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;
404 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
405 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
406 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
407 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
408 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
409 #define QUEUE_FLAG_ELVSWITCH 6 /* don't use elevator, just do FIFO */
410 #define QUEUE_FLAG_BIDI 7 /* queue supports bidi requests */
411 #define QUEUE_FLAG_NOMERGES 8 /* disable merge attempts */
412 #define QUEUE_FLAG_SAME_COMP 9 /* complete on same CPU-group */
413 #define QUEUE_FLAG_FAIL_IO 10 /* fake timeout */
414 #define QUEUE_FLAG_STACKABLE 11 /* supports request stacking */
415 #define QUEUE_FLAG_NONROT 12 /* non-rotational device (SSD) */
416 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
417 #define QUEUE_FLAG_IO_STAT 13 /* do IO stats */
418 #define QUEUE_FLAG_DISCARD 14 /* supports DISCARD */
419 #define QUEUE_FLAG_NOXMERGES 15 /* No extended merges */
420 #define QUEUE_FLAG_ADD_RANDOM 16 /* Contributes to random pool */
421 #define QUEUE_FLAG_SECDISCARD 17 /* supports SECDISCARD */
422 #define QUEUE_FLAG_SAME_FORCE 18 /* force complete on same CPU */
424 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
425 (1 << QUEUE_FLAG_STACKABLE) | \
426 (1 << QUEUE_FLAG_SAME_COMP) | \
427 (1 << QUEUE_FLAG_ADD_RANDOM))
429 static inline void queue_lockdep_assert_held(struct request_queue *q)
432 lockdep_assert_held(q->queue_lock);
435 static inline void queue_flag_set_unlocked(unsigned int flag,
436 struct request_queue *q)
438 __set_bit(flag, &q->queue_flags);
441 static inline int queue_flag_test_and_clear(unsigned int flag,
442 struct request_queue *q)
444 queue_lockdep_assert_held(q);
446 if (test_bit(flag, &q->queue_flags)) {
447 __clear_bit(flag, &q->queue_flags);
454 static inline int queue_flag_test_and_set(unsigned int flag,
455 struct request_queue *q)
457 queue_lockdep_assert_held(q);
459 if (!test_bit(flag, &q->queue_flags)) {
460 __set_bit(flag, &q->queue_flags);
467 static inline void queue_flag_set(unsigned int flag, struct request_queue *q)
469 queue_lockdep_assert_held(q);
470 __set_bit(flag, &q->queue_flags);
473 static inline void queue_flag_clear_unlocked(unsigned int flag,
474 struct request_queue *q)
476 __clear_bit(flag, &q->queue_flags);
479 static inline int queue_in_flight(struct request_queue *q)
481 return q->in_flight[0] + q->in_flight[1];
484 static inline void queue_flag_clear(unsigned int flag, struct request_queue *q)
486 queue_lockdep_assert_held(q);
487 __clear_bit(flag, &q->queue_flags);
490 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
491 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
492 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
493 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
494 #define blk_queue_noxmerges(q) \
495 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
496 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
497 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
498 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
499 #define blk_queue_stackable(q) \
500 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
501 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
502 #define blk_queue_secdiscard(q) (blk_queue_discard(q) && \
503 test_bit(QUEUE_FLAG_SECDISCARD, &(q)->queue_flags))
505 #define blk_noretry_request(rq) \
506 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
507 REQ_FAILFAST_DRIVER))
509 #define blk_account_rq(rq) \
510 (((rq)->cmd_flags & REQ_STARTED) && \
511 ((rq)->cmd_type == REQ_TYPE_FS || \
512 ((rq)->cmd_flags & REQ_DISCARD)))
514 #define blk_pm_request(rq) \
515 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
516 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
518 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
519 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
520 /* rq->queuelist of dequeued request must be list_empty() */
521 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
523 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
525 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
527 static inline unsigned int blk_queue_cluster(struct request_queue *q)
529 return q->limits.cluster;
533 * We regard a request as sync, if either a read or a sync write
535 static inline bool rw_is_sync(unsigned int rw_flags)
537 return !(rw_flags & REQ_WRITE) || (rw_flags & REQ_SYNC);
540 static inline bool rq_is_sync(struct request *rq)
542 return rw_is_sync(rq->cmd_flags);
545 static inline int blk_queue_full(struct request_queue *q, int sync)
548 return test_bit(QUEUE_FLAG_SYNCFULL, &q->queue_flags);
549 return test_bit(QUEUE_FLAG_ASYNCFULL, &q->queue_flags);
552 static inline void blk_set_queue_full(struct request_queue *q, int sync)
555 queue_flag_set(QUEUE_FLAG_SYNCFULL, q);
557 queue_flag_set(QUEUE_FLAG_ASYNCFULL, q);
560 static inline void blk_clear_queue_full(struct request_queue *q, int sync)
563 queue_flag_clear(QUEUE_FLAG_SYNCFULL, q);
565 queue_flag_clear(QUEUE_FLAG_ASYNCFULL, q);
570 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
571 * it already be started by driver.
