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 */
63 REQ_TYPE_LINUX_BLOCK, /* generic block layer message */
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,
74 * For request of type REQ_TYPE_LINUX_BLOCK, rq->cmd[0] is the opcode being
75 * sent down (similar to how REQ_TYPE_BLOCK_PC means that ->cmd[] holds a
78 * 0x00 -> 0x3f are driver private, to be used for whatever purpose they need,
79 * typically to differentiate REQ_TYPE_SPECIAL requests.
83 REQ_LB_OP_EJECT = 0x40, /* eject request */
84 REQ_LB_OP_FLUSH = 0x41, /* flush request */
87 #define BLK_MAX_CDB 16
90 * try to put the fields that are referenced together in the same cacheline.
91 * if you modify this structure, be sure to check block/blk-core.c:rq_init()
95 struct list_head queuelist;
96 struct call_single_data csd;
98 struct request_queue *q;
100 unsigned int cmd_flags;
101 enum rq_cmd_type_bits cmd_type;
102 unsigned long atomic_flags;
106 /* the following two fields are internal, NEVER access directly */
107 unsigned int __data_len; /* total data len */
108 sector_t __sector; /* sector cursor */
113 struct hlist_node hash; /* merge hash */
115 * The rb_node is only used inside the io scheduler, requests
116 * are pruned when moved to the dispatch queue. So let the
117 * completion_data share space with the rb_node.
120 struct rb_node rb_node; /* sort/lookup */
121 void *completion_data;
125 * Three pointers are available for the IO schedulers, if they need
126 * more they have to dynamically allocate it.
128 void *elevator_private;
129 void *elevator_private2;
130 void *elevator_private3;
132 struct gendisk *rq_disk;
133 unsigned long start_time;
134 #ifdef CONFIG_BLK_CGROUP
135 unsigned long long start_time_ns;
136 unsigned long long io_start_time_ns; /* when passed to hardware */
138 /* Number of scatter-gather DMA addr+len pairs after
139 * physical address coalescing is performed.
141 unsigned short nr_phys_segments;
143 unsigned short ioprio;
147 void *special; /* opaque pointer available for LLD use */
148 char *buffer; /* kaddr of the current segment if available */
154 * when request is used as a packet command carrier
156 unsigned char __cmd[BLK_MAX_CDB];
158 unsigned short cmd_len;
160 unsigned int extra_len; /* length of alignment and padding */
161 unsigned int sense_len;
162 unsigned int resid_len; /* residual count */
165 unsigned long deadline;
166 struct list_head timeout_list;
167 unsigned int timeout;
171 * completion callback.
173 rq_end_io_fn *end_io;
177 struct request *next_rq;
180 static inline unsigned short req_get_ioprio(struct request *req)
186 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
187 * requests. Some step values could eventually be made generic.
189 struct request_pm_state
191 /* PM state machine step value, currently driver specific */
193 /* requested PM state value (S1, S2, S3, S4, ...) */
195 void* data; /* for driver use */
198 #include <linux/elevator.h>
200 typedef void (request_fn_proc) (struct request_queue *q);
201 typedef int (make_request_fn) (struct request_queue *q, struct bio *bio);
202 typedef int (prep_rq_fn) (struct request_queue *, struct request *);
203 typedef void (unplug_fn) (struct request_queue *);
206 struct bvec_merge_data {
207 struct block_device *bi_bdev;
212 typedef int (merge_bvec_fn) (struct request_queue *, struct bvec_merge_data *,
214 typedef void (prepare_flush_fn) (struct request_queue *, struct request *);
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);
219 enum blk_eh_timer_return {
225 typedef enum blk_eh_timer_return (rq_timed_out_fn)(struct request *);
227 enum blk_queue_state {
232 struct blk_queue_tag {
233 struct request **tag_index; /* map of busy tags */
234 unsigned long *tag_map; /* bit map of free/busy tags */
235 int busy; /* current depth */
236 int max_depth; /* what we will send to device */
237 int real_max_depth; /* what the array can hold */
238 atomic_t refcnt; /* map can be shared */
241 #define BLK_SCSI_MAX_CMDS (256)
