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 (unprep_rq_fn) (struct request_queue *, struct request *);
204 typedef void (unplug_fn) (struct request_queue *);
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 (prepare_flush_fn) (struct request_queue *, struct request *);
216 typedef void (softirq_done_fn)(struct request *);
217 typedef int (dma_drain_needed_fn)(struct request *);
218 typedef int (lld_busy_fn) (struct request_queue *q);
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;
263 unsigned char misaligned;
264 unsigned char discard_misaligned;
265 unsigned char no_cluster;
266 signed char discard_zeroes_data;
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 unplug_fn *unplug_fn;
288 merge_bvec_fn *merge_bvec_fn;
289 prepare_flush_fn *prepare_flush_fn;
290 softirq_done_fn *softirq_done_fn;
291 rq_timed_out_fn *rq_timed_out_fn;
292 dma_drain_needed_fn *dma_drain_needed;
293 lld_busy_fn *lld_busy_fn;
296 * Dispatch queue sorting
299 struct request *boundary_rq;
302 * Auto-unplugging state
304 struct timer_list unplug_timer;
305 int unplug_thresh; /* After this many requests */
306 unsigned long unplug_delay; /* After this many jiffies */
307 struct work_struct unplug_work;
309 struct backing_dev_info backing_dev_info;
312 * The queue owner gets to use this for whatever they like.
313 * ll_rw_blk doesn't touch it.
318 * queue needs bounce pages for pages above this limit
323 * various queue flags, see QUEUE_* below
325 unsigned long queue_flags;
328 * protects queue structures from reentrancy. ->__queue_lock should
329 * _never_ be used directly, it is queue private. always use
332 spinlock_t __queue_lock;
333 spinlock_t *queue_lock;
343 unsigned long nr_requests; /* Max # of requests */
344 unsigned int nr_congestion_on;
345 unsigned int nr_congestion_off;
346 unsigned int nr_batching;
348 void *dma_drain_buffer;
349 unsigned int dma_drain_size;
350 unsigned int dma_pad_mask;
351 unsigned int dma_alignment;
353 struct blk_queue_tag *queue_tags;
354 struct list_head tag_busy_list;
356 unsigned int nr_sorted;
357 unsigned int in_flight[2];
359 unsigned int rq_timeout;
360 struct timer_list timeout;
361 struct list_head timeout_list;
363 struct queue_limits limits;
368 unsigned int sg_timeout;
369 unsigned int sg_reserved_size;
371 #ifdef CONFIG_BLK_DEV_IO_TRACE
372 struct blk_trace *blk_trace;
375 * reserved for flush operations
377 unsigned int ordered, next_ordered, ordseq;
378 int orderr, ordcolor;
379 struct request pre_flush_rq, bar_rq, post_flush_rq;
380 struct request *orig_bar_rq;
382 struct mutex sysfs_lock;
384 #if defined(CONFIG_BLK_DEV_BSG)
385 struct bsg_class_device bsg_dev;
389 #define QUEUE_FLAG_CLUSTER 0 /* cluster several segments into 1 */
390 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
391 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
392 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
393 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
394 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
395 #define QUEUE_FLAG_REENTER 6 /* Re-entrancy avoidance */
396 #define QUEUE_FLAG_PLUGGED 7 /* queue is plugged */