573 #define RQ_NOMERGE_FLAGS \
574 (REQ_NOMERGE | REQ_STARTED | REQ_SOFTBARRIER | REQ_FLUSH | REQ_FUA)
575 #define rq_mergeable(rq) \
576 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
577 (((rq)->cmd_flags & REQ_DISCARD) || \
578 (rq)->cmd_type == REQ_TYPE_FS))
581 * q->prep_rq_fn return values
583 #define BLKPREP_OK 0 /* serve it */
584 #define BLKPREP_KILL 1 /* fatal error, kill */
585 #define BLKPREP_DEFER 2 /* leave on queue */
587 extern unsigned long blk_max_low_pfn, blk_max_pfn;
590 * standard bounce addresses:
592 * BLK_BOUNCE_HIGH : bounce all highmem pages
593 * BLK_BOUNCE_ANY : don't bounce anything
594 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
597 #if BITS_PER_LONG == 32
598 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
600 #define BLK_BOUNCE_HIGH -1ULL
602 #define BLK_BOUNCE_ANY (-1ULL)
603 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
606 * default timeout for SG_IO if none specified
608 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
609 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
612 extern int init_emergency_isa_pool(void);
613 extern void blk_queue_bounce(struct request_queue *q, struct bio **bio);
615 static inline int init_emergency_isa_pool(void)
619 static inline void blk_queue_bounce(struct request_queue *q, struct bio **bio)
622 #endif /* CONFIG_MMU */
628 unsigned long offset;
633 struct req_iterator {
638 /* This should not be used directly - use rq_for_each_segment */
639 #define for_each_bio(_bio) \
640 for (; _bio; _bio = _bio->bi_next)
641 #define __rq_for_each_bio(_bio, rq) \
643 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
645 #define rq_for_each_segment(bvl, _rq, _iter) \
646 __rq_for_each_bio(_iter.bio, _rq) \
647 bio_for_each_segment(bvl, _iter.bio, _iter.i)
649 #define rq_iter_last(rq, _iter) \
650 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
652 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
653 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
655 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
656 extern void rq_flush_dcache_pages(struct request *rq);
658 static inline void rq_flush_dcache_pages(struct request *rq)
663 extern int blk_register_queue(struct gendisk *disk);
664 extern void blk_unregister_queue(struct gendisk *disk);
665 extern void generic_make_request(struct bio *bio);
666 extern void blk_rq_init(struct request_queue *q, struct request *rq);
667 extern void blk_put_request(struct request *);
668 extern void __blk_put_request(struct request_queue *, struct request *);
669 extern struct request *blk_get_request(struct request_queue *, int, gfp_t);
670 extern struct request *blk_make_request(struct request_queue *, struct bio *,
672 extern void blk_requeue_request(struct request_queue *, struct request *);
673 extern void blk_add_request_payload(struct request *rq, struct page *page,
675 extern int blk_rq_check_limits(struct request_queue *q, struct request *rq);
676 extern int blk_lld_busy(struct request_queue *q);
677 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
678 struct bio_set *bs, gfp_t gfp_mask,
679 int (*bio_ctr)(struct bio *, struct bio *, void *),
681 extern void blk_rq_unprep_clone(struct request *rq);
682 extern int blk_insert_cloned_request(struct request_queue *q,
684 extern void blk_delay_queue(struct request_queue *, unsigned long);
685 extern void blk_recount_segments(struct request_queue *, struct bio *);
686 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
687 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
688 unsigned int, void __user *);
689 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
690 unsigned int, void __user *);
691 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
692 struct scsi_ioctl_command __user *);
694 extern void blk_queue_bio(struct request_queue *q, struct bio *bio);
697 * A queue has just exitted congestion. Note this in the global counter of
698 * congested queues, and wake up anyone who was waiting for requests to be
701 static inline void blk_clear_queue_congested(struct request_queue *q, int sync)
703 clear_bdi_congested(&q->backing_dev_info, sync);
707 * A queue has just entered congestion. Flag that in the queue's VM-visible
708 * state flags and increment the global gounter of congested queues.