242 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
244 struct queue_limits {
245 unsigned long bounce_pfn;
246 unsigned long seg_boundary_mask;
248 unsigned int max_hw_sectors;
249 unsigned int max_sectors;
250 unsigned int max_segment_size;
251 unsigned int physical_block_size;
252 unsigned int alignment_offset;
255 unsigned int max_discard_sectors;
256 unsigned int discard_granularity;
257 unsigned int discard_alignment;
259 unsigned short logical_block_size;
260 unsigned short max_segments;
262 unsigned char misaligned;
263 unsigned char discard_misaligned;
264 unsigned char no_cluster;
265 signed char discard_zeroes_data;
271 * Together with queue_head for cacheline sharing
273 struct list_head queue_head;
274 struct request *last_merge;
275 struct elevator_queue *elevator;
278 * the queue request freelist, one for reads and one for writes
280 struct request_list rq;
282 request_fn_proc *request_fn;
283 make_request_fn *make_request_fn;
284 prep_rq_fn *prep_rq_fn;
285 unplug_fn *unplug_fn;
286 merge_bvec_fn *merge_bvec_fn;
287 prepare_flush_fn *prepare_flush_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 * Auto-unplugging state
302 struct timer_list unplug_timer;
303 int unplug_thresh; /* After this many requests */
304 unsigned long unplug_delay; /* After this many jiffies */
305 struct work_struct unplug_work;
307 struct backing_dev_info backing_dev_info;
310 * The queue owner gets to use this for whatever they like.
311 * ll_rw_blk doesn't touch it.
316 * queue needs bounce pages for pages above this limit
321 * various queue flags, see QUEUE_* below
323 unsigned long queue_flags;
326 * protects queue structures from reentrancy. ->__queue_lock should
327 * _never_ be used directly, it is queue private. always use
330 spinlock_t __queue_lock;
331 spinlock_t *queue_lock;
341 unsigned long nr_requests; /* Max # of requests */
342 unsigned int nr_congestion_on;
343 unsigned int nr_congestion_off;
344 unsigned int nr_batching;
346 void *dma_drain_buffer;
347 unsigned int dma_drain_size;
348 unsigned int dma_pad_mask;
349 unsigned int dma_alignment;
351 struct blk_queue_tag *queue_tags;
352 struct list_head tag_busy_list;
354 unsigned int nr_sorted;
355 unsigned int in_flight[2];
357 unsigned int rq_timeout;
358 struct timer_list timeout;
359 struct list_head timeout_list;
361 struct queue_limits limits;
366 unsigned int sg_timeout;
367 unsigned int sg_reserved_size;
369 #ifdef CONFIG_BLK_DEV_IO_TRACE
370 struct blk_trace *blk_trace;
373 * reserved for flush operations
375 unsigned int ordered, next_ordered, ordseq;
376 int orderr, ordcolor;
377 struct request pre_flush_rq, bar_rq, post_flush_rq;
378 struct request *orig_bar_rq;
380 struct mutex sysfs_lock;
382 #if defined(CONFIG_BLK_DEV_BSG)
383 struct bsg_class_device bsg_dev;
387 #define QUEUE_FLAG_CLUSTER 0 /* cluster several segments into 1 */
388 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
389 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
390 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
391 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
392 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
393 #define QUEUE_FLAG_REENTER 6 /* Re-entrancy avoidance */
394 #define QUEUE_FLAG_PLUGGED 7 /* queue is plugged */
395 #define QUEUE_FLAG_ELVSWITCH 8 /* don't use elevator, just do FIFO */
396 #define QUEUE_FLAG_BIDI 9 /* queue supports bidi requests */
397 #define QUEUE_FLAG_NOMERGES 10 /* disable merge attempts */
398 #define QUEUE_FLAG_SAME_COMP 11 /* force complete on same CPU */
399 #define QUEUE_FLAG_FAIL_IO 12 /* fake timeout */
400 #define QUEUE_FLAG_STACKABLE 13 /* supports request stacking */
401 #define QUEUE_FLAG_NONROT 14 /* non-rotational device (SSD) */
402 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
403 #define QUEUE_FLAG_IO_STAT 15 /* do IO stats */
404 #define QUEUE_FLAG_DISCARD 16 /* supports DISCARD */
405 #define QUEUE_FLAG_NOXMERGES 17 /* No extended merges */
