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;
34 #define BLKDEV_MIN_RQ 4
35 #define BLKDEV_MAX_RQ 128 /* Default maximum */
38 typedef void (rq_end_io_fn)(struct request *, int);
42 * count[], starved[], and wait[] are indexed by
43 * BLK_RW_SYNC/BLK_RW_ASYNC
49 wait_queue_head_t wait[2];
53 * request command types
55 enum rq_cmd_type_bits {
56 REQ_TYPE_FS = 1, /* fs request */
57 REQ_TYPE_BLOCK_PC, /* scsi command */
58 REQ_TYPE_SENSE, /* sense request */
59 REQ_TYPE_PM_SUSPEND, /* suspend request */
60 REQ_TYPE_PM_RESUME, /* resume request */
61 REQ_TYPE_PM_SHUTDOWN, /* shutdown request */
62 REQ_TYPE_SPECIAL, /* driver defined type */
64 * for ATA/ATAPI devices. this really doesn't belong here, ide should
65 * use REQ_TYPE_SPECIAL and use rq->cmd[0] with the range of driver
66 * private REQ_LB opcodes to differentiate what type of request this is
68 REQ_TYPE_ATA_TASKFILE,
72 #define BLK_MAX_CDB 16
75 * try to put the fields that are referenced together in the same cacheline.
76 * if you modify this structure, be sure to check block/blk-core.c:rq_init()
80 struct list_head queuelist;
81 struct call_single_data csd;
83 struct request_queue *q;
85 unsigned int cmd_flags;
86 enum rq_cmd_type_bits cmd_type;
87 unsigned long atomic_flags;
91 /* the following two fields are internal, NEVER access directly */
92 unsigned int __data_len; /* total data len */
93 sector_t __sector; /* sector cursor */
98 struct hlist_node hash; /* merge hash */
100 * The rb_node is only used inside the io scheduler, requests
101 * are pruned when moved to the dispatch queue. So let the
102 * completion_data share space with the rb_node.
105 struct rb_node rb_node; /* sort/lookup */
106 void *completion_data;
110 * Three pointers are available for the IO schedulers, if they need
111 * more they have to dynamically allocate it.
113 void *elevator_private;
114 void *elevator_private2;
115 void *elevator_private3;
117 struct gendisk *rq_disk;
118 unsigned long start_time;
119 #ifdef CONFIG_BLK_CGROUP
120 unsigned long long start_time_ns;
121 unsigned long long io_start_time_ns; /* when passed to hardware */
123 /* Number of scatter-gather DMA addr+len pairs after
124 * physical address coalescing is performed.
126 unsigned short nr_phys_segments;
127 #if defined(CONFIG_BLK_DEV_INTEGRITY)
128 unsigned short nr_integrity_segments;
131 unsigned short ioprio;
135 void *special; /* opaque pointer available for LLD use */
136 char *buffer; /* kaddr of the current segment if available */
142 * when request is used as a packet command carrier
144 unsigned char __cmd[BLK_MAX_CDB];
146 unsigned short cmd_len;
148 unsigned int extra_len; /* length of alignment and padding */
149 unsigned int sense_len;
150 unsigned int resid_len; /* residual count */
153 unsigned long deadline;
154 struct list_head timeout_list;
155 unsigned int timeout;
159 * completion callback.
161 rq_end_io_fn *end_io;
165 struct request *next_rq;
168 static inline unsigned short req_get_ioprio(struct request *req)
174 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
175 * requests. Some step values could eventually be made generic.