710 static inline void blk_set_queue_congested(struct request_queue *q, int sync)
712 set_bdi_congested(&q->backing_dev_info, sync);
715 extern void blk_start_queue(struct request_queue *q);
716 extern void blk_stop_queue(struct request_queue *q);
717 extern void blk_sync_queue(struct request_queue *q);
718 extern void __blk_stop_queue(struct request_queue *q);
719 extern void __blk_run_queue(struct request_queue *q);
720 extern void blk_run_queue(struct request_queue *);
721 extern void blk_run_queue_async(struct request_queue *q);
722 extern int blk_rq_map_user(struct request_queue *, struct request *,
723 struct rq_map_data *, void __user *, unsigned long,
725 extern int blk_rq_unmap_user(struct bio *);
726 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
727 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
728 struct rq_map_data *, struct sg_iovec *, int,
729 unsigned int, gfp_t);
730 extern int blk_execute_rq(struct request_queue *, struct gendisk *,
731 struct request *, int);
732 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
733 struct request *, int, rq_end_io_fn *);
735 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
737 return bdev->bd_disk->queue;
741 * blk_rq_pos() : the current sector
742 * blk_rq_bytes() : bytes left in the entire request
743 * blk_rq_cur_bytes() : bytes left in the current segment
744 * blk_rq_err_bytes() : bytes left till the next error boundary
745 * blk_rq_sectors() : sectors left in the entire request
746 * blk_rq_cur_sectors() : sectors left in the current segment
748 static inline sector_t blk_rq_pos(const struct request *rq)
753 static inline unsigned int blk_rq_bytes(const struct request *rq)
755 return rq->__data_len;
758 static inline int blk_rq_cur_bytes(const struct request *rq)
760 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
763 extern unsigned int blk_rq_err_bytes(const struct request *rq);
765 static inline unsigned int blk_rq_sectors(const struct request *rq)
767 return blk_rq_bytes(rq) >> 9;
770 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
772 return blk_rq_cur_bytes(rq) >> 9;
776 * Request issue related functions.
778 extern struct request *blk_peek_request(struct request_queue *q);
779 extern void blk_start_request(struct request *rq);
780 extern struct request *blk_fetch_request(struct request_queue *q);
783 * Request completion related functions.
785 * blk_update_request() completes given number of bytes and updates
786 * the request without completing it.
788 * blk_end_request() and friends. __blk_end_request() must be called
789 * with the request queue spinlock acquired.
791 * Several drivers define their own end_request and call
792 * blk_end_request() for parts of the original function.
793 * This prevents code duplication in drivers.
