Merge tag 'v3.3.7' into 3.3/master
[zen-stable.git] / include / linux / blkdev.h
blob606cf339bb561f6b526c2a58ddcf1378c8376bf1
1 #ifndef _LINUX_BLKDEV_H
2 #define _LINUX_BLKDEV_H
4 #ifdef CONFIG_BLOCK
6 #include <linux/sched.h>
7 #include <linux/major.h>
8 #include <linux/genhd.h>
9 #include <linux/list.h>
10 #include <linux/timer.h>
11 #include <linux/workqueue.h>
12 #include <linux/pagemap.h>
13 #include <linux/backing-dev.h>
14 #include <linux/wait.h>
15 #include <linux/mempool.h>
16 #include <linux/bio.h>
17 #include <linux/stringify.h>
18 #include <linux/gfp.h>
19 #include <linux/bsg.h>
20 #include <linux/smp.h>
22 #include <asm/scatterlist.h>
24 struct module;
25 struct scsi_ioctl_command;
27 struct request_queue;
28 struct elevator_queue;
29 struct request_pm_state;
30 struct blk_trace;
31 struct request;
32 struct sg_io_hdr;
33 struct bsg_job;
35 #define BLKDEV_MIN_RQ 4
36 #define BLKDEV_MAX_RQ 128 /* Default maximum */
38 struct request;
39 typedef void (rq_end_io_fn)(struct request *, int);
41 struct request_list {
43 * count[], starved[], and wait[] are indexed by
44 * BLK_RW_SYNC/BLK_RW_ASYNC
46 int count[2];
47 int starved[2];
48 int elvpriv;
49 mempool_t *rq_pool;
50 wait_queue_head_t wait[2];
54 * request command types
56 enum rq_cmd_type_bits {
57 REQ_TYPE_FS = 1, /* fs request */
58 REQ_TYPE_BLOCK_PC, /* scsi command */
59 REQ_TYPE_SENSE, /* sense request */
60 REQ_TYPE_PM_SUSPEND, /* suspend request */
61 REQ_TYPE_PM_RESUME, /* resume request */
62 REQ_TYPE_PM_SHUTDOWN, /* shutdown request */
63 REQ_TYPE_SPECIAL, /* driver defined type */
65 * for ATA/ATAPI devices. this really doesn't belong here, ide should
66 * use REQ_TYPE_SPECIAL and use rq->cmd[0] with the range of driver
67 * private REQ_LB opcodes to differentiate what type of request this is
69 REQ_TYPE_ATA_TASKFILE,
70 REQ_TYPE_ATA_PC,
73 #define BLK_MAX_CDB 16
76 * try to put the fields that are referenced together in the same cacheline.
77 * if you modify this structure, be sure to check block/blk-core.c:blk_rq_init()
78 * as well!
80 struct request {
81 struct list_head queuelist;
82 struct call_single_data csd;
84 struct request_queue *q;
86 unsigned int cmd_flags;
87 enum rq_cmd_type_bits cmd_type;
88 unsigned long atomic_flags;
90 int cpu;
92 /* the following two fields are internal, NEVER access directly */
93 unsigned int __data_len; /* total data len */
94 sector_t __sector; /* sector cursor */
96 struct bio *bio;
97 struct bio *biotail;
99 struct hlist_node hash; /* merge hash */
101 * The rb_node is only used inside the io scheduler, requests
102 * are pruned when moved to the dispatch queue. So let the
103 * completion_data share space with the rb_node.
105 union {
106 struct rb_node rb_node; /* sort/lookup */
107 void *completion_data;
111 * Three pointers are available for the IO schedulers, if they need
112 * more they have to dynamically allocate it. Flush requests are
113 * never put on the IO scheduler. So let the flush fields share
114 * space with the elevator data.
116 union {
117 struct {
118 struct io_cq *icq;
119 void *priv[2];
120 } elv;
122 struct {
123 unsigned int seq;
124 struct list_head list;
125 rq_end_io_fn *saved_end_io;
126 } flush;
129 struct gendisk *rq_disk;
130 struct hd_struct *part;
131 unsigned long start_time;
132 #ifdef CONFIG_BLK_CGROUP
133 unsigned long long start_time_ns;
134 unsigned long long io_start_time_ns; /* when passed to hardware */
135 #endif
136 /* Number of scatter-gather DMA addr+len pairs after
137 * physical address coalescing is performed.
139 unsigned short nr_phys_segments;
140 #if defined(CONFIG_BLK_DEV_INTEGRITY)
141 unsigned short nr_integrity_segments;
142 #endif
144 unsigned short ioprio;
146 int ref_count;
148 void *special; /* opaque pointer available for LLD use */
149 char *buffer; /* kaddr of the current segment if available */
151 int tag;
152 int errors;
155 * when request is used as a packet command carrier
157 unsigned char __cmd[BLK_MAX_CDB];
158 unsigned char *cmd;
159 unsigned short cmd_len;
161 unsigned int extra_len; /* length of alignment and padding */
162 unsigned int sense_len;
163 unsigned int resid_len; /* residual count */
164 void *sense;
166 unsigned long deadline;
167 struct list_head timeout_list;
168 unsigned int timeout;
169 int retries;
172 * completion callback.
