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;
366 struct queue_limits limits;
371 unsigned int sg_timeout;
372 unsigned int sg_reserved_size;
374 #ifdef CONFIG_BLK_DEV_IO_TRACE
375 struct blk_trace *blk_trace;
378 * for flush operations
380 unsigned int flush_flags;
381 unsigned int flush_not_queueable:1;
382 unsigned int flush_queue_delayed:1;
383 unsigned int flush_pending_idx:1;
384 unsigned int flush_running_idx:1;
385 unsigned long flush_pending_since;
386 struct list_head flush_queue[2];
387 struct list_head flush_data_in_flight;
388 struct request flush_rq;
390 struct mutex sysfs_lock;
392 #if defined(CONFIG_BLK_DEV_BSG)
393 bsg_job_fn *bsg_job_fn;
395 struct bsg_class_device bsg_dev;
398 #ifdef CONFIG_BLK_DEV_THROTTLING
400 struct throtl_data *td;
402 #ifdef CONFIG_LOCKDEP
403 int ioc_release_depth;
407 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
408 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
409 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
410 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
411 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
412 #define QUEUE_FLAG_ELVSWITCH 6 /* don't use elevator, just do FIFO */
413 #define QUEUE_FLAG_BIDI 7 /* queue supports bidi requests */
414 #define QUEUE_FLAG_NOMERGES 8 /* disable merge attempts */
415 #define QUEUE_FLAG_SAME_COMP 9 /* complete on same CPU-group */
416 #define QUEUE_FLAG_FAIL_IO 10 /* fake timeout */
417 #define QUEUE_FLAG_STACKABLE 11 /* supports request stacking */
418 #define QUEUE_FLAG_NONROT 12 /* non-rotational device (SSD) */
419 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
420 #define QUEUE_FLAG_IO_STAT 13 /* do IO stats */
421 #define QUEUE_FLAG_DISCARD 14 /* supports DISCARD */
422 #define QUEUE_FLAG_NOXMERGES 15 /* No extended merges */
423 #define QUEUE_FLAG_ADD_RANDOM 16 /* Contributes to random pool */
424 #define QUEUE_FLAG_SECDISCARD 17 /* supports SECDISCARD */
425 #define QUEUE_FLAG_SAME_FORCE 18 /* force complete on same CPU */
427 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
428 (1 << QUEUE_FLAG_STACKABLE) | \
429 (1 << QUEUE_FLAG_SAME_COMP) | \
430 (1 << QUEUE_FLAG_ADD_RANDOM))
432 static inline int queue_is_locked(struct request_queue *q)
435 spinlock_t *lock = q->queue_lock;
436 return lock && spin_is_locked(lock);
442 static inline void queue_flag_set_unlocked(unsigned int flag,
443 struct request_queue *q)
445 __set_bit(flag, &q->queue_flags);
448 static inline int queue_flag_test_and_clear(unsigned int flag,
449 struct request_queue *q)
451 WARN_ON_ONCE(!queue_is_locked(q));
453 if (test_bit(flag, &q->queue_flags)) {
454 __clear_bit(flag, &q->queue_flags);
461 static inline int queue_flag_test_and_set(unsigned int flag,
462 struct request_queue *q)
464 WARN_ON_ONCE(!queue_is_locked(q));
466 if (!test_bit(flag, &q->queue_flags)) {
467 __set_bit(flag, &q->queue_flags);
474 static inline void queue_flag_set(unsigned int flag, struct request_queue *q)
476 WARN_ON_ONCE(!queue_is_locked(q));
477 __set_bit(flag, &q->queue_flags);
480 static inline void queue_flag_clear_unlocked(unsigned int flag,
481 struct request_queue *q)
483 __clear_bit(flag, &q->queue_flags);
486 static inline int queue_in_flight(struct request_queue *q)
488 return q->in_flight[0] + q->in_flight[1];
491 static inline void queue_flag_clear(unsigned int flag, struct request_queue *q)
493 WARN_ON_ONCE(!queue_is_locked(q));
494 __clear_bit(flag, &q->queue_flags);
497 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
498 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
499 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
500 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
501 #define blk_queue_noxmerges(q) \
502 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
503 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
504 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
505 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
506 #define blk_queue_stackable(q) \
507 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
508 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
509 #define blk_queue_secdiscard(q) (blk_queue_discard(q) && \
510 test_bit(QUEUE_FLAG_SECDISCARD, &(q)->queue_flags))
512 #define blk_noretry_request(rq) \