177 struct request_pm_state
179 /* PM state machine step value, currently driver specific */
181 /* requested PM state value (S1, S2, S3, S4, ...) */
183 void* data; /* for driver use */
186 #include <linux/elevator.h>
188 typedef void (request_fn_proc) (struct request_queue *q);
189 typedef int (make_request_fn) (struct request_queue *q, struct bio *bio);
190 typedef int (prep_rq_fn) (struct request_queue *, struct request *);
191 typedef void (unprep_rq_fn) (struct request_queue *, struct request *);
192 typedef void (unplug_fn) (struct request_queue *);
195 struct bvec_merge_data {
196 struct block_device *bi_bdev;
201 typedef int (merge_bvec_fn) (struct request_queue *, struct bvec_merge_data *,
203 typedef void (softirq_done_fn)(struct request *);
204 typedef int (dma_drain_needed_fn)(struct request *);
205 typedef int (lld_busy_fn) (struct request_queue *q);
207 enum blk_eh_timer_return {
213 typedef enum blk_eh_timer_return (rq_timed_out_fn)(struct request *);
215 enum blk_queue_state {
220 struct blk_queue_tag {
221 struct request **tag_index; /* map of busy tags */
222 unsigned long *tag_map; /* bit map of free/busy tags */
223 int busy; /* current depth */
224 int max_depth; /* what we will send to device */
225 int real_max_depth; /* what the array can hold */
226 atomic_t refcnt; /* map can be shared */
229 #define BLK_SCSI_MAX_CMDS (256)
230 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
232 struct queue_limits {
233 unsigned long bounce_pfn;
234 unsigned long seg_boundary_mask;
236 unsigned int max_hw_sectors;
237 unsigned int max_sectors;
238 unsigned int max_segment_size;
239 unsigned int physical_block_size;
240 unsigned int alignment_offset;
243 unsigned int max_discard_sectors;
244 unsigned int discard_granularity;
245 unsigned int discard_alignment;
247 unsigned short logical_block_size;
248 unsigned short max_segments;
249 unsigned short max_integrity_segments;
251 unsigned char misaligned;
252 unsigned char discard_misaligned;
253 unsigned char no_cluster;
254 signed char discard_zeroes_data;
260 * Together with queue_head for cacheline sharing
262 struct list_head queue_head;
263 struct request *last_merge;
264 struct elevator_queue *elevator;
267 * the queue request freelist, one for reads and one for writes
269 struct request_list rq;
271 request_fn_proc *request_fn;
272 make_request_fn *make_request_fn;
273 prep_rq_fn *prep_rq_fn;
274 unprep_rq_fn *unprep_rq_fn;
275 unplug_fn *unplug_fn;
276 merge_bvec_fn *merge_bvec_fn;
277 softirq_done_fn *softirq_done_fn;
278 rq_timed_out_fn *rq_timed_out_fn;
279 dma_drain_needed_fn *dma_drain_needed;
280 lld_busy_fn *lld_busy_fn;
283 * Dispatch queue sorting
286 struct request *boundary_rq;
289 * Auto-unplugging state
291 struct timer_list unplug_timer;
292 int unplug_thresh; /* After this many requests */
293 unsigned long unplug_delay; /* After this many jiffies */
294 struct work_struct unplug_work;
296 struct backing_dev_info backing_dev_info;
299 * The queue owner gets to use this for whatever they like.
300 * ll_rw_blk doesn't touch it.
305 * queue needs bounce pages for pages above this limit
310 * various queue flags, see QUEUE_* below
312 unsigned long queue_flags;
315 * protects queue structures from reentrancy. ->__queue_lock should
316 * _never_ be used directly, it is queue private. always use
319 spinlock_t __queue_lock;
320 spinlock_t *queue_lock;
330 unsigned long nr_requests; /* Max # of requests */
331 unsigned int nr_congestion_on;
332 unsigned int nr_congestion_off;
333 unsigned int nr_batching;
335 void *dma_drain_buffer;
336 unsigned int dma_drain_size;
337 unsigned int dma_pad_mask;
338 unsigned int dma_alignment;
340 struct blk_queue_tag *queue_tags;
341 struct list_head tag_busy_list;
343 unsigned int nr_sorted;
344 unsigned int in_flight[2];
346 unsigned int rq_timeout;
347 struct timer_list timeout;
348 struct list_head timeout_list;
350 struct queue_limits limits;
355 unsigned int sg_timeout;
356 unsigned int sg_reserved_size;
358 #ifdef CONFIG_BLK_DEV_IO_TRACE
359 struct blk_trace *blk_trace;
362 * reserved for flush operations
364 unsigned int ordered, next_ordered, ordseq;
365 int orderr, ordcolor;
366 struct request pre_flush_rq, bar_rq, post_flush_rq;
367 struct request *orig_bar_rq;
369 struct mutex sysfs_lock;
371 #if defined(CONFIG_BLK_DEV_BSG)
372 struct bsg_class_device bsg_dev;
375 #ifdef CONFIG_BLK_DEV_THROTTLING
377 struct throtl_data *td;
381 #define QUEUE_FLAG_CLUSTER 0 /* cluster several segments into 1 */
382 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
383 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
384 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
385 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
386 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
387 #define QUEUE_FLAG_REENTER 6 /* Re-entrancy avoidance */
388 #define QUEUE_FLAG_PLUGGED 7 /* queue is plugged */
389 #define QUEUE_FLAG_ELVSWITCH 8 /* don't use elevator, just do FIFO */
390 #define QUEUE_FLAG_BIDI 9 /* queue supports bidi requests */
391 #define QUEUE_FLAG_NOMERGES 10 /* disable merge attempts */
392 #define QUEUE_FLAG_SAME_COMP 11 /* force complete on same CPU */
393 #define QUEUE_FLAG_FAIL_IO 12 /* fake timeout */