174 rq_end_io_fn *end_io;
175 void *end_io_data;
177 /* for bidi */
178 struct request *next_rq;
181 static inline unsigned short req_get_ioprio(struct request *req)
183 return req->ioprio;
187 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
188 * requests. Some step values could eventually be made generic.
190 struct request_pm_state
192 /* PM state machine step value, currently driver specific */
193 int pm_step;
194 /* requested PM state value (S1, S2, S3, S4, ...) */
195 u32 pm_state;
196 void* data; /* for driver use */
199 #include <linux/elevator.h>
201 typedef void (request_fn_proc) (struct request_queue *q);
202 typedef void (make_request_fn) (struct request_queue *q, struct bio *bio);
203 typedef int (prep_rq_fn) (struct request_queue *, struct request *);
204 typedef void (unprep_rq_fn) (struct request_queue *, struct request *);
206 struct bio_vec;
207 struct bvec_merge_data {
208 struct block_device *bi_bdev;
209 sector_t bi_sector;
210 unsigned bi_size;
211 unsigned long bi_rw;
213 typedef int (merge_bvec_fn) (struct request_queue *, struct bvec_merge_data *,
214 struct bio_vec *);
215 typedef void (softirq_done_fn)(struct request *);
216 typedef int (dma_drain_needed_fn)(struct request *);
217 typedef int (lld_busy_fn) (struct request_queue *q);
218 typedef int (bsg_job_fn) (struct bsg_job *);
220 enum blk_eh_timer_return {
221 BLK_EH_NOT_HANDLED,
222 BLK_EH_HANDLED,
223 BLK_EH_RESET_TIMER,
226 typedef enum blk_eh_timer_return (rq_timed_out_fn)(struct request *);
228 enum blk_queue_state {
229 Queue_down,
230 Queue_up,
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;
254 unsigned int io_min;
255 unsigned int io_opt;
256 unsigned int max_discard_sectors;
257 unsigned int discard_granularity;
258 unsigned int discard_alignment;
260 unsigned short logical_block_size;
261 unsigned short max_segments;
262 unsigned short max_integrity_segments;
264 unsigned char misaligned;
265 unsigned char discard_misaligned;
266 unsigned char cluster;
267 unsigned char discard_zeroes_data;
270 struct request_queue {
272 * Together with queue_head for cacheline sharing
274 struct list_head queue_head;
275 struct request *last_merge;
276 struct elevator_queue *elevator;
279 * the queue request freelist, one for reads and one for writes
281 struct request_list rq;
283 request_fn_proc *request_fn;
284 make_request_fn *make_request_fn;
285 prep_rq_fn *prep_rq_fn;
286 unprep_rq_fn *unprep_rq_fn;
287 merge_bvec_fn *merge_bvec_fn;
288 softirq_done_fn *softirq_done_fn;
289 rq_timed_out_fn *rq_timed_out_fn;
290 dma_drain_needed_fn *dma_drain_needed;
291 lld_busy_fn *lld_busy_fn;
294 * Dispatch queue sorting
296 sector_t end_sector;
297 struct request *boundary_rq;
300 * Delayed queue handling
302 struct delayed_work delay_work;
304 struct backing_dev_info backing_dev_info;
307 * The queue owner gets to use this for whatever they like.
308 * ll_rw_blk doesn't touch it.
310 void *queuedata;
313 * various queue flags, see QUEUE_* below
315 unsigned long queue_flags;
318 * ida allocated id for this queue. Used to index queues from
319 * ioctx.
321 int id;
324 * queue needs bounce pages for pages above this limit
326 gfp_t bounce_gfp;
329 * protects queue structures from reentrancy. ->__queue_lock should
330 * _never_ be used directly, it is queue private. always use
331 * ->queue_lock.