406 #define QUEUE_FLAG_ADD_RANDOM 18 /* Contributes to random pool */
408 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
409 (1 << QUEUE_FLAG_CLUSTER) | \
410 (1 << QUEUE_FLAG_STACKABLE) | \
411 (1 << QUEUE_FLAG_SAME_COMP) | \
412 (1 << QUEUE_FLAG_ADD_RANDOM))
414 static inline int queue_is_locked(struct request_queue *q)
417 spinlock_t *lock = q->queue_lock;
418 return lock && spin_is_locked(lock);
424 static inline void queue_flag_set_unlocked(unsigned int flag,
425 struct request_queue *q)
427 __set_bit(flag, &q->queue_flags);
430 static inline int queue_flag_test_and_clear(unsigned int flag,
431 struct request_queue *q)
433 WARN_ON_ONCE(!queue_is_locked(q));
435 if (test_bit(flag, &q->queue_flags)) {
436 __clear_bit(flag, &q->queue_flags);
443 static inline int queue_flag_test_and_set(unsigned int flag,
444 struct request_queue *q)
446 WARN_ON_ONCE(!queue_is_locked(q));
448 if (!test_bit(flag, &q->queue_flags)) {
449 __set_bit(flag, &q->queue_flags);
456 static inline void queue_flag_set(unsigned int flag, struct request_queue *q)
458 WARN_ON_ONCE(!queue_is_locked(q));
459 __set_bit(flag, &q->queue_flags);
462 static inline void queue_flag_clear_unlocked(unsigned int flag,
463 struct request_queue *q)
465 __clear_bit(flag, &q->queue_flags);
468 static inline int queue_in_flight(struct request_queue *q)
470 return q->in_flight[0] + q->in_flight[1];
473 static inline void queue_flag_clear(unsigned int flag, struct request_queue *q)
475 WARN_ON_ONCE(!queue_is_locked(q));
476 __clear_bit(flag, &q->queue_flags);
481 * Hardbarrier is supported with one of the following methods.
483 * NONE : hardbarrier unsupported
484 * DRAIN : ordering by draining is enough
485 * DRAIN_FLUSH : ordering by draining w/ pre and post flushes
486 * DRAIN_FUA : ordering by draining w/ pre flush and FUA write
487 * TAG : ordering by tag is enough
488 * TAG_FLUSH : ordering by tag w/ pre and post flushes
489 * TAG_FUA : ordering by tag w/ pre flush and FUA write
491 QUEUE_ORDERED_BY_DRAIN = 0x01,
492 QUEUE_ORDERED_BY_TAG = 0x02,
493 QUEUE_ORDERED_DO_PREFLUSH = 0x10,
494 QUEUE_ORDERED_DO_BAR = 0x20,
495 QUEUE_ORDERED_DO_POSTFLUSH = 0x40,
496 QUEUE_ORDERED_DO_FUA = 0x80,
498 QUEUE_ORDERED_NONE = 0x00,
500 QUEUE_ORDERED_DRAIN = QUEUE_ORDERED_BY_DRAIN |
501 QUEUE_ORDERED_DO_BAR,
502 QUEUE_ORDERED_DRAIN_FLUSH = QUEUE_ORDERED_DRAIN |
503 QUEUE_ORDERED_DO_PREFLUSH |
504 QUEUE_ORDERED_DO_POSTFLUSH,
505 QUEUE_ORDERED_DRAIN_FUA = QUEUE_ORDERED_DRAIN |
506 QUEUE_ORDERED_DO_PREFLUSH |
507 QUEUE_ORDERED_DO_FUA,
509 QUEUE_ORDERED_TAG = QUEUE_ORDERED_BY_TAG |
510 QUEUE_ORDERED_DO_BAR,
511 QUEUE_ORDERED_TAG_FLUSH = QUEUE_ORDERED_TAG |
512 QUEUE_ORDERED_DO_PREFLUSH |
513 QUEUE_ORDERED_DO_POSTFLUSH,
514 QUEUE_ORDERED_TAG_FUA = QUEUE_ORDERED_TAG |
515 QUEUE_ORDERED_DO_PREFLUSH |
516 QUEUE_ORDERED_DO_FUA,
519 * Ordered operation sequence
521 QUEUE_ORDSEQ_STARTED = 0x01, /* flushing in progress */
522 QUEUE_ORDSEQ_DRAIN = 0x02, /* waiting for the queue to be drained */
523 QUEUE_ORDSEQ_PREFLUSH = 0x04, /* pre-flushing in progress */
524 QUEUE_ORDSEQ_BAR = 0x08, /* original barrier req in progress */
525 QUEUE_ORDSEQ_POSTFLUSH = 0x10, /* post-flushing in progress */
526 QUEUE_ORDSEQ_DONE = 0x20,
529 #define blk_queue_plugged(q) test_bit(QUEUE_FLAG_PLUGGED, &(q)->queue_flags)
530 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
531 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
532 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
533 #define blk_queue_noxmerges(q) \
534 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
535 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
536 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
537 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
538 #define blk_queue_flushing(q) ((q)->ordseq)
539 #define blk_queue_stackable(q) \
540 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
541 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
543 #define blk_noretry_request(rq) \
544 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