394 #define QUEUE_FLAG_STACKABLE 13 /* supports request stacking */
395 #define QUEUE_FLAG_NONROT 14 /* non-rotational device (SSD) */
396 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
397 #define QUEUE_FLAG_IO_STAT 15 /* do IO stats */
398 #define QUEUE_FLAG_DISCARD 16 /* supports DISCARD */
399 #define QUEUE_FLAG_NOXMERGES 17 /* No extended merges */
400 #define QUEUE_FLAG_ADD_RANDOM 18 /* Contributes to random pool */
401 #define QUEUE_FLAG_SECDISCARD 19 /* supports SECDISCARD */
403 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
404 (1 << QUEUE_FLAG_CLUSTER) | \
405 (1 << QUEUE_FLAG_STACKABLE) | \
406 (1 << QUEUE_FLAG_SAME_COMP) | \
407 (1 << QUEUE_FLAG_ADD_RANDOM))
409 static inline int queue_is_locked(struct request_queue *q)
412 spinlock_t *lock = q->queue_lock;
413 return lock && spin_is_locked(lock);
419 static inline void queue_flag_set_unlocked(unsigned int flag,
420 struct request_queue *q)
422 __set_bit(flag, &q->queue_flags);
425 static inline int queue_flag_test_and_clear(unsigned int flag,
426 struct request_queue *q)
428 WARN_ON_ONCE(!queue_is_locked(q));
430 if (test_bit(flag, &q->queue_flags)) {
431 __clear_bit(flag, &q->queue_flags);
438 static inline int queue_flag_test_and_set(unsigned int flag,
439 struct request_queue *q)
441 WARN_ON_ONCE(!queue_is_locked(q));
443 if (!test_bit(flag, &q->queue_flags)) {
444 __set_bit(flag, &q->queue_flags);
451 static inline void queue_flag_set(unsigned int flag, struct request_queue *q)
453 WARN_ON_ONCE(!queue_is_locked(q));
454 __set_bit(flag, &q->queue_flags);
457 static inline void queue_flag_clear_unlocked(unsigned int flag,
458 struct request_queue *q)
460 __clear_bit(flag, &q->queue_flags);
463 static inline int queue_in_flight(struct request_queue *q)
465 return q->in_flight[0] + q->in_flight[1];
468 static inline void queue_flag_clear(unsigned int flag, struct request_queue *q)
470 WARN_ON_ONCE(!queue_is_locked(q));
471 __clear_bit(flag, &q->queue_flags);
476 * Hardbarrier is supported with one of the following methods.
478 * NONE : hardbarrier unsupported
479 * DRAIN : ordering by draining is enough
480 * DRAIN_FLUSH : ordering by draining w/ pre and post flushes
481 * DRAIN_FUA : ordering by draining w/ pre flush and FUA write
482 * TAG : ordering by tag is enough
483 * TAG_FLUSH : ordering by tag w/ pre and post flushes
484 * TAG_FUA : ordering by tag w/ pre flush and FUA write
486 QUEUE_ORDERED_BY_DRAIN = 0x01,
487 QUEUE_ORDERED_BY_TAG = 0x02,
488 QUEUE_ORDERED_DO_PREFLUSH = 0x10,
489 QUEUE_ORDERED_DO_BAR = 0x20,
490 QUEUE_ORDERED_DO_POSTFLUSH = 0x40,
491 QUEUE_ORDERED_DO_FUA = 0x80,
493 QUEUE_ORDERED_NONE = 0x00,
495 QUEUE_ORDERED_DRAIN = QUEUE_ORDERED_BY_DRAIN |
496 QUEUE_ORDERED_DO_BAR,
497 QUEUE_ORDERED_DRAIN_FLUSH = QUEUE_ORDERED_DRAIN |
498 QUEUE_ORDERED_DO_PREFLUSH |
499 QUEUE_ORDERED_DO_POSTFLUSH,
500 QUEUE_ORDERED_DRAIN_FUA = QUEUE_ORDERED_DRAIN |
501 QUEUE_ORDERED_DO_PREFLUSH |
502 QUEUE_ORDERED_DO_FUA,
504 QUEUE_ORDERED_TAG = QUEUE_ORDERED_BY_TAG |
505 QUEUE_ORDERED_DO_BAR,
506 QUEUE_ORDERED_TAG_FLUSH = QUEUE_ORDERED_TAG |
507 QUEUE_ORDERED_DO_PREFLUSH |
508 QUEUE_ORDERED_DO_POSTFLUSH,
509 QUEUE_ORDERED_TAG_FUA = QUEUE_ORDERED_TAG |
510 QUEUE_ORDERED_DO_PREFLUSH |
511 QUEUE_ORDERED_DO_FUA,
514 * Ordered operation sequence
516 QUEUE_ORDSEQ_STARTED = 0x01, /* flushing in progress */
517 QUEUE_ORDSEQ_DRAIN = 0x02, /* waiting for the queue to be drained */
518 QUEUE_ORDSEQ_PREFLUSH = 0x04, /* pre-flushing in progress */
519 QUEUE_ORDSEQ_BAR = 0x08, /* original barrier req in progress */
520 QUEUE_ORDSEQ_POSTFLUSH = 0x10, /* post-flushing in progress */
521 QUEUE_ORDSEQ_DONE = 0x20,
524 #define blk_queue_plugged(q) test_bit(QUEUE_FLAG_PLUGGED, &(q)->queue_flags)
525 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
526 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
527 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
528 #define blk_queue_noxmerges(q) \
529 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
530 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
531 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
532 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
533 #define blk_queue_flushing(q) ((q)->ordseq)
534 #define blk_queue_stackable(q) \
535 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
536 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
537 #define blk_queue_secdiscard(q) (blk_queue_discard(q) && \
538 test_bit(QUEUE_FLAG_SECDISCARD, &(q)->queue_flags))
540 #define blk_noretry_request(rq) \
541 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
542 REQ_FAILFAST_DRIVER))
544 #define blk_account_rq(rq) \
545 (((rq)->cmd_flags & REQ_STARTED) && \
546 ((rq)->cmd_type == REQ_TYPE_FS || \
547 ((rq)->cmd_flags & REQ_DISCARD)))
549 #define blk_pm_request(rq) \
550 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
551 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
553 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
554 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
555 /* rq->queuelist of dequeued request must be list_empty() */
556 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
558 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
560 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
