struct request_queue
*, int);
921 extern void blk_queue_invalidate_tags(struct request_queue
*);
922 extern struct blk_queue_tag
*blk_init_tags(int);
923 extern void blk_free_tags(struct blk_queue_tag
*);
925 static inline struct request
*blk_map_queue_find_tag(struct blk_queue_tag
*bqt
,
928 if (unlikely(bqt
== NULL
|| tag
>= bqt
->real_max_depth
))
930 return bqt
->tag_index
[tag
];
933 #define BLKDEV_DISCARD_SECURE 0x01 /* secure discard */
935 extern int blkdev_issue_flush(struct block_device
*, gfp_t
, sector_t
*);
936 extern int blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
937 sector_t nr_sects
, gfp_t gfp_mask
, unsigned long flags
);
938 extern int blkdev_issue_zeroout(struct block_device
*bdev
, sector_t sector
,
939 sector_t nr_sects
, gfp_t gfp_mask
);
940 static inline int sb_issue_discard(struct super_block
*sb
, sector_t block
,
941 sector_t nr_blocks
, gfp_t gfp_mask
, unsigned long flags
)
943 return blkdev_issue_discard(sb
->s_bdev
, block
<< (sb
->s_blocksize_bits
- 9),
944 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
947 static inline int sb_issue_zeroout(struct super_block
*sb
, sector_t block
,
948 sector_t nr_blocks
, gfp_t gfp_mask
)
950 return blkdev_issue_zeroout(sb
->s_bdev
,
951 block
<< (sb
->s_blocksize_bits
- 9),
952 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
956 extern int blk_verify_command(unsigned char *cmd
, fmode_t has_write_perm
);
958 enum blk_default_limits
{
959 BLK_MAX_SEGMENTS
= 128,
960 BLK_SAFE_MAX_SECTORS
= 255,
961 BLK_DEF_MAX_SECTORS
= 1024,
962 BLK_MAX_SEGMENT_SIZE
= 65536,
963 BLK_SEG_BOUNDARY_MASK
= 0xFFFFFFFFUL
,
966 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
968 static inline unsigned long queue_bounce_pfn(struct request_queue
*q
)
970 return q
->limits
.bounce_pfn
;
973 static inline unsigned long queue_segment_boundary(struct request_queue
*q
)
975 return q
->limits
.seg_boundary_mask
;
978 static inline unsigned int queue_max_sectors(struct request_queue
*q
)
980 return q
->limits
.max_sectors
;
983 static inline unsigned int queue_max_hw_sectors(struct request_queue
*q
)
985 return q
->limits
.max_hw_sectors
;
988 static inline unsigned short queue_max_segments(struct request_queue
*q
)
990 return q
->limits
.max_segments
;
993 static inline unsigned int queue_max_segment_size(struct request_queue
*q
)
995 return q
->limits
.max_segment_size
;
998 static inline unsigned short queue_logical_block_size(struct request_queue
*q
)
1002 if (q
&& q
->limits
.logical_block_size
)
1003 retval
= q
->limits
.logical_block_size
;
1008 static inline unsigned short bdev_logical_block_size(struct block_device
*bdev
)
1010 return queue_logical_block_size(bdev_get_queue(bdev
));
1013 static inline unsigned int queue_physical_block_size(struct request_queue
*q
)
1015 return q
->limits
.physical_block_size
;
1018 static inline unsigned int bdev_physical_block_size(struct block_device
*bdev
)
1020 return queue_physical_block_size(bdev_get_queue(bdev
));
1023 static inline unsigned int queue_io_min(struct request_queue
*q
)
1025 return q
->limits
.io_min
;
1028 static inline int bdev_io_min(struct block_device
*bdev
)
1030 return queue_io_min(bdev_get_queue(bdev
));
1033 static inline unsigned int queue_io_opt(struct request_queue
*q
)
1035 return q
->limits
.io_opt
;
1038 static inline int bdev_io_opt(struct block_device
*bdev
)
1040 return queue_io_opt(bdev_get_queue(bdev
));
1043 static inline int queue_alignment_offset(struct request_queue
*q
)
1045 if (q
->limits
.misaligned
)
1048 return q
->limits
.alignment_offset
;
1051 static inline int queue_limit_alignment_offset(struct queue_limits
*lim
, sector_t sector
)
1053 unsigned int granularity
= max(lim
->physical_block_size
, lim
->io_min
);
1054 unsigned int alignment
= (sector
<< 9) & (granularity
- 1);
1056 return (granularity
+ lim
->alignment_offset
- alignment
)
1057 & (granularity
- 1);
1060 static inline int bdev_alignment_offset(struct block_device
*bdev
)
1062 struct request_queue
*q
= bdev_get_queue(bdev
);
1064 if (q