545 REQ_FAILFAST_DRIVER))
547 #define blk_account_rq(rq) \
548 (((rq)->cmd_flags & REQ_STARTED) && \
549 ((rq)->cmd_type == REQ_TYPE_FS || \
550 ((rq)->cmd_flags & REQ_DISCARD)))
552 #define blk_pm_request(rq) \
553 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
554 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
556 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
557 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
558 /* rq->queuelist of dequeued request must be list_empty() */
559 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
561 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
563 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
566 * We regard a request as sync, if either a read or a sync write
568 static inline bool rw_is_sync(unsigned int rw_flags)
570 return !(rw_flags & REQ_WRITE) || (rw_flags & REQ_SYNC);
573 static inline bool rq_is_sync(struct request *rq)
575 return rw_is_sync(rq->cmd_flags);
578 static inline int blk_queue_full(struct request_queue *q, int sync)
581 return test_bit(QUEUE_FLAG_SYNCFULL, &q->queue_flags);
582 return test_bit(QUEUE_FLAG_ASYNCFULL, &q->queue_flags);
585 static inline void blk_set_queue_full(struct request_queue *q, int sync)
588 queue_flag_set(QUEUE_FLAG_SYNCFULL, q);
590 queue_flag_set(QUEUE_FLAG_ASYNCFULL, q);
593 static inline void blk_clear_queue_full(struct request_queue *q, int sync)
596 queue_flag_clear(QUEUE_FLAG_SYNCFULL, q);
598 queue_flag_clear(QUEUE_FLAG_ASYNCFULL, q);
603 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
604 * it already be started by driver.
606 #define RQ_NOMERGE_FLAGS \
607 (REQ_NOMERGE | REQ_STARTED | REQ_HARDBARRIER | REQ_SOFTBARRIER)
608 #define rq_mergeable(rq) \
609 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
610 (((rq)->cmd_flags & REQ_DISCARD) || \
611 (rq)->cmd_type == REQ_TYPE_FS))
614 * q->prep_rq_fn return values
616 #define BLKPREP_OK 0 /* serve it */
617 #define BLKPREP_KILL 1 /* fatal error, kill */
618 #define BLKPREP_DEFER 2 /* leave on queue */
620 extern unsigned long blk_max_low_pfn, blk_max_pfn;
623 * standard bounce addresses:
625 * BLK_BOUNCE_HIGH : bounce all highmem pages
626 * BLK_BOUNCE_ANY : don't bounce anything
627 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
630 #if BITS_PER_LONG == 32
631 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
633 #define BLK_BOUNCE_HIGH -1ULL
635 #define BLK_BOUNCE_ANY (-1ULL)
636 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
639 * default timeout for SG_IO if none specified
641 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
642 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
645 extern int init_emergency_isa_pool(void);
646 extern void blk_queue_bounce(struct request_queue *q, struct bio **bio);
648 static inline int init_emergency_isa_pool(void)
652 static inline void blk_queue_bounce(struct request_queue *q, struct bio **bio)
655 #endif /* CONFIG_MMU */
661 unsigned long offset;
666 struct req_iterator {
671 /* This should not be used directly - use rq_for_each_segment */
672 #define for_each_bio(_bio) \
673 for (; _bio; _bio = _bio->bi_next)
674 #define __rq_for_each_bio(_bio, rq) \
676 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
678 #define rq_for_each_segment(bvl, _rq, _iter) \
679 __rq_for_each_bio(_iter.bio, _rq) \
680 bio_for_each_segment(bvl, _iter.bio, _iter.i)
682 #define rq_iter_last(rq, _iter) \
683 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
685 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
686 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
688 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
689 extern void rq_flush_dcache_pages(struct request *rq);
691 static inline void rq_flush_dcache_pages(struct request *rq)
696 extern int blk_register_queue(struct gendisk *disk);
697 extern void blk_unregister_queue(struct gendisk *disk);
698 extern void register_disk(struct gendisk *dev);
699 extern void generic_make_request(struct bio *bio);
