1 #ifndef _LINUX_BLKDEV_H
2 #define _LINUX_BLKDEV_H
4 #include <linux/sched.h>
8 #include <linux/major.h>
9 #include <linux/genhd.h>
10 #include <linux/list.h>
11 #include <linux/timer.h>
12 #include <linux/workqueue.h>
13 #include <linux/pagemap.h>
14 #include <linux/backing-dev.h>
15 #include <linux/wait.h>
16 #include <linux/mempool.h>
17 #include <linux/bio.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>
26 struct scsi_ioctl_command;
29 struct elevator_queue;
30 struct request_pm_state;
37 #define BLKDEV_MIN_RQ 4
38 #define BLKDEV_MAX_RQ 128 /* Default maximum */
41 * Maximum number of blkcg policies allowed to be registered concurrently.
42 * Defined here to simplify include dependency.
44 #define BLKCG_MAX_POLS 2
47 typedef void (rq_end_io_fn)(struct request *, int);
49 #define BLK_RL_SYNCFULL (1U << 0)
50 #define BLK_RL_ASYNCFULL (1U << 1)
53 struct request_queue *q; /* the queue this rl belongs to */
54 #ifdef CONFIG_BLK_CGROUP
55 struct blkcg_gq *blkg; /* blkg this request pool belongs to */
58 * count[], starved[], and wait[] are indexed by
59 * BLK_RW_SYNC/BLK_RW_ASYNC
64 wait_queue_head_t wait[2];
69 * request command types
71 enum rq_cmd_type_bits {
72 REQ_TYPE_FS = 1, /* fs request */
73 REQ_TYPE_BLOCK_PC, /* scsi command */
74 REQ_TYPE_SENSE, /* sense request */
75 REQ_TYPE_PM_SUSPEND, /* suspend request */
76 REQ_TYPE_PM_RESUME, /* resume request */
77 REQ_TYPE_PM_SHUTDOWN, /* shutdown request */
78 REQ_TYPE_SPECIAL, /* driver defined type */
80 * for ATA/ATAPI devices. this really doesn't belong here, ide should
81 * use REQ_TYPE_SPECIAL and use rq->cmd[0] with the range of driver
82 * private REQ_LB opcodes to differentiate what type of request this is
84 REQ_TYPE_ATA_TASKFILE,
88 #define BLK_MAX_CDB 16
91 * try to put the fields that are referenced together in the same cacheline.
92 * if you modify this structure, be sure to check block/blk-core.c:blk_rq_init()
96 struct list_head queuelist;
97 struct call_single_data csd;
99 struct request_queue *q;
101 unsigned int cmd_flags;
102 enum rq_cmd_type_bits cmd_type;
103 unsigned long atomic_flags;
107 /* the following two fields are internal, NEVER access directly */
108 unsigned int __data_len; /* total data len */
109 sector_t __sector; /* sector cursor */
114 struct hlist_node hash; /* merge hash */
116 * The rb_node is only used inside the io scheduler, requests
117 * are pruned when moved to the dispatch queue. So let the
118 * completion_data share space with the rb_node.
121 struct rb_node rb_node; /* sort/lookup */
122 void *completion_data;
126 * Three pointers are available for the IO schedulers, if they need
127 * more they have to dynamically allocate it. Flush requests are
128 * never put on the IO scheduler. So let the flush fields share
129 * space with the elevator data.
139 struct list_head list;
140 rq_end_io_fn *saved_end_io;
144 struct gendisk *rq_disk;
145 struct hd_struct *part;
146 unsigned long start_time;
147 #ifdef CONFIG_BLK_CGROUP
148 struct request_list *rl; /* rl this rq is alloced from */
149 unsigned long long start_time_ns;
150 unsigned long long io_start_time_ns; /* when passed to hardware */
152 /* Number of scatter-gather DMA addr+len pairs after
153 * physical address coalescing is performed.
155 unsigned short nr_phys_segments;
156 #if defined(CONFIG_BLK_DEV_INTEGRITY)
157 unsigned short nr_integrity_segments;
160 unsigned short ioprio;
164 void *special; /* opaque pointer available for LLD use */
165 char *buffer; /* kaddr of the current segment if available */
171 * when request is used as a packet command carrier
173 unsigned char __cmd[BLK_MAX_CDB];
175 unsigned short cmd_len;
177 unsigned int extra_len; /* length of alignment and padding */
178 unsigned int sense_len;
179 unsigned int resid_len; /* residual count */
182 unsigned long deadline;
183 struct list_head timeout_list;
184 unsigned int timeout;
188 * completion callback.
190 rq_end_io_fn *end_io;
194 struct request *next_rq;
197 static inline unsigned short req_get_ioprio(struct request *req)
203 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
204 * requests. Some step values could eventually be made generic.