->limits
.misaligned
)
1067 if (bdev
!= bdev
->bd_contains
)
1068 return bdev
->bd_part
->alignment_offset
;
1070 return q
->limits
.alignment_offset
;
1073 static inline int queue_discard_alignment(struct request_queue
*q
)
1075 if (q
->limits
.discard_misaligned
)
1078 return q
->limits
.discard_alignment
;
1081 static inline int queue_limit_discard_alignment(struct queue_limits
*lim
, sector_t sector
)
1083 unsigned int alignment
= (sector
<< 9) & (lim
->discard_granularity
- 1);
1085 if (!lim
->max_discard_sectors
)
1088 return (lim
->discard_granularity
+ lim
->discard_alignment
- alignment
)
1089 & (lim
->discard_granularity
- 1);
1092 static inline unsigned int queue_discard_zeroes_data(struct request_queue
*q
)
1094 if (q
->limits
.max_discard_sectors
&& q
->limits
.discard_zeroes_data
== 1)
1100 static inline unsigned int bdev_discard_zeroes_data(struct block_device
*bdev
)
1102 return queue_discard_zeroes_data(bdev_get_queue(bdev
));
1105 static inline int queue_dma_alignment(struct request_queue
*q
)
1107 return q
? q
->dma_alignment
: 511;
1110 static inline int blk_rq_aligned(struct request_queue
*q
, unsigned long addr
,
1113 unsigned int alignment
= queue_dma_alignment(q
) | q
->dma_pad_mask
;
1114 return !(addr
& alignment
) && !(len
& alignment
);
1117 /* assumes size > 256 */
1118 static inline unsigned int blksize_bits(unsigned int size
)
1120 unsigned int bits
= 8;
1124 } while (size
> 256);
1128 static inline unsigned int block_size(struct block_device
*bdev
)
1130 return bdev
->bd_block_size
;
1133 static inline bool queue_flush_queueable(struct request_queue
*q
)
1135 return !q
->flush_not_queueable
;
1138 typedef struct {struct page
*v
;} Sector
;
1140 unsigned char *read_dev_sector(struct block_device
*, sector_t
, Sector
*);
1142 static inline void put_dev_sector(Sector p
)
1144 page_cache_release(p
.v
);
1148 int kblockd_schedule_work(struct request_queue
*q
, struct work_struct
*work
);
1150 #ifdef CONFIG_BLK_CGROUP
1152 * This should not be using sched_clock(). A real patch is in progress
1153 * to fix this up, until that is in place we need to disable preemption
1154 * around sched_clock() in this function and set_io_start_time_ns().
1156 static inline void set_start_time_ns(struct request
*req
)
1159 req
->start_time_ns
= sched_clock();
1163 static inline void set_io_start_time_ns(struct request
*req
)
1166 req
->io_start_time_ns
= sched_clock();
1170 static inline uint64_t rq_start_time_ns(struct request
*req
)
1172 return req
->start_time_ns
;
1175 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1177 return req
->io_start_time_ns
;
1180 static inline void set_start_time_ns(struct request
*req
) {}
1181 static inline void set_io_start_time_ns(struct request
*req
) {}
1182 static inline uint64_t rq_start_time_ns(struct request
*req
)
1186 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1192 #ifdef CONFIG_BLK_DEV_THROTTLING
1193 extern int blk_throtl_init(struct request_queue
*q
);
1194 extern void blk_throtl_exit(struct request_queue
*q
);
1195 extern int blk_throtl_bio(struct request_queue
*q
, struct bio
**bio
);
1196 #else /* CONFIG_BLK_DEV_THROTTLING */
1197 static inline int blk_throtl_bio(struct request_queue
*q
, struct bio
**bio
)
1202 static inline int blk_throtl_init(struct request_queue
*q
) { return 0; }
1203 static inline int blk_throtl_exit(struct request_queue
*q
) { return 0; }
1204 #endif /* CONFIG_BLK_DEV_THROTTLING */
1206 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1207 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1208 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1209 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1211 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1213 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1214 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1216 struct blk_integrity_exchg
{
1220 unsigned int data_size
;
1221 unsigned short sector_size
;
1222 const char *disk_name
;
1225 typedef void (integrity_gen_fn
) (struct blk_integrity_exchg