795 extern bool blk_update_request(struct request *rq, int error,
796 unsigned int nr_bytes);
797 extern bool blk_end_request(struct request *rq, int error,
798 unsigned int nr_bytes);
799 extern void blk_end_request_all(struct request *rq, int error);
800 extern bool blk_end_request_cur(struct request *rq, int error);
801 extern bool blk_end_request_err(struct request *rq, int error);
802 extern bool __blk_end_request(struct request *rq, int error,
803 unsigned int nr_bytes);
804 extern void __blk_end_request_all(struct request *rq, int error);
805 extern bool __blk_end_request_cur(struct request *rq, int error);
806 extern bool __blk_end_request_err(struct request *rq, int error);
808 extern void blk_complete_request(struct request *);
809 extern void __blk_complete_request(struct request *);
810 extern void blk_abort_request(struct request *);
811 extern void blk_abort_queue(struct request_queue *);
812 extern void blk_unprep_request(struct request *);
815 * Access functions for manipulating queue properties
817 extern struct request_queue *blk_init_queue_node(request_fn_proc *rfn,
818 spinlock_t *lock, int node_id);
819 extern struct request_queue *blk_init_queue(request_fn_proc *, spinlock_t *);
820 extern struct request_queue *blk_init_allocated_queue(struct request_queue *,
821 request_fn_proc *, spinlock_t *);
822 extern void blk_cleanup_queue(struct request_queue *);
823 extern void blk_queue_make_request(struct request_queue *, make_request_fn *);
824 extern void blk_queue_bounce_limit(struct request_queue *, u64);
825 extern void blk_limits_max_hw_sectors(struct queue_limits *, unsigned int);
826 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
827 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
828 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
829 extern void blk_queue_max_discard_sectors(struct request_queue *q,
830 unsigned int max_discard_sectors);
831 extern void blk_queue_logical_block_size(struct request_queue *, unsigned short);
832 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
833 extern void blk_queue_alignment_offset(struct request_queue *q,
834 unsigned int alignment);
835 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
836 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
837 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
838 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
839 extern void blk_set_default_limits(struct queue_limits *lim);
840 extern void blk_set_stacking_limits(struct queue_limits *lim);
841 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
843 extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev,
845 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
847 extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b);
848 extern void blk_queue_dma_pad(struct request_queue *, unsigned int);
849 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
850 extern int blk_queue_dma_drain(struct request_queue *q,
851 dma_drain_needed_fn *dma_drain_needed,
852 void *buf, unsigned int size);
853 extern void blk_queue_lld_busy(struct request_queue *q, lld_busy_fn *fn);
854 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
855 extern void blk_queue_prep_rq(struct request_queue *, prep_rq_fn *pfn);
856 extern void blk_queue_unprep_rq(struct request_queue *, unprep_rq_fn *ufn);
857 extern void blk_queue_merge_bvec(struct request_queue *, merge_bvec_fn *);
858 extern void blk_queue_dma_alignment(struct request_queue *, int);
859 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
860 extern void blk_queue_softirq_done(struct request_queue *, softirq_done_fn *);
861 extern void blk_queue_rq_timed_out(struct request_queue *, rq_timed_out_fn *);
862 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
863 extern void blk_queue_flush(struct request_queue *q, unsigned int flush);
864 extern void blk_queue_flush_queueable(struct request_queue *q, bool queueable);
865 extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
867 extern int blk_rq_map_sg(struct request_queue *, struct request *, struct scatterlist *);
868 extern void blk_dump_rq_flags(struct request *, char *);
869 extern long nr_blockdev_pages(void);
871 bool __must_check blk_get_queue(struct request_queue *);
872 struct request_queue *blk_alloc_queue(gfp_t);
873 struct request_queue *blk_alloc_queue_node(gfp_t, int);
874 extern void blk_put_queue(struct request_queue *);
877 * blk_plug permits building a queue of related requests by holding the I/O
878 * fragments for a short period. This allows merging of sequential requests
879 * into single larger request. As the requests are moved from a per-task list to
880 * the device's request_queue in a batch, this results in improved scalability
881 * as the lock contention for request_queue lock is reduced.
883 * It is ok not to disable preemption when adding the request to the plug list
884 * or when attempting a merge, because blk_schedule_flush_list() will only flush
885 * the plug list when the task sleeps by itself. For details, please see
886 * schedule() where blk_schedule_flush_plug() is called.