206 struct request_pm_state
208 /* PM state machine step value, currently driver specific */
210 /* requested PM state value (S1, S2, S3, S4, ...) */
212 void* data; /* for driver use */
215 #include <linux/elevator.h>
217 typedef void (request_fn_proc) (struct request_queue *q);
218 typedef void (make_request_fn) (struct request_queue *q, struct bio *bio);
219 typedef int (prep_rq_fn) (struct request_queue *, struct request *);
220 typedef void (unprep_rq_fn) (struct request_queue *, struct request *);
223 struct bvec_merge_data {
224 struct block_device *bi_bdev;
229 typedef int (merge_bvec_fn) (struct request_queue *, struct bvec_merge_data *,
231 typedef void (softirq_done_fn)(struct request *);
232 typedef int (dma_drain_needed_fn)(struct request *);
233 typedef int (lld_busy_fn) (struct request_queue *q);
234 typedef int (bsg_job_fn) (struct bsg_job *);
236 enum blk_eh_timer_return {
242 typedef enum blk_eh_timer_return (rq_timed_out_fn)(struct request *);
244 enum blk_queue_state {
249 struct blk_queue_tag {
250 struct request **tag_index; /* map of busy tags */
251 unsigned long *tag_map; /* bit map of free/busy tags */
252 int busy; /* current depth */
253 int max_depth; /* what we will send to device */
254 int real_max_depth; /* what the array can hold */
255 atomic_t refcnt; /* map can be shared */
258 #define BLK_SCSI_MAX_CMDS (256)
259 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
261 struct queue_limits {
262 unsigned long bounce_pfn;
263 unsigned long seg_boundary_mask;
265 unsigned int max_hw_sectors;
266 unsigned int max_sectors;
267 unsigned int max_segment_size;
268 unsigned int physical_block_size;
269 unsigned int alignment_offset;
272 unsigned int max_discard_sectors;
273 unsigned int max_write_same_sectors;
274 unsigned int discard_granularity;
275 unsigned int discard_alignment;
277 unsigned short logical_block_size;
278 unsigned short max_segments;
279 unsigned short max_integrity_segments;
281 unsigned char misaligned;
282 unsigned char discard_misaligned;
283 unsigned char cluster;
284 unsigned char discard_zeroes_data;
287 struct request_queue {
289 * Together with queue_head for cacheline sharing
291 struct list_head queue_head;
292 struct request *last_merge;
293 struct elevator_queue *elevator;
294 int nr_rqs[2]; /* # allocated [a]sync rqs */
295 int nr_rqs_elvpriv; /* # allocated rqs w/ elvpriv */
298 * If blkcg is not used, @q->root_rl serves all requests. If blkcg
299 * is used, root blkg allocates from @q->root_rl and all other
300 * blkgs from their own blkg->rl. Which one to use should be
301 * determined using bio_request_list().
303 struct request_list root_rl;
305 request_fn_proc *request_fn;
306 make_request_fn *make_request_fn;
307 prep_rq_fn *prep_rq_fn;
308 unprep_rq_fn *unprep_rq_fn;
309 merge_bvec_fn *merge_bvec_fn;
310 softirq_done_fn *softirq_done_fn;
311 rq_timed_out_fn *rq_timed_out_fn;
312 dma_drain_needed_fn *dma_drain_needed;
313 lld_busy_fn *lld_busy_fn;
316 * Dispatch queue sorting
319 struct request *boundary_rq;
322 * Delayed queue handling
324 struct delayed_work delay_work;
326 struct backing_dev_info backing_dev_info;
329 * The queue owner gets to use this for whatever they like.
330 * ll_rw_blk doesn't touch it.
335 * various queue flags, see QUEUE_* below
337 unsigned long queue_flags;
340 * ida allocated id for this queue. Used to index queues from
346 * queue needs bounce pages for pages above this limit
351 * protects queue structures from reentrancy. ->__queue_lock should
352 * _never_ be used directly, it is queue private. always use
355 spinlock_t __queue_lock;
356 spinlock_t *queue_lock;
366 unsigned long nr_requests; /* Max # of requests */
367 unsigned int nr_congestion_on;
368 unsigned int nr_congestion_off;
369 unsigned int nr_batching;
371 unsigned int dma_drain_size;
372 void *dma_drain_buffer;
373 unsigned int dma_pad_mask;
374 unsigned int dma_alignment;
376 struct blk_queue_tag *queue_tags;
377 struct list_head tag_busy_list;
379 unsigned int nr_sorted;
380 unsigned int in_flight[2];
382 unsigned int rq_timeout;
383 struct timer_list timeout;
384 struct list_head timeout_list;
386 struct list_head icq_list;
387 #ifdef CONFIG_BLK_CGROUP
388 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
389 struct blkcg_gq *root_blkg;
390 struct list_head blkg_list;
393 struct queue_limits limits;
398 unsigned int sg_timeout;
399 unsigned int sg_reserved_size;
401 #ifdef CONFIG_BLK_DEV_IO_TRACE
402 struct blk_trace *blk_trace;
405 * for flush operations
407 unsigned int flush_flags;
408 unsigned int flush_not_queueable:1;
409 unsigned int flush_queue_delayed:1;
410 unsigned int flush_pending_idx:1;
411 unsigned int flush_running_idx:1;
412 unsigned long flush_pending_since;
413 struct list_head flush_queue[2];
414 struct list_head flush_data_in_flight;
415 struct request flush_rq;
417 struct mutex sysfs_lock;