700 extern void blk_rq_init(struct request_queue *q, struct request *rq);
701 extern void blk_put_request(struct request *);
702 extern void __blk_put_request(struct request_queue *, struct request *);
703 extern struct request *blk_get_request(struct request_queue *, int, gfp_t);
704 extern struct request *blk_make_request(struct request_queue *, struct bio *,
706 extern void blk_insert_request(struct request_queue *, struct request *, int, void *);
707 extern void blk_requeue_request(struct request_queue *, struct request *);
708 extern int blk_rq_check_limits(struct request_queue *q, struct request *rq);
709 extern int blk_lld_busy(struct request_queue *q);
710 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
711 struct bio_set *bs, gfp_t gfp_mask,
712 int (*bio_ctr)(struct bio *, struct bio *, void *),
714 extern void blk_rq_unprep_clone(struct request *rq);
715 extern int blk_insert_cloned_request(struct request_queue *q,
717 extern void blk_plug_device(struct request_queue *);
718 extern void blk_plug_device_unlocked(struct request_queue *);
719 extern int blk_remove_plug(struct request_queue *);
720 extern void blk_recount_segments(struct request_queue *, struct bio *);
721 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
722 unsigned int, void __user *);
723 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
724 struct scsi_ioctl_command __user *);
727 * A queue has just exitted congestion. Note this in the global counter of
728 * congested queues, and wake up anyone who was waiting for requests to be
731 static inline void blk_clear_queue_congested(struct request_queue *q, int sync)
733 clear_bdi_congested(&q->backing_dev_info, sync);
737 * A queue has just entered congestion. Flag that in the queue's VM-visible
738 * state flags and increment the global gounter of congested queues.
740 static inline void blk_set_queue_congested(struct request_queue *q, int sync)
742 set_bdi_congested(&q->backing_dev_info, sync);
745 extern void blk_start_queue(struct request_queue *q);
746 extern void blk_stop_queue(struct request_queue *q);
747 extern void blk_sync_queue(struct request_queue *q);
748 extern void __blk_stop_queue(struct request_queue *q);
749 extern void __blk_run_queue(struct request_queue *);
750 extern void blk_run_queue(struct request_queue *);
751 extern int blk_rq_map_user(struct request_queue *, struct request *,
752 struct rq_map_data *, void __user *, unsigned long,
754 extern int blk_rq_unmap_user(struct bio *);
755 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
756 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
757 struct rq_map_data *, struct sg_iovec *, int,
758 unsigned int, gfp_t);
759 extern int blk_execute_rq(struct request_queue *, struct gendisk *,
760 struct request *, int);
761 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
762 struct request *, int, rq_end_io_fn *);
763 extern void blk_unplug(struct request_queue *q);
765 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
767 return bdev->bd_disk->queue;
771 * blk_rq_pos() : the current sector
772 * blk_rq_bytes() : bytes left in the entire request
773 * blk_rq_cur_bytes() : bytes left in the current segment
774 * blk_rq_err_bytes() : bytes left till the next error boundary
775 * blk_rq_sectors() : sectors left in the entire request
776 * blk_rq_cur_sectors() : sectors left in the current segment
778 static inline sector_t blk_rq_pos(const struct request *rq)
783 static inline unsigned int blk_rq_bytes(const struct request *rq)
785 return rq->__data_len;
788 static inline int blk_rq_cur_bytes(const struct request *rq)
790 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
793 extern unsigned int blk_rq_err_bytes(const struct request *rq);
795 static inline unsigned int blk_rq_sectors(const struct request *rq)
797 return blk_rq_bytes(rq) >> 9;
800 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
802 return blk_rq_cur_bytes(rq) >> 9;
806 * Request issue related functions.