889 unsigned long magic; /* detect uninitialized use-cases */
890 struct list_head list; /* requests */
891 struct list_head cb_list; /* md requires an unplug callback */
892 unsigned int should_sort; /* list to be sorted before flushing? */
894 #define BLK_MAX_REQUEST_COUNT 16
897 struct list_head list;
898 void (*callback)(struct blk_plug_cb *);
901 extern void blk_start_plug(struct blk_plug *);
902 extern void blk_finish_plug(struct blk_plug *);
903 extern void blk_flush_plug_list(struct blk_plug *, bool);
905 static inline void blk_flush_plug(struct task_struct *tsk)
907 struct blk_plug *plug = tsk->plug;
910 blk_flush_plug_list(plug, false);
913 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
915 struct blk_plug *plug = tsk->plug;
918 blk_flush_plug_list(plug, true);
921 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
923 struct blk_plug *plug = tsk->plug;
925 return plug && (!list_empty(&plug->list) || !list_empty(&plug->cb_list));
931 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
932 extern int blk_queue_start_tag(struct request_queue *, struct request *);
933 extern struct request *blk_queue_find_tag(struct request_queue *, int);
934 extern void blk_queue_end_tag(struct request_queue *, struct request *);
935 extern int blk_queue_init_tags(struct request_queue *, int, struct blk_queue_tag *);
936 extern void blk_queue_free_tags(struct request_queue *);
937 extern int blk_queue_resize_tags(struct request_queue *, int);
938 extern void blk_queue_invalidate_tags(struct request_queue *);
939 extern struct blk_queue_tag *blk_init_tags(int);
940 extern void blk_free_tags(struct blk_queue_tag *);
942 static inline struct request *blk_map_queue_find_tag(struct blk_queue_tag *bqt,
945 if (unlikely(bqt == NULL || tag >= bqt->real_max_depth))
947 return bqt->tag_index[tag];
950 #define BLKDEV_DISCARD_SECURE 0x01 /* secure discard */
952 extern int blkdev_issue_flush(struct block_device *, gfp_t, sector_t *);
953 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
954 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
955 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
956 sector_t nr_sects, gfp_t gfp_mask);
957 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
958 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
960 return blkdev_issue_discard(sb->s_bdev, block << (sb->s_blocksize_bits - 9),
961 nr_blocks << (sb->s_blocksize_bits - 9),
964 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
965 sector_t nr_blocks, gfp_t gfp_mask)
967 return blkdev_issue_zeroout(sb->s_bdev,
968 block << (sb->s_blocksize_bits - 9),
969 nr_blocks << (sb->s_blocksize_bits - 9),
973 extern int blk_verify_command(unsigned char *cmd, fmode_t has_write_perm);
975 enum blk_default_limits {
976 BLK_MAX_SEGMENTS = 128,
977 BLK_SAFE_MAX_SECTORS = 255,
978 BLK_DEF_MAX_SECTORS = 1024,
979 BLK_MAX_SEGMENT_SIZE = 65536,
980 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
983 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
985 static inline unsigned long queue_bounce_pfn(struct request_queue *q)
987 return q->limits.bounce_pfn;
990 static inline unsigned long queue_segment_boundary(struct request_queue *q)
992 return q->limits.seg_boundary_mask;
995 static inline unsigned int queue_max_sectors(struct request_queue *q)
997 return q->limits.max_sectors;
1000 static inline unsigned int queue_max_hw_sectors(struct request_queue *q)
1002 return q->limits.max_hw_sectors;
1005 static inline unsigned short queue_max_segments(struct request_queue *q)
1007 return q->limits.max_segments;
1010 static inline unsigned int queue_max_segment_size(struct request_queue *q)