421 #if defined(CONFIG_BLK_DEV_BSG)
422 bsg_job_fn *bsg_job_fn;
424 struct bsg_class_device bsg_dev;
427 #ifdef CONFIG_BLK_CGROUP
428 struct list_head all_q_node;
430 #ifdef CONFIG_BLK_DEV_THROTTLING
432 struct throtl_data *td;
436 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
437 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
438 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
439 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
440 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
441 #define QUEUE_FLAG_BYPASS 6 /* act as dumb FIFO queue */
442 #define QUEUE_FLAG_BIDI 7 /* queue supports bidi requests */
443 #define QUEUE_FLAG_NOMERGES 8 /* disable merge attempts */
444 #define QUEUE_FLAG_SAME_COMP 9 /* complete on same CPU-group */
445 #define QUEUE_FLAG_FAIL_IO 10 /* fake timeout */
446 #define QUEUE_FLAG_STACKABLE 11 /* supports request stacking */
447 #define QUEUE_FLAG_NONROT 12 /* non-rotational device (SSD) */
448 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
449 #define QUEUE_FLAG_IO_STAT 13 /* do IO stats */
450 #define QUEUE_FLAG_DISCARD 14 /* supports DISCARD */
451 #define QUEUE_FLAG_NOXMERGES 15 /* No extended merges */
452 #define QUEUE_FLAG_ADD_RANDOM 16 /* Contributes to random pool */
453 #define QUEUE_FLAG_SECDISCARD 17 /* supports SECDISCARD */
454 #define QUEUE_FLAG_SAME_FORCE 18 /* force complete on same CPU */
456 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
457 (1 << QUEUE_FLAG_STACKABLE) | \
458 (1 << QUEUE_FLAG_SAME_COMP) | \
459 (1 << QUEUE_FLAG_ADD_RANDOM))
461 static inline void queue_lockdep_assert_held(struct request_queue *q)
464 lockdep_assert_held(q->queue_lock);
467 static inline void queue_flag_set_unlocked(unsigned int flag,
468 struct request_queue *q)
470 __set_bit(flag, &q->queue_flags);
473 static inline int queue_flag_test_and_clear(unsigned int flag,
474 struct request_queue *q)
476 queue_lockdep_assert_held(q);
478 if (test_bit(flag, &q->queue_flags)) {
479 __clear_bit(flag, &q->queue_flags);
486 static inline int queue_flag_test_and_set(unsigned int flag,
487 struct request_queue *q)
489 queue_lockdep_assert_held(q);
491 if (!test_bit(flag, &q->queue_flags)) {
492 __set_bit(flag, &q->queue_flags);
499 static inline void queue_flag_set(unsigned int flag, struct request_queue *q)
501 queue_lockdep_assert_held(q);
502 __set_bit(flag, &q->queue_flags);
505 static inline void queue_flag_clear_unlocked(unsigned int flag,
506 struct request_queue *q)
508 __clear_bit(flag, &q->queue_flags);
511 static inline int queue_in_flight(struct request_queue *q)
513 return q->in_flight[0] + q->in_flight[1];
516 static inline void queue_flag_clear(unsigned int flag, struct request_queue *q)
518 queue_lockdep_assert_held(q);
519 __clear_bit(flag, &q->queue_flags);
522 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
523 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
524 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
525 #define blk_queue_bypass(q) test_bit(QUEUE_FLAG_BYPASS, &(q)->queue_flags)
526 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
527 #define blk_queue_noxmerges(q) \
528 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
529 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
530 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
531 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
532 #define blk_queue_stackable(q) \
533 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
534 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
535 #define blk_queue_secdiscard(q) (blk_queue_discard(q) && \
536 test_bit(QUEUE_FLAG_SECDISCARD, &(q)->queue_flags))
538 #define blk_noretry_request(rq) \
539 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
540 REQ_FAILFAST_DRIVER))
542 #define blk_account_rq(rq) \
543 (((rq)->cmd_flags & REQ_STARTED) && \
544 ((rq)->cmd_type == REQ_TYPE_FS))
546 #define blk_pm_request(rq) \
547 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
548 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
550 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
551 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
552 /* rq->queuelist of dequeued request must be list_empty() */
553 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
555 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
557 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
559 static inline unsigned int blk_queue_cluster(struct request_queue *q)
561 return q->limits.cluster;
565 * We regard a request as sync, if either a read or a sync write
567 static inline bool rw_is_sync(unsigned int rw_flags)
569 return !(rw_flags & REQ_WRITE) || (rw_flags & REQ_SYNC);
572 static inline bool rq_is_sync(struct request *rq)
574 return rw_is_sync(rq->cmd_flags);
577 static inline bool blk_rl_full(struct request_list *rl, bool sync)