397 #define QUEUE_FLAG_ELVSWITCH 8 /* don't use elevator, just do FIFO */
398 #define QUEUE_FLAG_BIDI 9 /* queue supports bidi requests */
399 #define QUEUE_FLAG_NOMERGES 10 /* disable merge attempts */
400 #define QUEUE_FLAG_SAME_COMP 11 /* force complete on same CPU */
401 #define QUEUE_FLAG_FAIL_IO 12 /* fake timeout */
402 #define QUEUE_FLAG_STACKABLE 13 /* supports request stacking */
403 #define QUEUE_FLAG_NONROT 14 /* non-rotational device (SSD) */
404 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
405 #define QUEUE_FLAG_IO_STAT 15 /* do IO stats */
406 #define QUEUE_FLAG_DISCARD 16 /* supports DISCARD */
407 #define QUEUE_FLAG_NOXMERGES 17 /* No extended merges */
408 #define QUEUE_FLAG_ADD_RANDOM 18 /* Contributes to random pool */
410 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
411 (1 << QUEUE_FLAG_CLUSTER) | \
412 (1 << QUEUE_FLAG_STACKABLE) | \
413 (1 << QUEUE_FLAG_SAME_COMP) | \
414 (1 << QUEUE_FLAG_ADD_RANDOM))
416 static inline int queue_is_locked(struct request_queue *q)
419 spinlock_t *lock = q->queue_lock;
420 return lock && spin_is_locked(lock);
426 static inline void queue_flag_set_unlocked(unsigned int flag,
427 struct request_queue *q)
429 __set_bit(flag, &q->queue_flags);
432 static inline int queue_flag_test_and_clear(unsigned int flag,
433 struct request_queue *q)
435 WARN_ON_ONCE(!queue_is_locked(q));
437 if (test_bit(flag, &q->queue_flags)) {
438 __clear_bit(flag, &q->queue_flags);
445 static inline int queue_flag_test_and_set(unsigned int flag,
446 struct request_queue *q)
448 WARN_ON_ONCE(!queue_is_locked(q));
450 if (!test_bit(flag, &q->queue_flags)) {
451 __set_bit(flag, &q->queue_flags);
458 static inline void queue_flag_set(unsigned int flag, struct request_queue *q)
460 WARN_ON_ONCE(!queue_is_locked(q));
461 __set_bit(flag, &q->queue_flags);
464 static inline void queue_flag_clear_unlocked(unsigned int flag,
465 struct request_queue *q)
467 __clear_bit(flag, &q->queue_flags);
470 static inline int queue_in_flight(struct request_queue *q)
472 return q->in_flight[0] + q->in_flight[1];
475 static inline void queue_flag_clear(unsigned int flag, struct request_queue *q)
477 WARN_ON_ONCE(!queue_is_locked(q));
478 __clear_bit(flag, &q->queue_flags);
483 * Hardbarrier is supported with one of the following methods.
485 * NONE : hardbarrier unsupported
486 * DRAIN : ordering by draining is enough
487 * DRAIN_FLUSH : ordering by draining w/ pre and post flushes
488 * DRAIN_FUA : ordering by draining w/ pre flush and FUA write
489 * TAG : ordering by tag is enough
490 * TAG_FLUSH : ordering by tag w/ pre and post flushes
491 * TAG_FUA : ordering by tag w/ pre flush and FUA write
493 QUEUE_ORDERED_BY_DRAIN = 0x01,
494 QUEUE_ORDERED_BY_TAG = 0x02,
495 QUEUE_ORDERED_DO_PREFLUSH = 0x10,
496 QUEUE_ORDERED_DO_BAR = 0x20,
497 QUEUE_ORDERED_DO_POSTFLUSH = 0x40,
498 QUEUE_ORDERED_DO_FUA = 0x80,
500 QUEUE_ORDERED_NONE = 0x00,
502 QUEUE_ORDERED_DRAIN = QUEUE_ORDERED_BY_DRAIN |
503 QUEUE_ORDERED_DO_BAR,
504 QUEUE_ORDERED_DRAIN_FLUSH = QUEUE_ORDERED_DRAIN |
505 QUEUE_ORDERED_DO_PREFLUSH |
506 QUEUE_ORDERED_DO_POSTFLUSH,
507 QUEUE_ORDERED_DRAIN_FUA = QUEUE_ORDERED_DRAIN |