1012 return q->limits.max_segment_size;
1015 static inline unsigned short queue_logical_block_size(struct request_queue *q)
1019 if (q && q->limits.logical_block_size)
1020 retval = q->limits.logical_block_size;
1025 static inline unsigned short bdev_logical_block_size(struct block_device *bdev)
1027 return queue_logical_block_size(bdev_get_queue(bdev));
1030 static inline unsigned int queue_physical_block_size(struct request_queue *q)
1032 return q->limits.physical_block_size;
1035 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1037 return queue_physical_block_size(bdev_get_queue(bdev));
1040 static inline unsigned int queue_io_min(struct request_queue *q)
1042 return q->limits.io_min;
1045 static inline int bdev_io_min(struct block_device *bdev)
1047 return queue_io_min(bdev_get_queue(bdev));
1050 static inline unsigned int queue_io_opt(struct request_queue *q)
1052 return q->limits.io_opt;
1055 static inline int bdev_io_opt(struct block_device *bdev)
1057 return queue_io_opt(bdev_get_queue(bdev));
1060 static inline int queue_alignment_offset(struct request_queue *q)
1062 if (q->limits.misaligned)
1065 return q->limits.alignment_offset;
1068 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1070 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1071 unsigned int alignment = (sector << 9) & (granularity - 1);
1073 return (granularity + lim->alignment_offset - alignment)
1074 & (granularity - 1);
1077 static inline int bdev_alignment_offset(struct block_device *bdev)
1079 struct request_queue *q = bdev_get_queue(bdev);
1081 if (q->limits.misaligned)
1084 if (bdev != bdev->bd_contains)
1085 return bdev->bd_part->alignment_offset;
1087 return q->limits.alignment_offset;
1090 static inline int queue_discard_alignment(struct request_queue *q)
1092 if (q->limits.discard_misaligned)
1095 return q->limits.discard_alignment;
1098 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1100 unsigned int alignment = (sector << 9) & (lim->discard_granularity - 1);
1102 if (!lim->max_discard_sectors)
1105 return (lim->discard_granularity + lim->discard_alignment - alignment)
1106 & (lim->discard_granularity - 1);
1109 static inline unsigned int queue_discard_zeroes_data(struct request_queue *q)
1111 if (q->limits.max_discard_sectors && q->limits.discard_zeroes_data == 1)
1117 static inline unsigned int bdev_discard_zeroes_data(struct block_device *bdev)
1119 return queue_discard_zeroes_data(bdev_get_queue(bdev));
1122 static inline int queue_dma_alignment(struct request_queue *q)
1124 return q ? q->dma_alignment : 511;
1127 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1130 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1131 return !(addr & alignment) && !(len & alignment);
1134 /* assumes size > 256 */
1135 static inline unsigned int blksize_bits(unsigned int size)
1137 unsigned int bits = 8;
1141 } while (size > 256);
1145 static inline unsigned int block_size(struct block_device *bdev)
1147 return bdev->bd_block_size;
1150 static inline bool queue_flush_queueable(struct request_queue *q)
1152 return !q->flush_not_queueable;
1155 typedef struct {struct page *v;} Sector;
1157 unsigned char *read_dev_sector(struct block_device *, sector_t, Sector *);
1159 static inline void put_dev_sector(Sector p)
1161 page_cache_release(p.v);
1165 int kblockd_schedule_work(struct request_queue *q, struct work_struct *work);
1167 #ifdef CONFIG_BLK_CGROUP
1169 * This should not be using sched_clock(). A real patch is in progress
1170 * to fix this up, until that is in place we need to disable preemption
1171 * around sched_clock() in this function and set_io_start_time_ns().