579 unsigned int flag = sync ? BLK_RL_SYNCFULL : BLK_RL_ASYNCFULL;
581 return rl->flags & flag;
584 static inline void blk_set_rl_full(struct request_list *rl, bool sync)
586 unsigned int flag = sync ? BLK_RL_SYNCFULL : BLK_RL_ASYNCFULL;
591 static inline void blk_clear_rl_full(struct request_list *rl, bool sync)
593 unsigned int flag = sync ? BLK_RL_SYNCFULL : BLK_RL_ASYNCFULL;
598 static inline bool rq_mergeable(struct request *rq)
600 if (rq->cmd_type != REQ_TYPE_FS)
603 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
609 static inline bool blk_check_merge_flags(unsigned int flags1,
612 if ((flags1 & REQ_DISCARD) != (flags2 & REQ_DISCARD))
615 if ((flags1 & REQ_SECURE) != (flags2 & REQ_SECURE))
618 if ((flags1 & REQ_WRITE_SAME) != (flags2 & REQ_WRITE_SAME))
624 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
626 if (bio_data(a) == bio_data(b))
633 * q->prep_rq_fn return values
635 #define BLKPREP_OK 0 /* serve it */
636 #define BLKPREP_KILL 1 /* fatal error, kill */
637 #define BLKPREP_DEFER 2 /* leave on queue */
639 extern unsigned long blk_max_low_pfn, blk_max_pfn;
642 * standard bounce addresses:
644 * BLK_BOUNCE_HIGH : bounce all highmem pages
645 * BLK_BOUNCE_ANY : don't bounce anything
646 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
649 #if BITS_PER_LONG == 32
650 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
652 #define BLK_BOUNCE_HIGH -1ULL
654 #define BLK_BOUNCE_ANY (-1ULL)
655 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
658 * default timeout for SG_IO if none specified
660 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
661 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
664 extern int init_emergency_isa_pool(void);
665 extern void blk_queue_bounce(struct request_queue *q, struct bio **bio);
667 static inline int init_emergency_isa_pool(void)
671 static inline void blk_queue_bounce(struct request_queue *q, struct bio **bio)
674 #endif /* CONFIG_MMU */
680 unsigned long offset;
685 struct req_iterator {
690 /* This should not be used directly - use rq_for_each_segment */
691 #define for_each_bio(_bio) \
692 for (; _bio; _bio = _bio->bi_next)
693 #define __rq_for_each_bio(_bio, rq) \
695 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
697 #define rq_for_each_segment(bvl, _rq, _iter) \
698 __rq_for_each_bio(_iter.bio, _rq) \
699 bio_for_each_segment(bvl, _iter.bio, _iter.i)
701 #define rq_iter_last(rq, _iter) \
702 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
704 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
705 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
707 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
708 extern void rq_flush_dcache_pages(struct request *rq);
710 static inline void rq_flush_dcache_pages(struct request *rq)
715 extern int blk_register_queue(struct gendisk *disk);
716 extern void blk_unregister_queue(struct gendisk *disk);
717 extern void generic_make_request(struct bio *bio);
718 extern void blk_rq_init(struct request_queue *q, struct request *rq);
719 extern void blk_put_request(struct request *);
720 extern void __blk_put_request(struct request_queue *, struct request *);
721 extern struct request *blk_get_request(struct request_queue *, int, gfp_t);
722 extern struct request *blk_make_request(struct request_queue *, struct bio *,
724 extern void blk_requeue_request(struct request_queue *, struct request *);
725 extern void blk_add_request_payload(struct request *rq, struct page *page,
727 extern int blk_rq_check_limits(struct request_queue *q, struct request *rq);
728 extern int blk_lld_busy(struct request_queue *q);
729 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
730 struct bio_set *bs, gfp_t gfp_mask,
731 int (*bio_ctr)(struct bio *, struct bio *, void *),
733 extern void blk_rq_unprep_clone(struct request *rq);
734 extern int blk_insert_cloned_request(struct request_queue *q,
736 extern void blk_delay_queue(struct request_queue *, unsigned long);
737 extern void blk_recount_segments(struct request_queue *, struct bio *);
738 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
739 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
740 unsigned int, void __user *);
741 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
742 unsigned int, void __user *);
743 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
744 struct scsi_ioctl_command __user *);
746 extern void blk_queue_bio(struct request_queue *q, struct bio *bio);
749 * A queue has just exitted congestion. Note this in the global counter of
750 * congested queues, and wake up anyone who was waiting for requests to be
753 static inline void blk_clear_queue_congested(struct request_queue *q, int sync)
755 clear_bdi_congested(&q->backing_dev_info, sync);
759 * A queue has just entered congestion. Flag that in the queue's VM-visible
760 * state flags and increment the global gounter of congested queues.