508 QUEUE_ORDERED_DO_PREFLUSH |
509 QUEUE_ORDERED_DO_FUA,
511 QUEUE_ORDERED_TAG = QUEUE_ORDERED_BY_TAG |
512 QUEUE_ORDERED_DO_BAR,
513 QUEUE_ORDERED_TAG_FLUSH = QUEUE_ORDERED_TAG |
514 QUEUE_ORDERED_DO_PREFLUSH |
515 QUEUE_ORDERED_DO_POSTFLUSH,
516 QUEUE_ORDERED_TAG_FUA = QUEUE_ORDERED_TAG |
517 QUEUE_ORDERED_DO_PREFLUSH |
518 QUEUE_ORDERED_DO_FUA,
521 * Ordered operation sequence
523 QUEUE_ORDSEQ_STARTED = 0x01, /* flushing in progress */
524 QUEUE_ORDSEQ_DRAIN = 0x02, /* waiting for the queue to be drained */
525 QUEUE_ORDSEQ_PREFLUSH = 0x04, /* pre-flushing in progress */
526 QUEUE_ORDSEQ_BAR = 0x08, /* original barrier req in progress */
527 QUEUE_ORDSEQ_POSTFLUSH = 0x10, /* post-flushing in progress */
528 QUEUE_ORDSEQ_DONE = 0x20,
531 #define blk_queue_plugged(q) test_bit(QUEUE_FLAG_PLUGGED, &(q)->queue_flags)
532 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
533 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
534 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
535 #define blk_queue_noxmerges(q) \
536 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
537 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
538 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
539 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
540 #define blk_queue_flushing(q) ((q)->ordseq)
541 #define blk_queue_stackable(q) \
542 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
543 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
545 #define blk_noretry_request(rq) \
546 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
547 REQ_FAILFAST_DRIVER))
549 #define blk_account_rq(rq) \
550 (((rq)->cmd_flags & REQ_STARTED) && \
551 ((rq)->cmd_type == REQ_TYPE_FS || \
552 ((rq)->cmd_flags & REQ_DISCARD)))
554 #define blk_pm_request(rq) \
555 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
556 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
558 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
559 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
560 /* rq->queuelist of dequeued request must be list_empty() */
561 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
563 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
565 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
568 * We regard a request as sync, if either a read or a sync write
570 static inline bool rw_is_sync(unsigned int rw_flags)
572 return !(rw_flags & REQ_WRITE) || (rw_flags & REQ_SYNC);
575 static inline bool rq_is_sync(struct request *rq)
577 return rw_is_sync(rq->cmd_flags);
580 static inline int blk_queue_full(struct request_queue *q, int sync)
583 return test_bit(QUEUE_FLAG_SYNCFULL, &q->queue_flags);
584 return test_bit(QUEUE_FLAG_ASYNCFULL, &q->queue_flags);
587 static inline void blk_set_queue_full(struct request_queue *q, int sync)
590 queue_flag_set(QUEUE_FLAG_SYNCFULL, q);
592 queue_flag_set(QUEUE_FLAG_ASYNCFULL, q);
595 static inline void blk_clear_queue_full(struct request_queue *q, int sync)
598 queue_flag_clear(QUEUE_FLAG_SYNCFULL, q);
600 queue_flag_clear(QUEUE_FLAG_ASYNCFULL, q);
605 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
606 * it already be started by driver.