808 extern struct request *blk_peek_request(struct request_queue *q);
809 extern void blk_start_request(struct request *rq);
810 extern struct request *blk_fetch_request(struct request_queue *q);
813 * Request completion related functions.
815 * blk_update_request() completes given number of bytes and updates
816 * the request without completing it.
818 * blk_end_request() and friends. __blk_end_request() must be called
819 * with the request queue spinlock acquired.
821 * Several drivers define their own end_request and call
822 * blk_end_request() for parts of the original function.
823 * This prevents code duplication in drivers.
825 extern bool blk_update_request(struct request *rq, int error,
826 unsigned int nr_bytes);
827 extern bool blk_end_request(struct request *rq, int error,
828 unsigned int nr_bytes);
829 extern void blk_end_request_all(struct request *rq, int error);
830 extern bool blk_end_request_cur(struct request *rq, int error);
831 extern bool blk_end_request_err(struct request *rq, int error);
832 extern bool __blk_end_request(struct request *rq, int error,
833 unsigned int nr_bytes);
834 extern void __blk_end_request_all(struct request *rq, int error);
835 extern bool __blk_end_request_cur(struct request *rq, int error);
836 extern bool __blk_end_request_err(struct request *rq, int error);
838 extern void blk_complete_request(struct request *);
839 extern void __blk_complete_request(struct request *);
840 extern void blk_abort_request(struct request *);
841 extern void blk_abort_queue(struct request_queue *);
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_merge_bvec(struct request_queue *, merge_bvec_fn *);
887 extern void blk_queue_dma_alignment(struct request_queue *, int);
888 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
889 extern void blk_queue_softirq_done(struct request_queue *, softirq_done_fn *);
890 extern void blk_queue_rq_timed_out(struct request_queue *, rq_timed_out_fn *);
891 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
892 extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
893 extern int blk_queue_ordered(struct request_queue *, unsigned, prepare_flush_fn *);
894 extern bool blk_do_ordered(struct request_queue *, struct request **);
895 extern unsigned blk_ordered_cur_seq(struct request_queue *);
896 extern unsigned blk_ordered_req_seq(struct request *);
897 extern bool blk_ordered_complete_seq(struct request_queue *, unsigned, int);
899 extern int blk_rq_map_sg(struct request_queue *, struct request *, struct scatterlist *);
900 extern void blk_dump_rq_flags(struct request *, char *);
901 extern void generic_unplug_device(struct request_queue *);
902 extern long nr_blockdev_pages(void);
904 int blk_get_queue(struct request_queue *);
905 struct request_queue *blk_alloc_queue(gfp_t);
906 struct request_queue *blk_alloc_queue_node(gfp_t, int);
907 extern void blk_put_queue(struct request_queue *);
912 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
913 extern int blk_queue_start_tag(struct request_queue *, struct request *);
914 extern struct request *blk_queue_find_tag(struct request_queue *, int);
915 extern void blk_queue_end_tag(struct request_queue *, struct request *);
916 extern int blk_queue_init_tags(struct request_queue *, int, struct blk_queue_tag *);
917 extern void blk_queue_free_tags(struct request_queue *);
918 extern int blk_queue_resize_tags(struct request_queue *, int);
919 extern void blk_queue_invalidate_tags(struct request_queue *);
920 extern struct blk_queue_tag *blk_init_tags(int);
921 extern void blk_free_tags(struct blk_queue_tag *);
923 static inline struct request *blk_map_queue_find_tag(struct blk_queue_tag *bqt,
926 if (unlikely(bqt == NULL || tag >= bqt->real_max_depth))
928 return bqt->tag_index[tag];
931 BLKDEV_WAIT, /* wait for completion */
932 BLKDEV_BARRIER, /*issue request with barrier */
934 #define BLKDEV_IFL_WAIT (1 << BLKDEV_WAIT)
935 #define BLKDEV_IFL_BARRIER (1 << BLKDEV_BARRIER)
936 extern int blkdev_issue_flush(struct block_device *, gfp_t, sector_t *,
938 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
939 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
940 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
941 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
942 static inline int sb_issue_discard(struct super_block *sb,
943 sector_t block, sector_t nr_blocks)
945 block <<= (sb->s_blocksize_bits - 9);
946 nr_blocks <<= (sb->s_blocksize_bits - 9);
947 return blkdev_issue_discard(sb->s_bdev, block, nr_blocks, GFP_KERNEL,