513 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
514 REQ_FAILFAST_DRIVER))
516 #define blk_account_rq(rq) \
517 (((rq)->cmd_flags & REQ_STARTED) && \
518 ((rq)->cmd_type == REQ_TYPE_FS || \
519 ((rq)->cmd_flags & REQ_DISCARD)))
521 #define blk_pm_request(rq) \
522 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
523 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
525 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
526 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
527 /* rq->queuelist of dequeued request must be list_empty() */
528 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
530 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
532 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
534 static inline unsigned int blk_queue_cluster(struct request_queue *q)
536 return q->limits.cluster;
540 * We regard a request as sync, if either a read or a sync write
542 static inline bool rw_is_sync(unsigned int rw_flags)
544 return !(rw_flags & REQ_WRITE) || (rw_flags & REQ_SYNC);
547 static inline bool rq_is_sync(struct request *rq)
549 return rw_is_sync(rq->cmd_flags);
552 static inline int blk_queue_full(struct request_queue *q, int sync)
555 return test_bit(QUEUE_FLAG_SYNCFULL, &q->queue_flags);
556 return test_bit(QUEUE_FLAG_ASYNCFULL, &q->queue_flags);
559 static inline void blk_set_queue_full(struct request_queue *q, int sync)
562 queue_flag_set(QUEUE_FLAG_SYNCFULL, q);
564 queue_flag_set(QUEUE_FLAG_ASYNCFULL, q);
567 static inline void blk_clear_queue_full(struct request_queue *q, int sync)
570 queue_flag_clear(QUEUE_FLAG_SYNCFULL, q);
572 queue_flag_clear(QUEUE_FLAG_ASYNCFULL, q);
577 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
578 * it already be started by driver.
580 #define RQ_NOMERGE_FLAGS \
581 (REQ_NOMERGE | REQ_STARTED | REQ_SOFTBARRIER | REQ_FLUSH | REQ_FUA)
582 #define rq_mergeable(rq) \
583 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
584 (((rq)->cmd_flags & REQ_DISCARD) || \
585 (rq)->cmd_type == REQ_TYPE_FS))
588 * q->prep_rq_fn return values
590 #define BLKPREP_OK 0 /* serve it */
591 #define BLKPREP_KILL 1 /* fatal error, kill */
592 #define BLKPREP_DEFER 2 /* leave on queue */
594 extern unsigned long blk_max_low_pfn, blk_max_pfn;
597 * standard bounce addresses:
599 * BLK_BOUNCE_HIGH : bounce all highmem pages
600 * BLK_BOUNCE_ANY : don't bounce anything
601 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
604 #if BITS_PER_LONG == 32
605 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
607 #define BLK_BOUNCE_HIGH -1ULL
609 #define BLK_BOUNCE_ANY (-1ULL)
610 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
613 * default timeout for SG_IO if none specified
615 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
616 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
619 extern int init_emergency_isa_pool(void);
620 extern void blk_queue_bounce(struct request_queue *q, struct bio **bio);
622 static inline int init_emergency_isa_pool(void)
626 static inline void blk_queue_bounce(struct request_queue *q, struct bio **bio)
629 #endif /* CONFIG_MMU */
635 unsigned long offset;
640 struct req_iterator {
645 /* This should not be used directly - use rq_for_each_segment */
646 #define for_each_bio(_bio) \
647 for (; _bio; _bio = _bio->bi_next)
648 #define __rq_for_each_bio(_bio, rq) \
650 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
652 #define rq_for_each_segment(bvl, _rq, _iter) \
653 __rq_for_each_bio(_iter.bio, _rq) \
654 bio_for_each_segment(bvl, _iter.bio, _iter.i)
656 #define rq_iter_last(rq, _iter) \
657 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
659 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
660 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
662 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
663 extern void rq_flush_dcache_pages(struct request *rq);
665 static inline void rq_flush_dcache_pages(struct request *rq)
670 extern int blk_register_queue(struct gendisk *disk);
671 extern void blk_unregister_queue(struct gendisk *disk);
672 extern void generic_make_request(struct bio *bio);
673 extern void blk_rq_init(struct request_queue *q, struct request *rq);
674 extern void blk_put_request(struct request *);