608 #define RQ_NOMERGE_FLAGS \
609 (REQ_NOMERGE | REQ_STARTED | REQ_HARDBARRIER | REQ_SOFTBARRIER)
610 #define rq_mergeable(rq) \
611 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
612 (((rq)->cmd_flags & REQ_DISCARD) || \
613 (rq)->cmd_type == REQ_TYPE_FS))
616 * q->prep_rq_fn return values
618 #define BLKPREP_OK 0 /* serve it */
619 #define BLKPREP_KILL 1 /* fatal error, kill */
620 #define BLKPREP_DEFER 2 /* leave on queue */
622 extern unsigned long blk_max_low_pfn, blk_max_pfn;
625 * standard bounce addresses:
627 * BLK_BOUNCE_HIGH : bounce all highmem pages
628 * BLK_BOUNCE_ANY : don't bounce anything
629 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
632 #if BITS_PER_LONG == 32
633 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
635 #define BLK_BOUNCE_HIGH -1ULL
637 #define BLK_BOUNCE_ANY (-1ULL)
638 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
641 * default timeout for SG_IO if none specified
643 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
644 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
647 extern int init_emergency_isa_pool(void);
648 extern void blk_queue_bounce(struct request_queue *q, struct bio **bio);
650 static inline int init_emergency_isa_pool(void)
654 static inline void blk_queue_bounce(struct request_queue *q, struct bio **bio)
657 #endif /* CONFIG_MMU */
663 unsigned long offset;
668 struct req_iterator {
673 /* This should not be used directly - use rq_for_each_segment */
674 #define for_each_bio(_bio) \
675 for (; _bio; _bio = _bio->bi_next)
676 #define __rq_for_each_bio(_bio, rq) \
678 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
680 #define rq_for_each_segment(bvl, _rq, _iter) \
681 __rq_for_each_bio(_iter.bio, _rq) \
682 bio_for_each_segment(bvl, _iter.bio, _iter.i)
684 #define rq_iter_last(rq, _iter) \
685 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
687 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
688 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
690 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
691 extern void rq_flush_dcache_pages(struct request *rq);
693 static inline void rq_flush_dcache_pages(struct request *rq)
698 extern int blk_register_queue(struct gendisk *disk);
699 extern void blk_unregister_queue(struct gendisk *disk);
700 extern void register_disk(struct gendisk *dev);
701 extern void generic_make_request(struct bio *bio);
702 extern void blk_rq_init(struct request_queue *q, struct request *rq);
703 extern void blk_put_request(struct request *);
704 extern void __blk_put_request(struct request_queue *, struct request *);
705 extern struct request *blk_get_request(struct request_queue *, int, gfp_t);
706 extern struct request *blk_make_request(struct request_queue *, struct bio *,
708 extern void blk_insert_request(struct request_queue *, struct request *, int, void *);
709 extern void blk_requeue_request(struct request_queue *, struct request *);
710 extern void blk_add_request_payload(struct request *rq, struct page *page,
712 extern int blk_rq_check_limits(struct request_queue *q, struct request *rq);
713 extern int blk_lld_busy(struct request_queue *q);
714 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
715 struct bio_set *bs, gfp_t gfp_mask,
716 int (*bio_ctr)(struct bio *, struct bio *, void *),
718 extern void blk_rq_unprep_clone(struct request *rq);
719 extern int blk_insert_cloned_request(struct request_queue *q,
721 extern void blk_plug_device(struct request_queue *);
722 extern void blk_plug_device_unlocked(struct request_queue *);
723 extern int blk_remove_plug(struct request_queue *);
724 extern void blk_recount_segments(struct request_queue *, struct bio *);
725 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
726 unsigned int, void __user *);
727 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
728 struct scsi_ioctl_command __user *);
731 * A queue has just exitted congestion. Note this in the global counter of
732 * congested queues, and wake up anyone who was waiting for requests to be
735 static inline void blk_clear_queue_congested(struct request_queue *q, int sync)
737 clear_bdi_congested(&q->backing_dev_info, sync);
741 * A queue has just entered congestion. Flag that in the queue's VM-visible
742 * state flags and increment the global gounter of congested queues.