948 BLKDEV_IFL_WAIT | BLKDEV_IFL_BARRIER);
951 extern int blk_verify_command(unsigned char *cmd, fmode_t has_write_perm);
953 enum blk_default_limits {
954 BLK_MAX_SEGMENTS = 128,
955 BLK_SAFE_MAX_SECTORS = 255,
956 BLK_DEF_MAX_SECTORS = 1024,
957 BLK_MAX_SEGMENT_SIZE = 65536,
958 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
961 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
963 static inline unsigned long queue_bounce_pfn(struct request_queue *q)
965 return q->limits.bounce_pfn;
968 static inline unsigned long queue_segment_boundary(struct request_queue *q)
970 return q->limits.seg_boundary_mask;
973 static inline unsigned int queue_max_sectors(struct request_queue *q)
975 return q->limits.max_sectors;
978 static inline unsigned int queue_max_hw_sectors(struct request_queue *q)
980 return q->limits.max_hw_sectors;
983 static inline unsigned short queue_max_segments(struct request_queue *q)
985 return q->limits.max_segments;
988 static inline unsigned int queue_max_segment_size(struct request_queue *q)
990 return q->limits.max_segment_size;
993 static inline unsigned short queue_logical_block_size(struct request_queue *q)
997 if (q && q->limits.logical_block_size)
998 retval = q->limits.logical_block_size;
1003 static inline unsigned short bdev_logical_block_size(struct block_device *bdev)
1005 return queue_logical_block_size(bdev_get_queue(bdev));
1008 static inline unsigned int queue_physical_block_size(struct request_queue *q)
1010 return q->limits.physical_block_size;
1013 static inline int bdev_physical_block_size(struct block_device *bdev)
1015 return queue_physical_block_size(bdev_get_queue(bdev));
1018 static inline unsigned int queue_io_min(struct request_queue *q)
1020 return q->limits.io_min;
1023 static inline int bdev_io_min(struct block_device *bdev)
1025 return queue_io_min(bdev_get_queue(bdev));
1028 static inline unsigned int queue_io_opt(struct request_queue *q)
1030 return q->limits.io_opt;
1033 static inline int bdev_io_opt(struct block_device *bdev)
1035 return queue_io_opt(bdev_get_queue(bdev));
1038 static inline int queue_alignment_offset(struct request_queue *q)
1040 if (q->limits.misaligned)
1043 return q->limits.alignment_offset;
1046 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1048 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1049 unsigned int alignment = (sector << 9) & (granularity - 1);
1051 return (granularity + lim->alignment_offset - alignment)
1052 & (granularity - 1);
1055 static inline int bdev_alignment_offset(struct block_device *bdev)
1057 struct request_queue *q = bdev_get_queue(bdev);
1059 if (q->limits.misaligned)
1062 if (bdev != bdev->bd_contains)
1063 return bdev->bd_part->alignment_offset;
1065 return q->limits.alignment_offset;
1068 static inline int queue_discard_alignment(struct request_queue *q)
1070 if (q->limits.discard_misaligned)
1073 return q->limits.discard_alignment;
1076 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1078 unsigned int alignment = (sector << 9) & (lim->discard_granularity - 1);
1080 return (lim->discard_granularity + lim->discard_alignment - alignment)
1081 & (lim->discard_granularity - 1);
1084 static inline unsigned int queue_discard_zeroes_data(struct request_queue *q)
1086 if (q->limits.discard_zeroes_data == 1)
1092 static inline unsigned int bdev_discard_zeroes_data(struct block_device *bdev)
1094 return queue_discard_zeroes_data(bdev_get_queue(bdev));
1097 static inline int queue_dma_alignment(struct request_queue *q)
1099 return q ? q->dma_alignment : 511;
1102 static inline int blk_rq_aligned(struct request_queue *q, void *addr,
1105 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1106 return !((unsigned long)addr & alignment) && !(len & alignment);
1109 /* assumes size > 256 */
1110 static inline unsigned int blksize_bits(unsigned int size)
1112 unsigned int bits = 8;
1116 } while (size > 256);
1120 static inline unsigned int block_size(struct block_device *bdev)
1122 return bdev->bd_block_size;
1125 typedef struct {struct page *v;} Sector;
1127 unsigned char *read_dev_sector(struct block_device *, sector_t, Sector *);
1129 static inline void put_dev_sector(Sector p)
1131 page_cache_release(p.v);
1135 int kblockd_schedule_work(struct request_queue *q, struct work_struct *work);
1137 #ifdef CONFIG_BLK_CGROUP
1139 * This should not be using sched_clock(). A real patch is in progress
1140 * to fix this up, until that is in place we need to disable preemption
1141 * around sched_clock() in this function and set_io_start_time_ns().