563 * We regard a request as sync, if either a read or a sync write
565 static inline bool rw_is_sync(unsigned int rw_flags)
567 return !(rw_flags & REQ_WRITE) || (rw_flags & REQ_SYNC);
570 static inline bool rq_is_sync(struct request *rq)
572 return rw_is_sync(rq->cmd_flags);
575 static inline int blk_queue_full(struct request_queue *q, int sync)
578 return test_bit(QUEUE_FLAG_SYNCFULL, &q->queue_flags);
579 return test_bit(QUEUE_FLAG_ASYNCFULL, &q->queue_flags);
582 static inline void blk_set_queue_full(struct request_queue *q, int sync)
585 queue_flag_set(QUEUE_FLAG_SYNCFULL, q);
587 queue_flag_set(QUEUE_FLAG_ASYNCFULL, q);
590 static inline void blk_clear_queue_full(struct request_queue *q, int sync)
593 queue_flag_clear(QUEUE_FLAG_SYNCFULL, q);
595 queue_flag_clear(QUEUE_FLAG_ASYNCFULL, q);
600 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
601 * it already be started by driver.
603 #define RQ_NOMERGE_FLAGS \
604 (REQ_NOMERGE | REQ_STARTED | REQ_HARDBARRIER | REQ_SOFTBARRIER)
605 #define rq_mergeable(rq) \
606 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
607 (((rq)->cmd_flags & REQ_DISCARD) || \
608 (rq)->cmd_type == REQ_TYPE_FS))
611 * q->prep_rq_fn return values
613 #define BLKPREP_OK 0 /* serve it */
614 #define BLKPREP_KILL 1 /* fatal error, kill */
615 #define BLKPREP_DEFER 2 /* leave on queue */
617 extern unsigned long blk_max_low_pfn, blk_max_pfn;
620 * standard bounce addresses:
622 * BLK_BOUNCE_HIGH : bounce all highmem pages
623 * BLK_BOUNCE_ANY : don't bounce anything
624 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
627 #if BITS_PER_LONG == 32
628 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
630 #define BLK_BOUNCE_HIGH -1ULL
632 #define BLK_BOUNCE_ANY (-1ULL)
633 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
636 * default timeout for SG_IO if none specified
638 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
639 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
642 extern int init_emergency_isa_pool(void);
643 extern void blk_queue_bounce(struct request_queue *q, struct bio **bio);
645 static inline int init_emergency_isa_pool(void)
649 static inline void blk_queue_bounce(struct request_queue *q, struct bio **bio)
652 #endif /* CONFIG_MMU */
658 unsigned long offset;
663 struct req_iterator {
668 /* This should not be used directly - use rq_for_each_segment */
669 #define for_each_bio(_bio) \
670 for (; _bio; _bio = _bio->bi_next)
671 #define __rq_for_each_bio(_bio, rq) \
673 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
675 #define rq_for_each_segment(bvl, _rq, _iter) \
676 __rq_for_each_bio(_iter.bio, _rq) \
677 bio_for_each_segment(bvl, _iter.bio, _iter.i)
679 #define rq_iter_last(rq, _iter) \
680 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
682 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
683 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
685 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
686 extern void rq_flush_dcache_pages(struct request *rq);
688 static inline void rq_flush_dcache_pages(struct request *rq)
693 extern int blk_register_queue(struct gendisk *disk);
694 extern void blk_unregister_queue(struct gendisk *disk);
695 extern void register_disk(struct gendisk *dev);
696 extern void generic_make_request(struct bio *bio);
697 extern void blk_rq_init(struct request_queue *q, struct request *rq);
698 extern void blk_put_request(struct request *);
699 extern void __blk_put_request(struct request_queue *, struct request *);
700 extern struct request *blk_get_request(struct request_queue *, int, gfp_t);
701 extern struct request *blk_make_request(struct request_queue *, struct bio *,
703 extern void blk_insert_request(struct request_queue *, struct request *, int, void *);
704 extern void blk_requeue_request(struct request_queue *, struct request *);
705 extern void blk_add_request_payload(struct request *rq, struct page *page,
707 extern int blk_rq_check_limits(struct request_queue *q, struct request *rq);
708 extern int blk_lld_busy(struct request_queue *q);
709 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
710 struct bio_set *bs, gfp_t gfp_mask,
711 int (*bio_ctr)(struct bio *, struct bio *, void *),
713 extern void blk_rq_unprep_clone(struct request *rq);
714 extern int blk_insert_cloned_request(struct request_queue *q,
716 extern void blk_plug_device(struct request_queue *);
717 extern void blk_plug_device_unlocked(struct request_queue *);
718 extern int blk_remove_plug(struct request_queue *);
719 extern void blk_recount_segments(struct request_queue *, struct bio *);
720 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
721 unsigned int, void __user *);
722 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
723 struct scsi_ioctl_command __user *);
726 * A queue has just exitted congestion. Note this in the global counter of
727 * congested queues, and wake up anyone who was waiting for requests to be
730 static inline void blk_clear_queue_congested(struct request_queue *q, int sync)
732 clear_bdi_congested(&q->backing_dev_info, sync);
736 * A queue has just entered congestion. Flag that in the queue's VM-visible
737 * state flags and increment the global gounter of congested queues.