675 extern void __blk_put_request(struct request_queue *, struct request *);
676 extern struct request *blk_get_request(struct request_queue *, int, gfp_t);
677 extern struct request *blk_make_request(struct request_queue *, struct bio *,
679 extern void blk_requeue_request(struct request_queue *, struct request *);
680 extern void blk_add_request_payload(struct request *rq, struct page *page,
682 extern int blk_rq_check_limits(struct request_queue *q, struct request *rq);
683 extern int blk_lld_busy(struct request_queue *q);
684 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
685 struct bio_set *bs, gfp_t gfp_mask,
686 int (*bio_ctr)(struct bio *, struct bio *, void *),
688 extern void blk_rq_unprep_clone(struct request *rq);
689 extern int blk_insert_cloned_request(struct request_queue *q,
691 extern void blk_delay_queue(struct request_queue *, unsigned long);
692 extern void blk_recount_segments(struct request_queue *, struct bio *);
693 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
694 unsigned int, void __user *);
695 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
696 struct scsi_ioctl_command __user *);
698 extern void blk_queue_bio(struct request_queue *q, struct bio *bio);
701 * A queue has just exitted congestion. Note this in the global counter of
702 * congested queues, and wake up anyone who was waiting for requests to be
705 static inline void blk_clear_queue_congested(struct request_queue *q, int sync)
707 clear_bdi_congested(&q->backing_dev_info, sync);
711 * A queue has just entered congestion. Flag that in the queue's VM-visible
712 * state flags and increment the global gounter of congested queues.
714 static inline void blk_set_queue_congested(struct request_queue *q, int sync)
716 set_bdi_congested(&q->backing_dev_info, sync);
719 extern void blk_start_queue(struct request_queue *q);
720 extern void blk_stop_queue(struct request_queue *q);
721 extern void blk_sync_queue(struct request_queue *q);
722 extern void __blk_stop_queue(struct request_queue *q);
723 extern void __blk_run_queue(struct request_queue *q);
724 extern void blk_run_queue(struct request_queue *);
725 extern void blk_run_queue_async(struct request_queue *q);
726 extern int blk_rq_map_user(struct request_queue *, struct request *,
727 struct rq_map_data *, void __user *, unsigned long,
729 extern int blk_rq_unmap_user(struct bio *);
730 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
731 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
732 struct rq_map_data *, struct sg_iovec *, int,
733 unsigned int, gfp_t);
734 extern int blk_execute_rq(struct request_queue *, struct gendisk *,
735 struct request *, int);
736 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
737 struct request *, int, rq_end_io_fn *);
739 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
741 return bdev->bd_disk->queue;
745 * blk_rq_pos() : the current sector
746 * blk_rq_bytes() : bytes left in the entire request
747 * blk_rq_cur_bytes() : bytes left in the current segment
748 * blk_rq_err_bytes() : bytes left till the next error boundary
749 * blk_rq_sectors() : sectors left in the entire request
750 * blk_rq_cur_sectors() : sectors left in the current segment
752 static inline sector_t blk_rq_pos(const struct request *rq)
757 static inline unsigned int blk_rq_bytes(const struct request *rq)
759 return rq->__data_len;
762 static inline int blk_rq_cur_bytes(const struct request *rq)
764 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
767 extern unsigned int blk_rq_err_bytes(const struct request *rq);
769 static inline unsigned int blk_rq_sectors(const struct request *rq)
771 return blk_rq_bytes(rq) >> 9;
774 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
776 return blk_rq_cur_bytes(rq) >> 9;
780 * Request issue related functions.
782 extern struct request *blk_peek_request(struct request_queue *q);
783 extern void blk_start_request(struct request *rq);
784 extern struct request *blk_fetch_request(struct request_queue *q);
787 * Request completion related functions.
789 * blk_update_request() completes given number of bytes and updates
790 * the request without completing it.
792 * blk_end_request() and friends. __blk_end_request() must be called
793 * with the request queue spinlock acquired.