744 static inline void blk_set_queue_congested(struct request_queue *q, int sync)
746 set_bdi_congested(&q->backing_dev_info, sync);
749 extern void blk_start_queue(struct request_queue *q);
750 extern void blk_stop_queue(struct request_queue *q);
751 extern void blk_sync_queue(struct request_queue *q);
752 extern void __blk_stop_queue(struct request_queue *q);
753 extern void __blk_run_queue(struct request_queue *);
754 extern void blk_run_queue(struct request_queue *);
755 extern int blk_rq_map_user(struct request_queue *, struct request *,
756 struct rq_map_data *, void __user *, unsigned long,
758 extern int blk_rq_unmap_user(struct bio *);
759 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
760 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
761 struct rq_map_data *, struct sg_iovec *, int,
762 unsigned int, gfp_t);
763 extern int blk_execute_rq(struct request_queue *, struct gendisk *,
764 struct request *, int);
765 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
766 struct request *, int, rq_end_io_fn *);
767 extern void blk_unplug(struct request_queue *q);
769 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
771 return bdev->bd_disk->queue;
775 * blk_rq_pos() : the current sector
776 * blk_rq_bytes() : bytes left in the entire request
777 * blk_rq_cur_bytes() : bytes left in the current segment
778 * blk_rq_err_bytes() : bytes left till the next error boundary
779 * blk_rq_sectors() : sectors left in the entire request
780 * blk_rq_cur_sectors() : sectors left in the current segment
782 static inline sector_t blk_rq_pos(const struct request *rq)
787 static inline unsigned int blk_rq_bytes(const struct request *rq)
789 return rq->__data_len;
792 static inline int blk_rq_cur_bytes(const struct request *rq)
794 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
797 extern unsigned int blk_rq_err_bytes(const struct request *rq);
799 static inline unsigned int blk_rq_sectors(const struct request *rq)
801 return blk_rq_bytes(rq) >> 9;
804 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
806 return blk_rq_cur_bytes(rq) >> 9;
810 * Request issue related functions.
812 extern struct request *blk_peek_request(struct request_queue *q);
813 extern void blk_start_request(struct request *rq);
814 extern struct request *blk_fetch_request(struct request_queue *q);
817 * Request completion related functions.
819 * blk_update_request() completes given number of bytes and updates
820 * the request without completing it.
822 * blk_end_request() and friends. __blk_end_request() must be called
823 * with the request queue spinlock acquired.
825 * Several drivers define their own end_request and call
826 * blk_end_request() for parts of the original function.
827 * This prevents code duplication in drivers.
829 extern bool blk_update_request(struct request *rq, int error,
830 unsigned int nr_bytes);
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);
836 extern bool __blk_end_request(struct request *rq, int error,
837 unsigned int nr_bytes);
838 extern void __blk_end_request_all(struct request *rq, int error);
839 extern bool __blk_end_request_cur(struct request *rq, int error);
840 extern bool __blk_end_request_err(struct request *rq, int error);
842 extern void blk_complete_request(struct request *);
843 extern void __blk_complete_request(struct request *);
844 extern void blk_abort_request(struct request *);
845 extern void blk_abort_queue(struct request_queue *);
846 extern void blk_unprep_request(struct request *);
849 * Access functions for manipulating queue properties
851 extern struct request_queue *blk_init_queue_node(request_fn_proc *rfn,
852 spinlock_t *lock, int node_id);
853 extern struct request_queue *blk_init_allocated_queue_node(struct request_queue *,
855 spinlock_t *, int node_id);
856 extern struct request_queue *blk_init_queue(request_fn_proc *, spinlock_t *);
857 extern struct request_queue *blk_init_allocated_queue(struct request_queue *,
858 request_fn_proc *, spinlock_t *);
859 extern void blk_cleanup_queue(struct request_queue *);
860 extern void blk_queue_make_request(struct request_queue *, make_request_fn *);
861 extern void blk_queue_bounce_limit(struct request_queue *, u64);
862 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
863 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
864 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
865 extern void blk_queue_max_discard_sectors(struct request_queue *q,