739 static inline void blk_set_queue_congested(struct request_queue *q, int sync)
741 set_bdi_congested(&q->backing_dev_info, sync);
744 extern void blk_start_queue(struct request_queue *q);
745 extern void blk_stop_queue(struct request_queue *q);
746 extern void blk_sync_queue(struct request_queue *q);
747 extern void __blk_stop_queue(struct request_queue *q);
748 extern void __blk_run_queue(struct request_queue *);
749 extern void blk_run_queue(struct request_queue *);
750 extern int blk_rq_map_user(struct request_queue *, struct request *,
751 struct rq_map_data *, void __user *, unsigned long,
753 extern int blk_rq_unmap_user(struct bio *);
754 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
755 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
756 struct rq_map_data *, struct sg_iovec *, int,
757 unsigned int, gfp_t);
758 extern int blk_execute_rq(struct request_queue *, struct gendisk *,
759 struct request *, int);
760 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
761 struct request *, int, rq_end_io_fn *);
762 extern void blk_unplug(struct request_queue *q);
764 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
766 return bdev->bd_disk->queue;
770 * blk_rq_pos() : the current sector
771 * blk_rq_bytes() : bytes left in the entire request
772 * blk_rq_cur_bytes() : bytes left in the current segment
773 * blk_rq_err_bytes() : bytes left till the next error boundary
774 * blk_rq_sectors() : sectors left in the entire request
775 * blk_rq_cur_sectors() : sectors left in the current segment
777 static inline sector_t blk_rq_pos(const struct request *rq)
782 static inline unsigned int blk_rq_bytes(const struct request *rq)
784 return rq->__data_len;
787 static inline int blk_rq_cur_bytes(const struct request *rq)
789 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
792 extern unsigned int blk_rq_err_bytes(const struct request *rq);
794 static inline unsigned int blk_rq_sectors(const struct request *rq)
796 return blk_rq_bytes(rq) >> 9;
799 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
801 return blk_rq_cur_bytes(rq) >> 9;
805 * Request issue related functions.
807 extern struct request *blk_peek_request(struct request_queue *q);
808 extern void blk_start_request(struct request *rq);
809 extern struct request *blk_fetch_request(struct request_queue *q);
812 * Request completion related functions.
814 * blk_update_request() completes given number of bytes and updates
815 * the request without completing it.
817 * blk_end_request() and friends. __blk_end_request() must be called
818 * with the request queue spinlock acquired.
820 * Several drivers define their own end_request and call
821 * blk_end_request() for parts of the original function.
822 * This prevents code duplication in drivers.