762 static inline void blk_set_queue_congested(struct request_queue *q, int sync)
764 set_bdi_congested(&q->backing_dev_info, sync);
767 extern void blk_start_queue(struct request_queue *q);
768 extern void blk_stop_queue(struct request_queue *q);
769 extern void blk_sync_queue(struct request_queue *q);
770 extern void __blk_stop_queue(struct request_queue *q);
771 extern void __blk_run_queue(struct request_queue *q);
772 extern void blk_run_queue(struct request_queue *);
773 extern void blk_run_queue_async(struct request_queue *q);
774 extern int blk_rq_map_user(struct request_queue *, struct request *,
775 struct rq_map_data *, void __user *, unsigned long,
777 extern int blk_rq_unmap_user(struct bio *);
778 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
779 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
780 struct rq_map_data *, struct sg_iovec *, int,
781 unsigned int, gfp_t);
782 extern int blk_execute_rq(struct request_queue *, struct gendisk *,
783 struct request *, int);
784 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
785 struct request *, int, rq_end_io_fn *);
787 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
789 return bdev->bd_disk->queue;
793 * blk_rq_pos() : the current sector
794 * blk_rq_bytes() : bytes left in the entire request
795 * blk_rq_cur_bytes() : bytes left in the current segment
796 * blk_rq_err_bytes() : bytes left till the next error boundary
797 * blk_rq_sectors() : sectors left in the entire request
798 * blk_rq_cur_sectors() : sectors left in the current segment
800 static inline sector_t blk_rq_pos(const struct request *rq)
805 static inline unsigned int blk_rq_bytes(const struct request *rq)
807 return rq->__data_len;
810 static inline int blk_rq_cur_bytes(const struct request *rq)
812 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
815 extern unsigned int blk_rq_err_bytes(const struct request *rq);
817 static inline unsigned int blk_rq_sectors(const struct request *rq)
819 return blk_rq_bytes(rq) >> 9;
822 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
824 return blk_rq_cur_bytes(rq) >> 9;
827 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
828 unsigned int cmd_flags)
830 if (unlikely(cmd_flags & REQ_DISCARD))
831 return q->limits.max_discard_sectors;
833 if (unlikely(cmd_flags & REQ_WRITE_SAME))
834 return q->limits.max_write_same_sectors;
836 return q->limits.max_sectors;
839 static inline unsigned int blk_rq_get_max_sectors(struct request *rq)
841 struct request_queue *q = rq->q;
843 if (unlikely(rq->cmd_type == REQ_TYPE_BLOCK_PC))
844 return q->limits.max_hw_sectors;
846 return blk_queue_get_max_sectors(q, rq->cmd_flags);
850 * Request issue related functions.
852 extern struct request *blk_peek_request(struct request_queue *q);
853 extern void blk_start_request(struct request *rq);
854 extern struct request *blk_fetch_request(struct request_queue *q);
857 * Request completion related functions.
859 * blk_update_request() completes given number of bytes and updates
860 * the request without completing it.
862 * blk_end_request() and friends. __blk_end_request() must be called
863 * with the request queue spinlock acquired.
865 * Several drivers define their own end_request and call
866 * blk_end_request() for parts of the original function.
867 * This prevents code duplication in drivers.
869 extern bool blk_update_request(struct request *rq, int error,
870 unsigned int nr_bytes);
871 extern bool blk_end_request(struct request *rq, int error,
872 unsigned int nr_bytes);
873 extern void blk_end_request_all(struct request *rq, int error);
874 extern bool blk_end_request_cur(struct request *rq, int error);
875 extern bool blk_end_request_err(struct request *rq, int error);
876 extern bool __blk_end_request(struct request *rq, int error,
877 unsigned int nr_bytes);
878 extern void __blk_end_request_all(struct request *rq, int error);
879 extern bool __blk_end_request_cur(struct request *rq, int error);
880 extern bool __blk_end_request_err(struct request *rq, int error);
882 extern void blk_complete_request(struct request *);
883 extern void __blk_complete_request(struct request *);
884 extern void blk_abort_request(struct request *);
885 extern void blk_unprep_request(struct request *);
888 * Access functions for manipulating queue properties
890 extern struct request_queue *blk_init_queue_node(request_fn_proc *rfn,
891 spinlock_t *lock, int node_id);
892 extern struct request_queue *blk_init_queue(request_fn_proc *, spinlock_t *);
893 extern struct request_queue *blk_init_allocated_queue(struct request_queue *,
894 request_fn_proc *, spinlock_t *);
895 extern void blk_cleanup_queue(struct request_queue *);
896 extern void blk_queue_make_request(struct request_queue *, make_request_fn *);
897 extern void blk_queue_bounce_limit(struct request_queue *, u64);
898 extern void blk_limits_max_hw_sectors(struct queue_limits *, unsigned int);
899 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
900 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
901 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
902 extern void blk_queue_max_discard_sectors(struct request_queue *q,
903 unsigned int max_discard_sectors);
904 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
905 unsigned int max_write_same_sectors);
906 extern void blk_queue_logical_block_size(struct request_queue *, unsigned short);
907 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
908 extern void blk_queue_alignment_offset(struct request_queue *q,