866 unsigned int max_discard_sectors);
867 extern void blk_queue_logical_block_size(struct request_queue *, unsigned short);
868 extern void blk_queue_physical_block_size(struct request_queue *, unsigned short);
869 extern void blk_queue_alignment_offset(struct request_queue *q,
870 unsigned int alignment);
871 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
872 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
873 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
874 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
875 extern void blk_set_default_limits(struct queue_limits *lim);
876 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
878 extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev,
880 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
882 extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b);
883 extern void blk_queue_dma_pad(struct request_queue *, unsigned int);
884 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
885 extern int blk_queue_dma_drain(struct request_queue *q,
886 dma_drain_needed_fn *dma_drain_needed,
887 void *buf, unsigned int size);
888 extern void blk_queue_lld_busy(struct request_queue *q, lld_busy_fn *fn);
889 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
890 extern void blk_queue_prep_rq(struct request_queue *, prep_rq_fn *pfn);
891 extern void blk_queue_unprep_rq(struct request_queue *, unprep_rq_fn *ufn);
892 extern void blk_queue_merge_bvec(struct request_queue *, merge_bvec_fn *);
893 extern void blk_queue_dma_alignment(struct request_queue *, int);
894 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
895 extern void blk_queue_softirq_done(struct request_queue *, softirq_done_fn *);
896 extern void blk_queue_rq_timed_out(struct request_queue *, rq_timed_out_fn *);
897 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
898 extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
899 extern int blk_queue_ordered(struct request_queue *, unsigned, prepare_flush_fn *);
900 extern bool blk_do_ordered(struct request_queue *, struct request **);
901 extern unsigned blk_ordered_cur_seq(struct request_queue *);
902 extern unsigned blk_ordered_req_seq(struct request *);
903 extern bool blk_ordered_complete_seq(struct request_queue *, unsigned, int);
905 extern int blk_rq_map_sg(struct request_queue *, struct request *, struct scatterlist *);
906 extern void blk_dump_rq_flags(struct request *, char *);
907 extern void generic_unplug_device(struct request_queue *);
908 extern long nr_blockdev_pages(void);
910 int blk_get_queue(struct request_queue *);
911 struct request_queue *blk_alloc_queue(gfp_t);
912 struct request_queue *blk_alloc_queue_node(gfp_t, int);
913 extern void blk_put_queue(struct request_queue *);
918 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
919 extern int blk_queue_start_tag(struct request_queue *, struct request *);
920 extern struct request *blk_queue_find_tag(struct request_queue *, int);
921 extern void blk_queue_end_tag(struct request_queue *, struct request *);
922 extern int blk_queue_init_tags(struct request_queue *, int, struct blk_queue_tag *);
923 extern void blk_queue_free_tags(struct request_queue *);
924 extern int blk_queue_resize_tags(struct request_queue *, int);
925 extern void blk_queue_invalidate_tags(struct request_queue *);
926 extern struct blk_queue_tag *blk_init_tags(int);
927 extern void blk_free_tags(struct blk_queue_tag *);
929 static inline struct request *blk_map_queue_find_tag(struct blk_queue_tag *bqt,
932 if (unlikely(bqt == NULL || tag >= bqt->real_max_depth))
934 return bqt->tag_index[tag];
937 BLKDEV_WAIT, /* wait for completion */
938 BLKDEV_BARRIER, /*issue request with barrier */
940 #define BLKDEV_IFL_WAIT (1 << BLKDEV_WAIT)
941 #define BLKDEV_IFL_BARRIER (1 << BLKDEV_BARRIER)
942 extern int blkdev_issue_flush(struct block_device *, gfp_t, sector_t *,
944 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
945 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
946 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
947 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
948 static inline int sb_issue_discard(struct super_block *sb,
949 sector_t block, sector_t nr_blocks)
951 block <<= (sb->s_blocksize_bits - 9);
952 nr_blocks <<= (sb->s_blocksize_bits - 9);
953 return blkdev_issue_discard(sb->s_bdev, block, nr_blocks, GFP_KERNEL,
954 BLKDEV_IFL_WAIT | BLKDEV_IFL_BARRIER);
957 extern int blk_verify_command(unsigned char *cmd, fmode_t has_write_perm);