824 extern bool blk_update_request(struct request *rq, int error,
825 unsigned int nr_bytes);
826 extern bool blk_end_request(struct request *rq, int error,
827 unsigned int nr_bytes);
828 extern void blk_end_request_all(struct request *rq, int error);
829 extern bool blk_end_request_cur(struct request *rq, int error);
830 extern bool blk_end_request_err(struct request *rq, int error);
831 extern bool __blk_end_request(struct request *rq, int error,
832 unsigned int nr_bytes);
833 extern void __blk_end_request_all(struct request *rq, int error);
834 extern bool __blk_end_request_cur(struct request *rq, int error);
835 extern bool __blk_end_request_err(struct request *rq, int error);
837 extern void blk_complete_request(struct request *);
838 extern void __blk_complete_request(struct request *);
839 extern void blk_abort_request(struct request *);
840 extern void blk_abort_queue(struct request_queue *);
841 extern void blk_unprep_request(struct request *);
844 * Access functions for manipulating queue properties
846 extern struct request_queue *blk_init_queue_node(request_fn_proc *rfn,
847 spinlock_t *lock, int node_id);
848 extern struct request_queue *blk_init_allocated_queue_node(struct request_queue *,
850 spinlock_t *, int node_id);
851 extern struct request_queue *blk_init_queue(request_fn_proc *, spinlock_t *);
852 extern struct request_queue *blk_init_allocated_queue(struct request_queue *,
853 request_fn_proc *, spinlock_t *);
854 extern void blk_cleanup_queue(struct request_queue *);
855 extern void blk_queue_make_request(struct request_queue *, make_request_fn *);
856 extern void blk_queue_bounce_limit(struct request_queue *, u64);
857 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
858 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
859 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
860 extern void blk_queue_max_discard_sectors(struct request_queue *q,
861 unsigned int max_discard_sectors);
862 extern void blk_queue_logical_block_size(struct request_queue *, unsigned short);
863 extern void blk_queue_physical_block_size(struct request_queue *, unsigned short);
864 extern void blk_queue_alignment_offset(struct request_queue *q,
865 unsigned int alignment);
866 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
867 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
868 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
869 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
870 extern void blk_set_default_limits(struct queue_limits *lim);
871 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
873 extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev,
875 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
877 extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b);
878 extern void blk_queue_dma_pad(struct request_queue *, unsigned int);
879 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
880 extern int blk_queue_dma_drain(struct request_queue *q,
881 dma_drain_needed_fn *dma_drain_needed,
882 void *buf, unsigned int size);
883 extern void blk_queue_lld_busy(struct request_queue *q, lld_busy_fn *fn);
884 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
885 extern void blk_queue_prep_rq(struct request_queue *, prep_rq_fn *pfn);
886 extern void blk_queue_unprep_rq(struct request_queue *, unprep_rq_fn *ufn);
887 extern void blk_queue_merge_bvec(struct request_queue *, merge_bvec_fn *);
888 extern void blk_queue_dma_alignment(struct request_queue *, int);
889 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
890 extern void blk_queue_softirq_done(struct request_queue *, softirq_done_fn *);
891 extern void blk_queue_rq_timed_out(struct request_queue *, rq_timed_out_fn *);
892 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
893 extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
894 extern int blk_queue_ordered(struct request_queue *, unsigned);
895 extern bool blk_do_ordered(struct request_queue *, struct request **);
896 extern unsigned blk_ordered_cur_seq(struct request_queue *);
897 extern unsigned blk_ordered_req_seq(struct request *);
898 extern bool blk_ordered_complete_seq(struct request_queue *, unsigned, int);
900 extern int blk_rq_map_sg(struct request_queue *, struct request *, struct scatterlist *);
901 extern void blk_dump_rq_flags(struct request *, char *);
902 extern void generic_unplug_device(struct request_queue *);
903 extern long nr_blockdev_pages(void);
905 int blk_get_queue(struct request_queue *);
906 struct request_queue *blk_alloc_queue(gfp_t);
907 struct request_queue *blk_alloc_queue_node(gfp_t, int);
908 extern void blk_put_queue(struct request_queue *);
913 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
914 extern int blk_queue_start_tag(struct request_queue *, struct request *);
915 extern struct request *blk_queue_find_tag(struct request_queue *, int);
916 extern void blk_queue_end_tag(struct request_queue *, struct request *);
917 extern int blk_queue_init_tags(struct request_queue *, int, struct blk_queue_tag *);
918 extern void blk_queue_free_tags(struct request_queue *);
919 extern int blk_queue_resize_tags(struct request_queue *, int);
920 extern void blk_queue_invalidate_tags(struct request_queue *);
921 extern struct blk_queue_tag *blk_init_tags(int);
922 extern void blk_free_tags(struct blk_queue_tag *);
924 static inline struct request *blk_map_queue_find_tag(struct blk_queue_tag *bqt,
927 if (unlikely(bqt == NULL || tag >= bqt->real_max_depth))
929 return bqt->tag_index[tag];