1173 static inline void set_start_time_ns(struct request *req)
1176 req->start_time_ns = sched_clock();
1180 static inline void set_io_start_time_ns(struct request *req)
1183 req->io_start_time_ns = sched_clock();
1187 static inline uint64_t rq_start_time_ns(struct request *req)
1189 return req->start_time_ns;
1192 static inline uint64_t rq_io_start_time_ns(struct request *req)
1194 return req->io_start_time_ns;
1197 static inline void set_start_time_ns(struct request *req) {}
1198 static inline void set_io_start_time_ns(struct request *req) {}
1199 static inline uint64_t rq_start_time_ns(struct request *req)
1203 static inline uint64_t rq_io_start_time_ns(struct request *req)
1209 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1210 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1211 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1212 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1214 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1216 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1217 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1219 struct blk_integrity_exchg {
1223 unsigned int data_size;
1224 unsigned short sector_size;
1225 const char *disk_name;
1228 typedef void (integrity_gen_fn) (struct blk_integrity_exchg *);
1229 typedef int (integrity_vrfy_fn) (struct blk_integrity_exchg *);
1230 typedef void (integrity_set_tag_fn) (void *, void *, unsigned int);
1231 typedef void (integrity_get_tag_fn) (void *, void *, unsigned int);
1233 struct blk_integrity {
1234 integrity_gen_fn *generate_fn;
1235 integrity_vrfy_fn *verify_fn;
1236 integrity_set_tag_fn *set_tag_fn;
1237 integrity_get_tag_fn *get_tag_fn;
1239 unsigned short flags;
1240 unsigned short tuple_size;
1241 unsigned short sector_size;
1242 unsigned short tag_size;
1246 struct kobject kobj;
1249 extern bool blk_integrity_is_initialized(struct gendisk *);
1250 extern int blk_integrity_register(struct gendisk *, struct blk_integrity *);
1251 extern void blk_integrity_unregister(struct gendisk *);
1252 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1253 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1254 struct scatterlist *);
1255 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1256 extern int blk_integrity_merge_rq(struct request_queue *, struct request *,
1258 extern int blk_integrity_merge_bio(struct request_queue *, struct request *,
1262 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1264 return bdev->bd_disk->integrity;
1267 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1269 return disk->integrity;
1272 static inline int blk_integrity_rq(struct request *rq)
1274 if (rq->bio == NULL)
1277 return bio_integrity(rq->bio);
1280 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1283 q->limits.max_integrity_segments = segs;
1286 static inline unsigned short
1287 queue_max_integrity_segments(struct request_queue *q)
1289 return q->limits.max_integrity_segments;
1292 #else /* CONFIG_BLK_DEV_INTEGRITY */
1295 struct block_device;
1297 struct blk_integrity;
1299 static inline int blk_integrity_rq(struct request *rq)
1303 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1308 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1310 struct scatterlist *s)
1314 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1318 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1322 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1326 static inline int blk_integrity_register(struct gendisk *d,
1327 struct blk_integrity *b)
1331 static inline void blk_integrity_unregister(struct gendisk *d)
1334 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1338 static inline unsigned short queue_max_integrity_segments(struct request_queue *q)
1342 static inline int blk_integrity_merge_rq(struct request_queue *rq,
1348 static inline int blk_integrity_merge_bio(struct request_queue *rq,
1354 static inline bool blk_integrity_is_initialized(struct gendisk *g)
1359 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1361 struct block_device_operations {
1362 int (*open) (struct block_device *, fmode_t);
1363 int (*release) (struct gendisk *, fmode_t);
1364 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1365 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1366 int (*direct_access) (struct block_device *, sector_t,
1367 void **, unsigned long *);
1368 unsigned int (*check_events) (struct gendisk *disk,
1369 unsigned int clearing);
1370 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1371 int (*media_changed) (struct gendisk *);
1372 void (*unlock_native_capacity) (struct gendisk *);
1373 int (*revalidate_disk) (struct gendisk *);
1374 int (*getgeo)(struct block_device *, struct hd_geometry *);
1375 /* this callback is with swap_lock and sometimes page table lock held */
1376 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1377 struct module *owner;
1380 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1382 #else /* CONFIG_BLOCK */
1384 * stubs for when the block layer is configured out
1386 #define buffer_heads_over_limit 0
1388 static inline long nr_blockdev_pages(void)
1396 static inline void blk_start_plug(struct blk_plug *plug)
1400 static inline void blk_finish_plug(struct blk_plug *plug)
1404 static inline void blk_flush_plug(struct task_struct *task)
1408 static inline void blk_schedule_flush_plug(struct task_struct *task)
1413 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1418 #endif /* CONFIG_BLOCK */