*);
1226 typedef int (integrity_vrfy_fn
) (struct blk_integrity_exchg
*);
1227 typedef void (integrity_set_tag_fn
) (void *, void *, unsigned int);
1228 typedef void (integrity_get_tag_fn
) (void *, void *, unsigned int);
1230 struct blk_integrity
{
1231 integrity_gen_fn
*generate_fn
;
1232 integrity_vrfy_fn
*verify_fn
;
1233 integrity_set_tag_fn
*set_tag_fn
;
1234 integrity_get_tag_fn
*get_tag_fn
;
1236 unsigned short flags
;
1237 unsigned short tuple_size
;
1238 unsigned short sector_size
;
1239 unsigned short tag_size
;
1243 struct kobject kobj
;
1246 extern bool blk_integrity_is_initialized(struct gendisk
*);
1247 extern int blk_integrity_register(struct gendisk
*, struct blk_integrity
*);
1248 extern void blk_integrity_unregister(struct gendisk
*);
1249 extern int blk_integrity_compare(struct gendisk
*, struct gendisk
*);
1250 extern int blk_rq_map_integrity_sg(struct request_queue
*, struct bio
*,
1251 struct scatterlist
*);
1252 extern int blk_rq_count_integrity_sg(struct request_queue
*, struct bio
*);
1253 extern int blk_integrity_merge_rq(struct request_queue
*, struct request
*,
1255 extern int blk_integrity_merge_bio(struct request_queue
*, struct request
*,
1259 struct blk_integrity
*bdev_get_integrity(struct block_device
*bdev
)
1261 return bdev
->bd_disk
->integrity
;
1264 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1266 return disk
->integrity
;
1269 static inline int blk_integrity_rq(struct request
*rq
)
1271 if (rq
->bio
== NULL
)
1274 return bio_integrity(rq
->bio
);
1277 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1280 q
->limits
.max_integrity_segments
= segs
;
1283 static inline unsigned short
1284 queue_max_integrity_segments(struct request_queue
*q
)
1286 return q
->limits
.max_integrity_segments
;
1289 #else /* CONFIG_BLK_DEV_INTEGRITY */
1291 #define blk_integrity_rq(rq) (0)
1292 #define blk_rq_count_integrity_sg(a, b) (0)
1293 #define blk_rq_map_integrity_sg(a, b, c) (0)
1294 #define bdev_get_integrity(a) (0)
1295 #define blk_get_integrity(a) (0)
1296 #define blk_integrity_compare(a, b) (0)
1297 #define blk_integrity_register(a, b) (0)
1298 #define blk_integrity_unregister(a) do { } while (0)
1299 #define blk_queue_max_integrity_segments(a, b) do { } while (0)
1300 #define queue_max_integrity_segments(a) (0)
1301 #define blk_integrity_merge_rq(a, b, c) (0)
1302 #define blk_integrity_merge_bio(a, b, c) (0)
1303 #define blk_integrity_is_initialized(a) (0)
1305 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1307 struct block_device_operations
{
1308 int (*open
) (struct block_device
*, fmode_t
);
1309 int (*release
) (struct gendisk
*, fmode_t
);
1310 int (*ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1311 int (*compat_ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1312 int (*direct_access
) (struct block_device
*, sector_t
,
1313 void **, unsigned long *);
1314 unsigned int (*check_events
) (struct gendisk
*disk
,
1315 unsigned int clearing
);
1316 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1317 int (*media_changed
) (struct gendisk
*);
1318 void (*unlock_native_capacity
) (struct gendisk
*);
1319 int (*revalidate_disk
) (struct gendisk
*);
1320 int (*getgeo
)(struct block_device
*, struct hd_geometry
*);
1321 /* this callback is with swap_lock and sometimes page table lock held */
1322 void (*swap_slot_free_notify
) (struct block_device
*, unsigned long);
1323 struct module
*owner
;
1326 extern int __blkdev_driver_ioctl(struct block_device
*, fmode_t
, unsigned int,
1328 #else /* CONFIG_BLOCK */
1330 * stubs for when the block layer is configured out
1332 #define buffer_heads_over_limit 0
1334 static inline long nr_blockdev_pages(void)
1342 static inline void blk_start_plug(struct blk_plug
*plug
)
1346 static inline void blk_finish_plug(struct blk_plug
*plug
)
1350 static inline void blk_flush_plug(struct task_struct
*task
)
1354 static inline void blk_schedule_flush_plug(struct task_struct
*task
)
1359 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
1364 #endif /* CONFIG_BLOCK */