932 BLKDEV_WAIT, /* wait for completion */
933 BLKDEV_BARRIER, /* issue request with barrier */
934 BLKDEV_SECURE, /* secure discard */
936 #define BLKDEV_IFL_WAIT (1 << BLKDEV_WAIT)
937 #define BLKDEV_IFL_BARRIER (1 << BLKDEV_BARRIER)
938 #define BLKDEV_IFL_SECURE (1 << BLKDEV_SECURE)
939 extern int blkdev_issue_flush(struct block_device *, gfp_t, sector_t *,
941 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
942 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
943 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
944 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
945 static inline int sb_issue_discard(struct super_block *sb,
946 sector_t block, sector_t nr_blocks)
948 block <<= (sb->s_blocksize_bits - 9);
949 nr_blocks <<= (sb->s_blocksize_bits - 9);
950 return blkdev_issue_discard(sb->s_bdev, block, nr_blocks, GFP_NOFS,
951 BLKDEV_IFL_WAIT | BLKDEV_IFL_BARRIER);
954 extern int blk_verify_command(unsigned char *cmd, fmode_t has_write_perm);
956 enum blk_default_limits {
957 BLK_MAX_SEGMENTS = 128,
958 BLK_SAFE_MAX_SECTORS = 255,
959 BLK_DEF_MAX_SECTORS = 1024,
960 BLK_MAX_SEGMENT_SIZE = 65536,
961 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
964 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
966 static inline unsigned long queue_bounce_pfn(struct request_queue *q)
968 return q->limits.bounce_pfn;
971 static inline unsigned long queue_segment_boundary(struct request_queue *q)
973 return q->limits.seg_boundary_mask;
976 static inline unsigned int queue_max_sectors(struct request_queue *q)
978 return q->limits.max_sectors;
981 static inline unsigned int queue_max_hw_sectors(struct request_queue *q)
983 return q->limits.max_hw_sectors;
986 static inline unsigned short queue_max_segments(struct request_queue *q)
988 return q->limits.max_segments;
991 static inline unsigned int queue_max_segment_size(struct request_queue *q)
993 return q->limits.max_segment_size;
996 static inline unsigned short queue_logical_block_size(struct request_queue *q)
1000 if (q && q->limits.logical_block_size)
1001 retval = q->limits.logical_block_size;
1006 static inline unsigned short bdev_logical_block_size(struct block_device *bdev)
1008 return queue_logical_block_size(bdev_get_queue(bdev));
1011 static inline unsigned int queue_physical_block_size(struct request_queue *q)
1013 return q->limits.physical_block_size;
1016 static inline int bdev_physical_block_size(struct block_device *bdev)
1018 return queue_physical_block_size(bdev_get_queue(bdev));
1021 static inline unsigned int queue_io_min(struct request_queue *q)
1023 return q->limits.io_min;
1026 static inline int bdev_io_min(struct block_device *bdev)
1028 return queue_io_min(bdev_get_queue(bdev));
1031 static inline unsigned int queue_io_opt(struct request_queue *q)
1033 return q->limits.io_opt;
1036 static inline int bdev_io_opt(struct block_device *bdev)
1038 return queue_io_opt(bdev_get_queue(bdev));
1041 static inline int queue_alignment_offset(struct request_queue *q)
1043 if (q->limits.misaligned)
1046 return q->limits.alignment_offset;
1049 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1051 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1052 unsigned int alignment = (sector << 9) & (granularity - 1);
1054 return (granularity + lim->alignment_offset - alignment)
1055 & (granularity - 1);
1058 static inline int bdev_alignment_offset(struct block_device *bdev)
1060 struct request_queue *q = bdev_get_queue(bdev);
1062 if (q->limits.misaligned)
1065 if (bdev != bdev->bd_contains)
1066 return bdev->bd_part->alignment_offset;
1068 return q->limits.alignment_offset;
1071 static inline int queue_discard_alignment(struct request_queue *q)
1073 if (q->limits.discard_misaligned)
1076 return q->limits.discard_alignment;
1079 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1081 unsigned int alignment = (sector << 9) & (lim->discard_granularity - 1);
1083 return (lim->discard_granularity + lim->discard_alignment - alignment)
1084 & (lim->discard_granularity - 1);
1087 static inline unsigned int queue_discard_zeroes_data(struct request_queue *q)
1089 if (q->limits.discard_zeroes_data == 1)
1095 static inline unsigned int bdev_discard_zeroes_data(struct block_device *bdev)
1097 return queue_discard_zeroes_data(bdev_get_queue(bdev));
1100 static inline int queue_dma_alignment(struct request_queue *q)
1102 return q ? q->dma_alignment : 511;
1105 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1108 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1109 return !(addr & alignment) && !(len & alignment);
1112 /* assumes size > 256 */
1113 static inline unsigned int blksize_bits(unsigned int size)
1115 unsigned int bits = 8;
1119 } while (size > 256);
1123 static inline unsigned int block_size(struct block_device *bdev)
1125 return bdev->bd_block_size;
1128 typedef struct {struct page *v;} Sector;
1130 unsigned char *read_dev_sector(struct block_device *, sector_t, Sector *);
1132 static inline void put_dev_sector(Sector p)
1134 page_cache_release(p.v);
1138 int kblockd_schedule_work(struct request_queue *q, struct work_struct *work);
1139 int kblockd_schedule_delayed_work(struct request_queue *q, struct delayed_work *dwork, unsigned long delay);
1141 #ifdef CONFIG_BLK_CGROUP
1143 * This should not be using sched_clock(). A real patch is in progress
1144 * to fix this up, until that is in place we need to disable preemption
1145 * around sched_clock() in this function and set_io_start_time_ns().