959 enum blk_default_limits {
960 BLK_MAX_SEGMENTS = 128,
961 BLK_SAFE_MAX_SECTORS = 255,
962 BLK_DEF_MAX_SECTORS = 1024,
963 BLK_MAX_SEGMENT_SIZE = 65536,
964 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
967 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
969 static inline unsigned long queue_bounce_pfn(struct request_queue *q)
971 return q->limits.bounce_pfn;
974 static inline unsigned long queue_segment_boundary(struct request_queue *q)
976 return q->limits.seg_boundary_mask;
979 static inline unsigned int queue_max_sectors(struct request_queue *q)
981 return q->limits.max_sectors;
984 static inline unsigned int queue_max_hw_sectors(struct request_queue *q)
986 return q->limits.max_hw_sectors;
989 static inline unsigned short queue_max_segments(struct request_queue *q)
991 return q->limits.max_segments;
994 static inline unsigned int queue_max_segment_size(struct request_queue *q)
996 return q->limits.max_segment_size;
999 static inline unsigned short queue_logical_block_size(struct request_queue *q)
1003 if (q && q->limits.logical_block_size)
1004 retval = q->limits.logical_block_size;
1009 static inline unsigned short bdev_logical_block_size(struct block_device *bdev)
1011 return queue_logical_block_size(bdev_get_queue(bdev));
1014 static inline unsigned int queue_physical_block_size(struct request_queue *q)
1016 return q->limits.physical_block_size;
1019 static inline int bdev_physical_block_size(struct block_device *bdev)
1021 return queue_physical_block_size(bdev_get_queue(bdev));
1024 static inline unsigned int queue_io_min(struct request_queue *q)
1026 return q->limits.io_min;
1029 static inline int bdev_io_min(struct block_device *bdev)
1031 return queue_io_min(bdev_get_queue(bdev));
1034 static inline unsigned int queue_io_opt(struct request_queue *q)
1036 return q->limits.io_opt;
1039 static inline int bdev_io_opt(struct block_device *bdev)
1041 return queue_io_opt(bdev_get_queue(bdev));
1044 static inline int queue_alignment_offset(struct request_queue *q)
1046 if (q->limits.misaligned)
1049 return q->limits.alignment_offset;
1052 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1054 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1055 unsigned int alignment = (sector << 9) & (granularity - 1);
1057 return (granularity + lim->alignment_offset - alignment)
1058 & (granularity - 1);
1061 static inline int bdev_alignment_offset(struct block_device *bdev)
1063 struct request_queue *q = bdev_get_queue(bdev);
1065 if (q->limits.misaligned)
1068 if (bdev != bdev->bd_contains)
1069 return bdev->bd_part->alignment_offset;
1071 return q->limits.alignment_offset;
1074 static inline int queue_discard_alignment(struct request_queue *q)
1076 if (q->limits.discard_misaligned)
1079 return q->limits.discard_alignment;
1082 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1084 unsigned int alignment = (sector << 9) & (lim->discard_granularity - 1);
1086 return (lim->discard_granularity + lim->discard_alignment - alignment)
1087 & (lim->discard_granularity - 1);
1090 static inline unsigned int queue_discard_zeroes_data(struct request_queue *q)
1092 if (q->limits.discard_zeroes_data == 1)
1098 static inline unsigned int bdev_discard_zeroes_data(struct block_device *bdev)
1100 return queue_discard_zeroes_data(bdev_get_queue(bdev));
1103 static inline int queue_dma_alignment(struct request_queue *q)
1105 return q ? q->dma_alignment : 511;
1108 static inline int blk_rq_aligned(struct request_queue *q, void *addr,
1111 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1112 return !((unsigned long)addr & alignment) && !(len & alignment);
1115 /* assumes size > 256 */
1116 static inline unsigned int blksize_bits(unsigned int size)
1118 unsigned int bits = 8;
1122 } while (size > 256);
1126 static inline unsigned int block_size(struct block_device *bdev)
1128 return bdev->bd_block_size;
1131 typedef struct {struct page *v;} Sector;
1133 unsigned char *read_dev_sector(struct block_device *, sector_t, Sector *);
1135 static inline void put_dev_sector(Sector p)
1137 page_cache_release(p.v);
1141 int kblockd_schedule_work(struct request_queue *q, struct work_struct *work);
1143 #ifdef CONFIG_BLK_CGROUP
1145 * This should not be using sched_clock(). A real patch is in progress
1146 * to fix this up, until that is in place we need to disable preemption
1147 * around sched_clock() in this function and set_io_start_time_ns().