795 * Several drivers define their own end_request and call
796 * blk_end_request() for parts of the original function.
797 * This prevents code duplication in drivers.
799 extern bool blk_update_request(struct request *rq, int error,
800 unsigned int nr_bytes);
801 extern bool blk_end_request(struct request *rq, int error,
802 unsigned int nr_bytes);
803 extern void blk_end_request_all(struct request *rq, int error);
804 extern bool blk_end_request_cur(struct request *rq, int error);
805 extern bool blk_end_request_err(struct request *rq, int error);
806 extern bool __blk_end_request(struct request *rq, int error,
807 unsigned int nr_bytes);
808 extern void __blk_end_request_all(struct request *rq, int error);
809 extern bool __blk_end_request_cur(struct request *rq, int error);
810 extern bool __blk_end_request_err(struct request *rq, int error);
812 extern void blk_complete_request(struct request *);
813 extern void __blk_complete_request(struct request *);
814 extern void blk_abort_request(struct request *);
815 extern void blk_abort_queue(struct request_queue *);
816 extern void blk_unprep_request(struct request *);
819 * Access functions for manipulating queue properties
821 extern struct request_queue *blk_init_queue_node(request_fn_proc *rfn,
822 spinlock_t *lock, int node_id);
823 extern struct request_queue *blk_init_allocated_queue_node(struct request_queue *,
825 spinlock_t *, int node_id);
826 extern struct request_queue *blk_init_queue(request_fn_proc *, spinlock_t *);
827 extern struct request_queue *blk_init_allocated_queue(struct request_queue *,
828 request_fn_proc *, spinlock_t *);
829 extern void blk_cleanup_queue(struct request_queue *);
830 extern void blk_queue_make_request(struct request_queue *, make_request_fn *);
831 extern void blk_queue_bounce_limit(struct request_queue *, u64);
832 extern void blk_limits_max_hw_sectors(struct queue_limits *, unsigned int);
833 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
834 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
835 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
836 extern void blk_queue_max_discard_sectors(struct request_queue *q,
837 unsigned int max_discard_sectors);
838 extern void blk_queue_logical_block_size(struct request_queue *, unsigned short);
839 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
840 extern void blk_queue_alignment_offset(struct request_queue *q,
841 unsigned int alignment);
842 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
843 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
844 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
845 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
846 extern void blk_set_default_limits(struct queue_limits *lim);
847 extern void blk_set_stacking_limits(struct queue_limits *lim);
848 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
850 extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev,
852 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
854 extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b);
855 extern void blk_queue_dma_pad(struct request_queue *, unsigned int);
856 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
857 extern int blk_queue_dma_drain(struct request_queue *q,
858 dma_drain_needed_fn *dma_drain_needed,
859 void *buf, unsigned int size);
860 extern void blk_queue_lld_busy(struct request_queue *q, lld_busy_fn *fn);
861 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
862 extern void blk_queue_prep_rq(struct request_queue *, prep_rq_fn *pfn);
863 extern void blk_queue_unprep_rq(struct request_queue *, unprep_rq_fn *ufn);
864 extern void blk_queue_merge_bvec(struct request_queue *, merge_bvec_fn *);
865 extern void blk_queue_dma_alignment(struct request_queue *, int);
866 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
867 extern void blk_queue_softirq_done(struct request_queue *, softirq_done_fn *);
868 extern void blk_queue_rq_timed_out(struct request_queue *, rq_timed_out_fn *);
869 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
870 extern void blk_queue_flush(struct request_queue *q, unsigned int flush);
871 extern void blk_queue_flush_queueable(struct request_queue *q, bool queueable);
872 extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
874 extern int blk_rq_map_sg(struct request_queue *, struct request *, struct scatterlist *);
875 extern void blk_dump_rq_flags(struct request *, char *);
876 extern long nr_blockdev_pages(void);
878 bool __must_check blk_get_queue(struct request_queue *);
879 struct request_queue *blk_alloc_queue(gfp_t);
880 struct request_queue *blk_alloc_queue_node(gfp_t, int);
881 extern void blk_put_queue(struct request_queue *);
884 * blk_plug permits building a queue of related requests by holding the I/O
885 * fragments for a short period. This allows merging of sequential requests
886 * into single larger request. As the requests are moved from a per-task list to
887 * the device's request_queue in a batch, this results in improved scalability
888 * as the lock contention for request_queue lock is reduced.