1147 static inline void set_start_time_ns(struct request *req)
1150 req->start_time_ns = sched_clock();
1154 static inline void set_io_start_time_ns(struct request *req)
1157 req->io_start_time_ns = sched_clock();
1161 static inline uint64_t rq_start_time_ns(struct request *req)
1163 return req->start_time_ns;
1166 static inline uint64_t rq_io_start_time_ns(struct request *req)
1168 return req->io_start_time_ns;
1171 static inline void set_start_time_ns(struct request *req) {}
1172 static inline void set_io_start_time_ns(struct request *req) {}
1173 static inline uint64_t rq_start_time_ns(struct request *req)
1177 static inline uint64_t rq_io_start_time_ns(struct request *req)
1183 #ifdef CONFIG_BLK_DEV_THROTTLING
1184 extern int blk_throtl_init(struct request_queue *q);
1185 extern void blk_throtl_exit(struct request_queue *q);
1186 extern int blk_throtl_bio(struct request_queue *q, struct bio **bio);
1187 extern void throtl_schedule_delayed_work(struct request_queue *q, unsigned long delay);
1188 extern void throtl_shutdown_timer_wq(struct request_queue *q);
1189 #else /* CONFIG_BLK_DEV_THROTTLING */
1190 static inline int blk_throtl_bio(struct request_queue *q, struct bio **bio)
1195 static inline int blk_throtl_init(struct request_queue *q) { return 0; }
1196 static inline int blk_throtl_exit(struct request_queue *q) { return 0; }
1197 static inline void throtl_schedule_delayed_work(struct request_queue *q, unsigned long delay) {}
1198 static inline void throtl_shutdown_timer_wq(struct request_queue *q) {}
1199 #endif /* CONFIG_BLK_DEV_THROTTLING */
1201 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1202 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1203 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1204 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1206 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1208 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1209 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1211 struct blk_integrity_exchg {
1215 unsigned int data_size;
1216 unsigned short sector_size;
1217 const char *disk_name;
1220 typedef void (integrity_gen_fn) (struct blk_integrity_exchg *);
1221 typedef int (integrity_vrfy_fn) (struct blk_integrity_exchg *);
1222 typedef void (integrity_set_tag_fn) (void *, void *, unsigned int);
1223 typedef void (integrity_get_tag_fn) (void *, void *, unsigned int);
1225 struct blk_integrity {
1226 integrity_gen_fn *generate_fn;
1227 integrity_vrfy_fn *verify_fn;
1228 integrity_set_tag_fn *set_tag_fn;
1229 integrity_get_tag_fn *get_tag_fn;
1231 unsigned short flags;
1232 unsigned short tuple_size;
1233 unsigned short sector_size;
1234 unsigned short tag_size;
1238 struct kobject kobj;
1241 extern int blk_integrity_register(struct gendisk *, struct blk_integrity *);
1242 extern void blk_integrity_unregister(struct gendisk *);
1243 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1244 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1245 struct scatterlist *);
1246 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1247 extern int blk_integrity_merge_rq(struct request_queue *, struct request *,
1249 extern int blk_integrity_merge_bio(struct request_queue *, struct request *,
1253 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1255 return bdev->bd_disk->integrity;
1258 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1260 return disk->integrity;
1263 static inline int blk_integrity_rq(struct request *rq)
1265 if (rq->bio == NULL)
1268 return bio_integrity(rq->bio);
1271 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1274 q->limits.max_integrity_segments = segs;
1277 static inline unsigned short
1278 queue_max_integrity_segments(struct request_queue *q)
1280 return q->limits.max_integrity_segments;
1283 #else /* CONFIG_BLK_DEV_INTEGRITY */
1285 #define blk_integrity_rq(rq) (0)
1286 #define blk_rq_count_integrity_sg(a, b) (0)
1287 #define blk_rq_map_integrity_sg(a, b, c) (0)
1288 #define bdev_get_integrity(a) (0)
1289 #define blk_get_integrity(a) (0)
1290 #define blk_integrity_compare(a, b) (0)
1291 #define blk_integrity_register(a, b) (0)
1292 #define blk_integrity_unregister(a) do { } while (0);
1293 #define blk_queue_max_integrity_segments(a, b) do { } while (0);
1294 #define queue_max_integrity_segments(a) (0)
1295 #define blk_integrity_merge_rq(a, b, c) (0)
1296 #define blk_integrity_merge_bio(a, b, c) (0)
1298 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1300 struct block_device_operations {
1301 int (*open) (struct block_device *, fmode_t);
1302 int (*release) (struct gendisk *, fmode_t);
1303 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1304 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1305 int (*direct_access) (struct block_device *, sector_t,
1306 void **, unsigned long *);
1307 int (*media_changed) (struct gendisk *);
1308 void (*unlock_native_capacity) (struct gendisk *);
1309 int (*revalidate_disk) (struct gendisk *);
1310 int (*getgeo)(struct block_device *, struct hd_geometry *);
1311 /* this callback is with swap_lock and sometimes page table lock held */
1312 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1313 struct module *owner;
1316 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1318 #else /* CONFIG_BLOCK */
1320 * stubs for when the block layer is configured out
1322 #define buffer_heads_over_limit 0
1324 static inline long nr_blockdev_pages(void)
1329 #endif /* CONFIG_BLOCK */