909 unsigned int alignment);
910 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
911 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
912 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
913 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
914 extern void blk_set_default_limits(struct queue_limits *lim);
915 extern void blk_set_stacking_limits(struct queue_limits *lim);
916 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
918 extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev,
920 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
922 extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b);
923 extern void blk_queue_dma_pad(struct request_queue *, unsigned int);
924 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
925 extern int blk_queue_dma_drain(struct request_queue *q,
926 dma_drain_needed_fn *dma_drain_needed,
927 void *buf, unsigned int size);
928 extern void blk_queue_lld_busy(struct request_queue *q, lld_busy_fn *fn);
929 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
930 extern void blk_queue_prep_rq(struct request_queue *, prep_rq_fn *pfn);
931 extern void blk_queue_unprep_rq(struct request_queue *, unprep_rq_fn *ufn);
932 extern void blk_queue_merge_bvec(struct request_queue *, merge_bvec_fn *);
933 extern void blk_queue_dma_alignment(struct request_queue *, int);
934 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
935 extern void blk_queue_softirq_done(struct request_queue *, softirq_done_fn *);
936 extern void blk_queue_rq_timed_out(struct request_queue *, rq_timed_out_fn *);
937 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
938 extern void blk_queue_flush(struct request_queue *q, unsigned int flush);
939 extern void blk_queue_flush_queueable(struct request_queue *q, bool queueable);
940 extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
942 extern int blk_rq_map_sg(struct request_queue *, struct request *, struct scatterlist *);
943 extern int blk_bio_map_sg(struct request_queue *q, struct bio *bio,
944 struct scatterlist *sglist);
945 extern void blk_dump_rq_flags(struct request *, char *);
946 extern long nr_blockdev_pages(void);
948 bool __must_check blk_get_queue(struct request_queue *);
949 struct request_queue *blk_alloc_queue(gfp_t);
950 struct request_queue *blk_alloc_queue_node(gfp_t, int);
951 extern void blk_put_queue(struct request_queue *);
954 * blk_plug permits building a queue of related requests by holding the I/O
955 * fragments for a short period. This allows merging of sequential requests
956 * into single larger request. As the requests are moved from a per-task list to
957 * the device's request_queue in a batch, this results in improved scalability
958 * as the lock contention for request_queue lock is reduced.
960 * It is ok not to disable preemption when adding the request to the plug list
961 * or when attempting a merge, because blk_schedule_flush_list() will only flush
962 * the plug list when the task sleeps by itself. For details, please see
963 * schedule() where blk_schedule_flush_plug() is called.
966 unsigned long magic; /* detect uninitialized use-cases */
967 struct list_head list; /* requests */
968 struct list_head cb_list; /* md requires an unplug callback */
969 unsigned int should_sort; /* list to be sorted before flushing? */
971 #define BLK_MAX_REQUEST_COUNT 16
974 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
976 struct list_head list;
977 blk_plug_cb_fn callback;
980 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
981 void *data, int size);
982 extern void blk_start_plug(struct blk_plug *);
983 extern void blk_finish_plug(struct blk_plug *);
984 extern void blk_flush_plug_list(struct blk_plug *, bool);
986 static inline void blk_flush_plug(struct task_struct *tsk)
988 struct blk_plug *plug = tsk->plug;
991 blk_flush_plug_list(plug, false);
994 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
996 struct blk_plug *plug = tsk->plug;
999 blk_flush_plug_list(plug, true);
1002 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1004 struct blk_plug *plug = tsk->plug;
1006 return plug && (!list_empty(&plug->list) || !list_empty(&plug->cb_list));
1012 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
1013 extern int blk_queue_start_tag(struct request_queue *, struct request *);
1014 extern struct request *blk_queue_find_tag(struct request_queue *, int);
1015 extern void blk_queue_end_tag(struct request_queue *, struct request *);
1016 extern int blk_queue_init_tags(struct request_queue *, int, struct blk_queue_tag *);
1017 extern void blk_queue_free_tags(struct request_queue *);
1018 extern int blk_queue_resize_tags(struct request_queue *, int);
1019 extern void blk_queue_invalidate_tags(struct request_queue *);
1020 extern struct blk_queue_tag *blk_init_tags(int);
1021 extern void blk_free_tags(struct blk_queue_tag *);
1023 static inline struct request *blk_map_queue_find_tag(struct blk_queue_tag *bqt,
1026 if (unlikely(bqt == NULL || tag >= bqt->real_max_depth))
1028 return bqt->tag_index[tag];
1031 #define BLKDEV_DISCARD_SECURE 0x01 /* secure discard */
1033 extern int blkdev_issue_flush(struct block_device *, gfp_t, sector_t *);
1034 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1035 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1036 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1037 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1038 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1039 sector_t nr_sects, gfp_t gfp_mask);