1143 static inline void set_start_time_ns(struct request *req)
1146 req->start_time_ns = sched_clock();
1150 static inline void set_io_start_time_ns(struct request *req)
1153 req->io_start_time_ns = sched_clock();
1157 static inline uint64_t rq_start_time_ns(struct request *req)
1159 return req->start_time_ns;
1162 static inline uint64_t rq_io_start_time_ns(struct request *req)
1164 return req->io_start_time_ns;
1167 static inline void set_start_time_ns(struct request *req) {}
1168 static inline void set_io_start_time_ns(struct request *req) {}
1169 static inline uint64_t rq_start_time_ns(struct request *req)
1173 static inline uint64_t rq_io_start_time_ns(struct request *req)
1179 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1180 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1181 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1182 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1184 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1186 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1187 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1189 struct blk_integrity_exchg {
1193 unsigned int data_size;
1194 unsigned short sector_size;
1195 const char *disk_name;
1198 typedef void (integrity_gen_fn) (struct blk_integrity_exchg *);
1199 typedef int (integrity_vrfy_fn) (struct blk_integrity_exchg *);
1200 typedef void (integrity_set_tag_fn) (void *, void *, unsigned int);
1201 typedef void (integrity_get_tag_fn) (void *, void *, unsigned int);
1203 struct blk_integrity {
1204 integrity_gen_fn *generate_fn;
1205 integrity_vrfy_fn *verify_fn;
1206 integrity_set_tag_fn *set_tag_fn;
1207 integrity_get_tag_fn *get_tag_fn;
1209 unsigned short flags;
1210 unsigned short tuple_size;
1211 unsigned short sector_size;
1212 unsigned short tag_size;
1216 struct kobject kobj;
1219 extern int blk_integrity_register(struct gendisk *, struct blk_integrity *);
1220 extern void blk_integrity_unregister(struct gendisk *);
1221 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1222 extern int blk_rq_map_integrity_sg(struct request *, struct scatterlist *);
1223 extern int blk_rq_count_integrity_sg(struct request *);
1226 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1228 return bdev->bd_disk->integrity;
1231 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1233 return disk->integrity;
1236 static inline int blk_integrity_rq(struct request *rq)
1238 if (rq->bio == NULL)
1241 return bio_integrity(rq->bio);
1244 #else /* CONFIG_BLK_DEV_INTEGRITY */
1246 #define blk_integrity_rq(rq) (0)
1247 #define blk_rq_count_integrity_sg(a) (0)
1248 #define blk_rq_map_integrity_sg(a, b) (0)
1249 #define bdev_get_integrity(a) (0)
1250 #define blk_get_integrity(a) (0)
1251 #define blk_integrity_compare(a, b) (0)
1252 #define blk_integrity_register(a, b) (0)
1253 #define blk_integrity_unregister(a) do { } while (0);
1255 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1257 struct block_device_operations {
1258 int (*open) (struct block_device *, fmode_t);
1259 int (*release) (struct gendisk *, fmode_t);
1260 int (*locked_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1261 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1262 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1263 int (*direct_access) (struct block_device *, sector_t,
1264 void **, unsigned long *);
1265 int (*media_changed) (struct gendisk *);
1266 void (*unlock_native_capacity) (struct gendisk *);
1267 int (*revalidate_disk) (struct gendisk *);
1268 int (*getgeo)(struct block_device *, struct hd_geometry *);
1269 /* this callback is with swap_lock and sometimes page table lock held */
1270 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1271 struct module *owner;
1274 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1276 #else /* CONFIG_BLOCK */
1278 * stubs for when the block layer is configured out
1280 #define buffer_heads_over_limit 0
1282 static inline long nr_blockdev_pages(void)
1287 #endif /* CONFIG_BLOCK */