890 * It is ok not to disable preemption when adding the request to the plug list
891 * or when attempting a merge, because blk_schedule_flush_list() will only flush
892 * the plug list when the task sleeps by itself. For details, please see
893 * schedule() where blk_schedule_flush_plug() is called.
896 unsigned long magic; /* detect uninitialized use-cases */
897 struct list_head list; /* requests */
898 struct list_head cb_list; /* md requires an unplug callback */
899 unsigned int should_sort; /* list to be sorted before flushing? */
901 #define BLK_MAX_REQUEST_COUNT 16
904 struct list_head list;
905 void (*callback)(struct blk_plug_cb *);
908 extern void blk_start_plug(struct blk_plug *);
909 extern void blk_finish_plug(struct blk_plug *);
910 extern void blk_flush_plug_list(struct blk_plug *, bool);
912 static inline void blk_flush_plug(struct task_struct *tsk)
914 struct blk_plug *plug = tsk->plug;
917 blk_flush_plug_list(plug, false);
920 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
922 struct blk_plug *plug = tsk->plug;
925 blk_flush_plug_list(plug, true);
928 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
930 struct blk_plug *plug = tsk->plug;
932 return plug && (!list_empty(&plug->list) || !list_empty(&plug->cb_list));
938 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
939 extern int blk_queue_start_tag(struct request_queue *, struct request *);
940 extern struct request *blk_queue_find_tag(struct request_queue *, int);
941 extern void blk_queue_end_tag(struct request_queue *, struct request *);
942 extern int blk_queue_init_tags(struct request_queue *, int, struct blk_queue_tag *);
943 extern void blk_queue_free_tags(struct request_queue *);
944 extern int blk_queue_resize_tags(struct request_queue *, int);
945 extern void blk_queue_invalidate_tags(struct request_queue *);
946 extern struct blk_queue_tag *blk_init_tags(int);
947 extern void blk_free_tags(struct blk_queue_tag *);
949 static inline struct request *blk_map_queue_find_tag(struct blk_queue_tag *bqt,
952 if (unlikely(bqt == NULL || tag >= bqt->real_max_depth))
954 return bqt->tag_index[tag];
957 #define BLKDEV_DISCARD_SECURE 0x01 /* secure discard */
959 extern int blkdev_issue_flush(struct block_device *, gfp_t, sector_t *);
960 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
961 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
962 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
963 sector_t nr_sects, gfp_t gfp_mask);
964 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
965 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
967 return blkdev_issue_discard(sb->s_bdev, block << (sb->s_blocksize_bits - 9),
968 nr_blocks << (sb->s_blocksize_bits - 9),
971 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
972 sector_t nr_blocks, gfp_t gfp_mask)
974 return blkdev_issue_zeroout(sb->s_bdev,
975 block << (sb->s_blocksize_bits - 9),
976 nr_blocks << (sb->s_blocksize_bits - 9),
980 extern int blk_verify_command(unsigned char *cmd, fmode_t has_write_perm);
982 enum blk_default_limits {
983 BLK_MAX_SEGMENTS = 128,
984 BLK_SAFE_MAX_SECTORS = 255,
985 BLK_DEF_MAX_SECTORS = 1024,
986 BLK_MAX_SEGMENT_SIZE = 65536,
987 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
990 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
992 static inline unsigned long queue_bounce_pfn(struct request_queue *q)
994 return q->limits.bounce_pfn;
997 static inline unsigned long queue_segment_boundary(struct request_queue *q)
999 return q->limits.seg_boundary_mask;
1002 static inline unsigned int queue_max_sectors(struct request_queue *q)
1004 return q->limits.max_sectors;
1007 static inline unsigned int queue_max_hw_sectors(struct request_queue *q)
1009 return q->limits.max_hw_sectors;
1012 static inline unsigned short queue_max_segments(struct request_queue *q)