1040 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1041 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1043 return blkdev_issue_discard(sb->s_bdev, block << (sb->s_blocksize_bits - 9),
1044 nr_blocks << (sb->s_blocksize_bits - 9),
1047 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1048 sector_t nr_blocks, gfp_t gfp_mask)
1050 return blkdev_issue_zeroout(sb->s_bdev,
1051 block << (sb->s_blocksize_bits - 9),
1052 nr_blocks << (sb->s_blocksize_bits - 9),
1056 extern int blk_verify_command(unsigned char *cmd, fmode_t has_write_perm);
1058 enum blk_default_limits {
1059 BLK_MAX_SEGMENTS = 128,
1060 BLK_SAFE_MAX_SECTORS = 255,
1061 BLK_DEF_MAX_SECTORS = 1024,
1062 BLK_MAX_SEGMENT_SIZE = 65536,
1063 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1066 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
1068 static inline unsigned long queue_bounce_pfn(struct request_queue *q)
1070 return q->limits.bounce_pfn;
1073 static inline unsigned long queue_segment_boundary(struct request_queue *q)
1075 return q->limits.seg_boundary_mask;
1078 static inline unsigned int queue_max_sectors(struct request_queue *q)
1080 return q->limits.max_sectors;
1083 static inline unsigned int queue_max_hw_sectors(struct request_queue *q)
1085 return q->limits.max_hw_sectors;
1088 static inline unsigned short queue_max_segments(struct request_queue *q)
1090 return q->limits.max_segments;
1093 static inline unsigned int queue_max_segment_size(struct request_queue *q)
1095 return q->limits.max_segment_size;
1098 static inline unsigned short queue_logical_block_size(struct request_queue *q)
1102 if (q && q->limits.logical_block_size)
1103 retval = q->limits.logical_block_size;
1108 static inline unsigned short bdev_logical_block_size(struct block_device *bdev)
1110 return queue_logical_block_size(bdev_get_queue(bdev));
1113 static inline unsigned int queue_physical_block_size(struct request_queue *q)
1115 return q->limits.physical_block_size;
1118 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1120 return queue_physical_block_size(bdev_get_queue(bdev));
1123 static inline unsigned int queue_io_min(struct request_queue *q)
1125 return q->limits.io_min;
1128 static inline int bdev_io_min(struct block_device *bdev)
1130 return queue_io_min(bdev_get_queue(bdev));
1133 static inline unsigned int queue_io_opt(struct request_queue *q)
1135 return q->limits.io_opt;
1138 static inline int bdev_io_opt(struct block_device *bdev)
1140 return queue_io_opt(bdev_get_queue(bdev));
1143 static inline int queue_alignment_offset(struct request_queue *q)
1145 if (q->limits.misaligned)
1148 return q->limits.alignment_offset;
1151 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1153 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1154 unsigned int alignment = (sector << 9) & (granularity - 1);
1156 return (granularity + lim->alignment_offset - alignment)
1157 & (granularity - 1);
1160 static inline int bdev_alignment_offset(struct block_device *bdev)
1162 struct request_queue *q = bdev_get_queue(bdev);
1164 if (q->limits.misaligned)
1167 if (bdev != bdev->bd_contains)
1168 return bdev->bd_part->alignment_offset;
1170 return q->limits.alignment_offset;
1173 static inline int queue_discard_alignment(struct request_queue *q)
1175 if (q->limits.discard_misaligned)
1178 return q->limits.discard_alignment;
1181 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1183 unsigned int alignment = (sector << 9) & (lim->discard_granularity - 1);
1185 if (!lim->max_discard_sectors)
1188 return (lim->discard_granularity + lim->discard_alignment - alignment)
1189 & (lim->discard_granularity - 1);
1192 static inline int bdev_discard_alignment(struct block_device *bdev)
1194 struct request_queue *q = bdev_get_queue(bdev);
1196 if (bdev != bdev->bd_contains)
1197 return bdev->bd_part->discard_alignment;
1199 return q->limits.discard_alignment;
1202 static inline unsigned int queue_discard_zeroes_data(struct request_queue *q)
1204 if (q->limits.max_discard_sectors && q->limits.discard_zeroes_data == 1)
1210 static inline unsigned int bdev_discard_zeroes_data(struct block_device *bdev)
1212 return queue_discard_zeroes_data(bdev_get_queue(bdev));
1215 static inline unsigned int bdev_write_same(struct block_device *bdev)
1217 struct request_queue *q = bdev_get_queue(bdev);
1220 return q->limits.max_write_same_sectors;
1225 static inline int queue_dma_alignment(struct request_queue *q)
1227 return q ? q->dma_alignment : 511;
1230 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1233 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1234 return !(addr & alignment) && !(len & alignment);
1237 /* assumes size > 256 */
1238 static inline unsigned int blksize_bits(unsigned int size)
1240 unsigned int bits = 8;
1244 } while (size > 256);
1248 static inline unsigned int block_size(struct block_device *bdev)
1250 return bdev->bd_block_size;
1253 static inline bool queue_flush_queueable(struct request_queue *q)
1255 return !q->flush_not_queueable;
1258 typedef struct {struct page *v;} Sector;
1260 unsigned char *read_dev_sector(struct block_device *, sector_t, Sector *);
1262 static inline void put_dev_sector(Sector p)
1264 page_cache_release(p.v);
1268 int kblockd_schedule_work(struct request_queue *q, struct work_struct *work);
1270 #ifdef CONFIG_BLK_CGROUP
1272 * This should not be using sched_clock(). A real patch is in progress
1273 * to fix this up, until that is in place we need to disable preemption
1274 * around sched_clock() in this function and set_io_start_time_ns().