1149 static inline void set_start_time_ns(struct request *req)
1152 req->start_time_ns = sched_clock();
1156 static inline void set_io_start_time_ns(struct request *req)
1159 req->io_start_time_ns = sched_clock();
1163 static inline uint64_t rq_start_time_ns(struct request *req)
1165 return req->start_time_ns;
1168 static inline uint64_t rq_io_start_time_ns(struct request *req)
1170 return req->io_start_time_ns;
1173 static inline void set_start_time_ns(struct request *req) {}
1174 static inline void set_io_start_time_ns(struct request *req) {}
1175 static inline uint64_t rq_start_time_ns(struct request *req)
1179 static inline uint64_t rq_io_start_time_ns(struct request *req)
1185 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1186 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1187 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1188 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1190 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1192 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1193 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1195 struct blk_integrity_exchg {
1199 unsigned int data_size;
1200 unsigned short sector_size;
1201 const char *disk_name;
1204 typedef void (integrity_gen_fn) (struct blk_integrity_exchg *);
1205 typedef int (integrity_vrfy_fn) (struct blk_integrity_exchg *);
1206 typedef void (integrity_set_tag_fn) (void *, void *, unsigned int);
1207 typedef void (integrity_get_tag_fn) (void *, void *, unsigned int);
1209 struct blk_integrity {
1210 integrity_gen_fn *generate_fn;
1211 integrity_vrfy_fn *verify_fn;
1212 integrity_set_tag_fn *set_tag_fn;
1213 integrity_get_tag_fn *get_tag_fn;
1215 unsigned short flags;
1216 unsigned short tuple_size;
1217 unsigned short sector_size;
1218 unsigned short tag_size;
1222 struct kobject kobj;
1225 extern int blk_integrity_register(struct gendisk *, struct blk_integrity *);
1226 extern void blk_integrity_unregister(struct gendisk *);
1227 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1228 extern int blk_rq_map_integrity_sg(struct request *, struct scatterlist *);
1229 extern int blk_rq_count_integrity_sg(struct request *);
1232 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1234 return bdev->bd_disk->integrity;
1237 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1239 return disk->integrity;
1242 static inline int blk_integrity_rq(struct request *rq)
1244 if (rq->bio == NULL)
1247 return bio_integrity(rq->bio);
1250 #else /* CONFIG_BLK_DEV_INTEGRITY */
1252 #define blk_integrity_rq(rq) (0)
1253 #define blk_rq_count_integrity_sg(a) (0)
1254 #define blk_rq_map_integrity_sg(a, b) (0)
1255 #define bdev_get_integrity(a) (0)
1256 #define blk_get_integrity(a) (0)
1257 #define blk_integrity_compare(a, b) (0)
1258 #define blk_integrity_register(a, b) (0)
1259 #define blk_integrity_unregister(a) do { } while (0);
1261 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1263 struct block_device_operations {
1264 int (*open) (struct block_device *, fmode_t);
1265 int (*release) (struct gendisk *, fmode_t);
1266 int (*locked_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1267 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1268 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1269 int (*direct_access) (struct block_device *, sector_t,
1270 void **, unsigned long *);
1271 int (*media_changed) (struct gendisk *);
1272 void (*unlock_native_capacity) (struct gendisk *);
1273 int (*revalidate_disk) (struct gendisk *);
1274 int (*getgeo)(struct block_device *, struct hd_geometry *);
1275 /* this callback is with swap_lock and sometimes page table lock held */
1276 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1277 struct module *owner;
1280 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1282 #else /* CONFIG_BLOCK */
1284 * stubs for when the block layer is configured out
1286 #define buffer_heads_over_limit 0
1288 static inline long nr_blockdev_pages(void)
1293 #endif /* CONFIG_BLOCK */