1014 return q->limits.max_segments;
1017 static inline unsigned int queue_max_segment_size(struct request_queue *q)
1019 return q->limits.max_segment_size;
1022 static inline unsigned short queue_logical_block_size(struct request_queue *q)
1026 if (q && q->limits.logical_block_size)
1027 retval = q->limits.logical_block_size;
1032 static inline unsigned short bdev_logical_block_size(struct block_device *bdev)
1034 return queue_logical_block_size(bdev_get_queue(bdev));
1037 static inline unsigned int queue_physical_block_size(struct request_queue *q)
1039 return q->limits.physical_block_size;
1042 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1044 return queue_physical_block_size(bdev_get_queue(bdev));
1047 static inline unsigned int queue_io_min(struct request_queue *q)
1049 return q->limits.io_min;
1052 static inline int bdev_io_min(struct block_device *bdev)
1054 return queue_io_min(bdev_get_queue(bdev));
1057 static inline unsigned int queue_io_opt(struct request_queue *q)
1059 return q->limits.io_opt;
1062 static inline int bdev_io_opt(struct block_device *bdev)
1064 return queue_io_opt(bdev_get_queue(bdev));
1067 static inline int queue_alignment_offset(struct request_queue *q)
1069 if (q->limits.misaligned)
1072 return q->limits.alignment_offset;
1075 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1077 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1078 unsigned int alignment = (sector << 9) & (granularity - 1);
1080 return (granularity + lim->alignment_offset - alignment)
1081 & (granularity - 1);
1084 static inline int bdev_alignment_offset(struct block_device *bdev)
1086 struct request_queue *q = bdev_get_queue(bdev);
1088 if (q->limits.misaligned)
1091 if (bdev != bdev->bd_contains)
1092 return bdev->bd_part->alignment_offset;
1094 return q->limits.alignment_offset;
1097 static inline int queue_discard_alignment(struct request_queue *q)
1099 if (q->limits.discard_misaligned)
1102 return q->limits.discard_alignment;
1105 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1107 unsigned int alignment = (sector << 9) & (lim->discard_granularity - 1);
1109 if (!lim->max_discard_sectors)
1112 return (lim->discard_granularity + lim->discard_alignment - alignment)
1113 & (lim->discard_granularity - 1);
1116 static inline unsigned int queue_discard_zeroes_data(struct request_queue *q)
1118 if (q->limits.max_discard_sectors && q->limits.discard_zeroes_data == 1)
1124 static inline unsigned int bdev_discard_zeroes_data(struct block_device *bdev)
1126 return queue_discard_zeroes_data(bdev_get_queue(bdev));
1129 static inline int queue_dma_alignment(struct request_queue *q)
1131 return q ? q->dma_alignment : 511;
1134 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1137 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1138 return !(addr & alignment) && !(len & alignment);
1141 /* assumes size > 256 */
1142 static inline unsigned int blksize_bits(unsigned int size)
1144 unsigned int bits = 8;
1148 } while (size > 256);
1152 static inline unsigned int block_size(struct block_device *bdev)
1154 return bdev->bd_block_size;
1157 static inline bool queue_flush_queueable(struct request_queue *q)
1159 return !q->flush_not_queueable;
1162 typedef struct {struct page *v;} Sector;
1164 unsigned char *read_dev_sector(struct block_device *, sector_t, Sector *);
1166 static inline void put_dev_sector(Sector p)
1168 page_cache_release(p.v);
1172 int kblockd_schedule_work(struct request_queue *q, struct work_struct *work);
1174 #ifdef CONFIG_BLK_CGROUP
1176 * This should not be using sched_clock(). A real patch is in progress
1177 * to fix this up, until that is in place we need to disable preemption
1178 * around sched_clock() in this function and set_io_start_time_ns().