1276 static inline void set_start_time_ns(struct request *req)
1279 req->start_time_ns = sched_clock();
1283 static inline void set_io_start_time_ns(struct request *req)
1286 req->io_start_time_ns = sched_clock();
1290 static inline uint64_t rq_start_time_ns(struct request *req)
1292 return req->start_time_ns;
1295 static inline uint64_t rq_io_start_time_ns(struct request *req)
1297 return req->io_start_time_ns;
1300 static inline void set_start_time_ns(struct request *req) {}
1301 static inline void set_io_start_time_ns(struct request *req) {}
1302 static inline uint64_t rq_start_time_ns(struct request *req)
1306 static inline uint64_t rq_io_start_time_ns(struct request *req)
1312 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1313 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1314 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1315 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1317 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1319 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1320 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1322 struct blk_integrity_exchg {
1326 unsigned int data_size;
1327 unsigned short sector_size;
1328 const char *disk_name;
1331 typedef void (integrity_gen_fn) (struct blk_integrity_exchg *);
1332 typedef int (integrity_vrfy_fn) (struct blk_integrity_exchg *);
1333 typedef void (integrity_set_tag_fn) (void *, void *, unsigned int);
1334 typedef void (integrity_get_tag_fn) (void *, void *, unsigned int);
1336 struct blk_integrity {
1337 integrity_gen_fn *generate_fn;
1338 integrity_vrfy_fn *verify_fn;
1339 integrity_set_tag_fn *set_tag_fn;
1340 integrity_get_tag_fn *get_tag_fn;
1342 unsigned short flags;
1343 unsigned short tuple_size;
1344 unsigned short sector_size;
1345 unsigned short tag_size;
1349 struct kobject kobj;
1352 extern bool blk_integrity_is_initialized(struct gendisk *);
1353 extern int blk_integrity_register(struct gendisk *, struct blk_integrity *);
1354 extern void blk_integrity_unregister(struct gendisk *);
1355 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1356 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1357 struct scatterlist *);
1358 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1359 extern int blk_integrity_merge_rq(struct request_queue *, struct request *,
1361 extern int blk_integrity_merge_bio(struct request_queue *, struct request *,
1365 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1367 return bdev->bd_disk->integrity;
1370 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1372 return disk->integrity;
1375 static inline int blk_integrity_rq(struct request *rq)
1377 if (rq->bio == NULL)
1380 return bio_integrity(rq->bio);
1383 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1386 q->limits.max_integrity_segments = segs;
1389 static inline unsigned short
1390 queue_max_integrity_segments(struct request_queue *q)
1392 return q->limits.max_integrity_segments;
1395 #else /* CONFIG_BLK_DEV_INTEGRITY */
1398 struct block_device;
1400 struct blk_integrity;
1402 static inline int blk_integrity_rq(struct request *rq)
1406 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1411 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1413 struct scatterlist *s)
1417 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1421 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1425 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1429 static inline int blk_integrity_register(struct gendisk *d,
1430 struct blk_integrity *b)
1434 static inline void blk_integrity_unregister(struct gendisk *d)
1437 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1441 static inline unsigned short queue_max_integrity_segments(struct request_queue *q)
1445 static inline int blk_integrity_merge_rq(struct request_queue *rq,
1451 static inline int blk_integrity_merge_bio(struct request_queue *rq,
1457 static inline bool blk_integrity_is_initialized(struct gendisk *g)
1462 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1464 struct block_device_operations {
1465 int (*open) (struct block_device *, fmode_t);
1466 int (*release) (struct gendisk *, fmode_t);
1467 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1468 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1469 int (*direct_access) (struct block_device *, sector_t,
1470 void **, unsigned long *);
1471 unsigned int (*check_events) (struct gendisk *disk,
1472 unsigned int clearing);
1473 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1474 int (*media_changed) (struct gendisk *);
1475 void (*unlock_native_capacity) (struct gendisk *);
1476 int (*revalidate_disk) (struct gendisk *);
1477 int (*getgeo)(struct block_device *, struct hd_geometry *);
1478 /* this callback is with swap_lock and sometimes page table lock held */
1479 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1480 struct module *owner;
1483 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1485 #else /* CONFIG_BLOCK */
1487 * stubs for when the block layer is configured out
1489 #define buffer_heads_over_limit 0
1491 static inline long nr_blockdev_pages(void)
1499 static inline void blk_start_plug(struct blk_plug *plug)
1503 static inline void blk_finish_plug(struct blk_plug *plug)
1507 static inline void blk_flush_plug(struct task_struct *task)
1511 static inline void blk_schedule_flush_plug(struct task_struct *task)
1516 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1521 #endif /* CONFIG_BLOCK */