1180 static inline void set_start_time_ns(struct request *req)
1183 req->start_time_ns = sched_clock();
1187 static inline void set_io_start_time_ns(struct request *req)
1190 req->io_start_time_ns = sched_clock();
1194 static inline uint64_t rq_start_time_ns(struct request *req)
1196 return req->start_time_ns;
1199 static inline uint64_t rq_io_start_time_ns(struct request *req)
1201 return req->io_start_time_ns;
1204 static inline void set_start_time_ns(struct request *req) {}
1205 static inline void set_io_start_time_ns(struct request *req) {}
1206 static inline uint64_t rq_start_time_ns(struct request *req)
1210 static inline uint64_t rq_io_start_time_ns(struct request *req)
1216 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1217 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1218 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1219 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1221 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1223 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1224 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1226 struct blk_integrity_exchg {
1230 unsigned int data_size;
1231 unsigned short sector_size;
1232 const char *disk_name;
1235 typedef void (integrity_gen_fn) (struct blk_integrity_exchg *);
1236 typedef int (integrity_vrfy_fn) (struct blk_integrity_exchg *);
1237 typedef void (integrity_set_tag_fn) (void *, void *, unsigned int);
1238 typedef void (integrity_get_tag_fn) (void *, void *, unsigned int);
1240 struct blk_integrity {
1241 integrity_gen_fn *generate_fn;
1242 integrity_vrfy_fn *verify_fn;
1243 integrity_set_tag_fn *set_tag_fn;
1244 integrity_get_tag_fn *get_tag_fn;
1246 unsigned short flags;
1247 unsigned short tuple_size;
1248 unsigned short sector_size;
1249 unsigned short tag_size;
1253 struct kobject kobj;
1256 extern bool blk_integrity_is_initialized(struct gendisk *);
1257 extern int blk_integrity_register(struct gendisk *, struct blk_integrity *);
1258 extern void blk_integrity_unregister(struct gendisk *);
1259 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1260 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1261 struct scatterlist *);
1262 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1263 extern int blk_integrity_merge_rq(struct request_queue *, struct request *,
1265 extern int blk_integrity_merge_bio(struct request_queue *, struct request *,
1269 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1271 return bdev->bd_disk->integrity;
1274 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1276 return disk->integrity;
1279 static inline int blk_integrity_rq(struct request *rq)
1281 if (rq->bio == NULL)
1284 return bio_integrity(rq->bio);
1287 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1290 q->limits.max_integrity_segments = segs;
1293 static inline unsigned short
1294 queue_max_integrity_segments(struct request_queue *q)
1296 return q->limits.max_integrity_segments;
1299 #else /* CONFIG_BLK_DEV_INTEGRITY */
1301 #define blk_integrity_rq(rq) (0)
1302 #define blk_rq_count_integrity_sg(a, b) (0)
1303 #define blk_rq_map_integrity_sg(a, b, c) (0)
1304 #define bdev_get_integrity(a) (0)
1305 #define blk_get_integrity(a) (0)
1306 #define blk_integrity_compare(a, b) (0)
1307 #define blk_integrity_register(a, b) (0)
1308 #define blk_integrity_unregister(a) do { } while (0)
1309 #define blk_queue_max_integrity_segments(a, b) do { } while (0)
1310 #define queue_max_integrity_segments(a) (0)
1311 #define blk_integrity_merge_rq(a, b, c) (0)
1312 #define blk_integrity_merge_bio(a, b, c) (0)
1313 #define blk_integrity_is_initialized(a) (0)
1315 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1317 struct block_device_operations {
1318 int (*open) (struct block_device *, fmode_t);
1319 int (*release) (struct gendisk *, fmode_t);
1320 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1321 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1322 int (*direct_access) (struct block_device *, sector_t,
1323 void **, unsigned long *);
1324 unsigned int (*check_events) (struct gendisk *disk,
1325 unsigned int clearing);
1326 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1327 int (*media_changed) (struct gendisk *);
1328 void (*unlock_native_capacity) (struct gendisk *);
1329 int (*revalidate_disk) (struct gendisk *);
1330 int (*getgeo)(struct block_device *, struct hd_geometry *);
1331 /* this callback is with swap_lock and sometimes page table lock held */
1332 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1333 struct module *owner;
1336 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1338 #else /* CONFIG_BLOCK */
1340 * stubs for when the block layer is configured out
1342 #define buffer_heads_over_limit 0
1344 static inline long nr_blockdev_pages(void)
1352 static inline void blk_start_plug(struct blk_plug *plug)
1356 static inline void blk_finish_plug(struct blk_plug *plug)
1360 static inline void blk_flush_plug(struct task_struct *task)
1364 static inline void blk_schedule_flush_plug(struct task_struct *task)
1369 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1374 #endif /* CONFIG_BLOCK */