333 spinlock_t __queue_lock;
334 spinlock_t *queue_lock;
337 * queue kobject
339 struct kobject kobj;
342 * queue settings
344 unsigned long nr_requests; /* Max # of requests */
345 unsigned int nr_congestion_on;
346 unsigned int nr_congestion_off;
347 unsigned int nr_batching;
349 unsigned int dma_drain_size;
350 void *dma_drain_buffer;
351 unsigned int dma_pad_mask;
352 unsigned int dma_alignment;
354 struct blk_queue_tag *queue_tags;
355 struct list_head tag_busy_list;
357 unsigned int nr_sorted;
358 unsigned int in_flight[2];
360 unsigned int rq_timeout;
361 struct timer_list timeout;
362 struct list_head timeout_list;
364 struct list_head icq_list;
366 struct queue_limits limits;
369 * sg stuff
371 unsigned int sg_timeout;
372 unsigned int sg_reserved_size;
373 int node;
374 #ifdef CONFIG_BLK_DEV_IO_TRACE
375 struct blk_trace *blk_trace;
376 #endif
378 * for flush operations
380 unsigned int flush_flags;
381 unsigned int flush_not_queueable:1;
382 unsigned int flush_queue_delayed:1;
383 unsigned int flush_pending_idx:1;
384 unsigned int flush_running_idx:1;
385 unsigned long flush_pending_since;
386 struct list_head flush_queue[2];
387 struct list_head flush_data_in_flight;
388 struct request flush_rq;
390 struct mutex sysfs_lock;
392 #if defined(CONFIG_BLK_DEV_BSG)
393 bsg_job_fn *bsg_job_fn;
394 int bsg_job_size;
395 struct bsg_class_device bsg_dev;
396 #endif
398 #ifdef CONFIG_BLK_DEV_THROTTLING
399 /* Throttle data */
400 struct throtl_data *td;
401 #endif
404 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
405 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
406 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
407 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
408 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
409 #define QUEUE_FLAG_ELVSWITCH 6 /* don't use elevator, just do FIFO */
410 #define QUEUE_FLAG_BIDI 7 /* queue supports bidi requests */
411 #define QUEUE_FLAG_NOMERGES 8 /* disable merge attempts */
412 #define QUEUE_FLAG_SAME_COMP 9 /* complete on same CPU-group */
413 #define QUEUE_FLAG_FAIL_IO 10 /* fake timeout */
414 #define QUEUE_FLAG_STACKABLE 11 /* supports request stacking */
415 #define QUEUE_FLAG_NONROT 12 /* non-rotational device (SSD) */
416 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
417 #define QUEUE_FLAG_IO_STAT 13 /* do IO stats */
418 #define QUEUE_FLAG_DISCARD 14 /* supports DISCARD */
419 #define QUEUE_FLAG_NOXMERGES 15 /* No extended merges */
420 #define QUEUE_FLAG_ADD_RANDOM 16 /* Contributes to random pool */
421 #define QUEUE_FLAG_SECDISCARD 17 /* supports SECDISCARD */
422 #define QUEUE_FLAG_SAME_FORCE 18 /* force complete on same CPU */
424 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
425 (1 << QUEUE_FLAG_STACKABLE) | \
426 (1 << QUEUE_FLAG_SAME_COMP) | \
427 (1 << QUEUE_FLAG_ADD_RANDOM))
429 static inline int queue_is_locked(struct request_queue *q)
431 #ifdef CONFIG_SMP
432 spinlock_t *lock = q->queue_lock;
433 return lock && spin_is_locked(lock);
434 #else
435 return 1;
436 #endif
439 static inline void queue_flag_set_unlocked(unsigned int flag,
440 struct request_queue *q)
442 __set_bit(flag, &q->queue_flags);
445 static inline int queue_flag_test_and_clear(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 __clear_bit(flag, &q->queue_flags);
452 return 1;
455 return 0;
458 static inline int queue_flag_test_and_set(unsigned int flag,
459 struct request_queue *q)
461 WARN_ON_ONCE(!queue_is_locked(q));
463 if (!test_bit(flag, &q->queue_flags)) {
464 __set_bit(flag, &q->queue_flags);
465 return 0;
468 return 1;
471 static inline void queue_flag_set(unsigned int flag, struct request_queue *q)
473 WARN_ON_ONCE(!queue_is_locked(q));
474 __set_bit(flag, &q->queue_flags);
477 static inline void queue_flag_clear_unlocked(unsigned int flag,
478 struct request_queue *q)
480 __clear_bit(flag, &q->queue_flags);
483 static inline int queue_in_flight(struct request_queue *q)
485 return q->in_flight[0] + q->in_flight[1];
488 static inline void queue_flag_clear(unsigned int flag, struct request_queue *q)
490 WARN_ON_ONCE(!queue_is_locked(q));
491 __clear_bit(flag, &q->queue_flags);
494 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
495 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
496 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
497 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
498 #define blk_queue_noxmerges(q) \
499 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
500 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
501 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
502 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
503 #define blk_queue_stackable(q) \
504 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
505 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
506 #define blk_queue_secdiscard(q) (blk_queue_discard(q) && \
507 test_bit(QUEUE_FLAG_SECDISCARD, &(q)->queue_flags))
509 #define blk_noretry_request(rq) \
510 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
511 REQ_FAILFAST_DRIVER))
513 #define blk_account_rq(rq) \
514 (((rq)->cmd_flags & REQ_STARTED) && \
515 ((rq)->cmd_type == REQ_TYPE_FS || \
516 ((rq)->cmd_flags & REQ_DISCARD)))
518 #define blk_pm_request(rq) \
519 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
520 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
522 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
523 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
524 /* rq->queuelist of dequeued request must be list_empty() */
525 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
527 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
529 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
531 static inline unsigned int blk_queue_cluster(struct request_queue *q)
533 return q->limits.cluster;
537 * We regard a request as sync, if either a read or a sync write
539 static inline bool rw_is_sync(unsigned int rw_flags)
541 return !(rw_flags & REQ_WRITE) || (rw_flags & REQ_SYNC);
544 static inline bool rq_is_sync(struct request *rq)
546 return rw_is_sync(rq->cmd_flags);
549 static inline int blk_queue_full(struct request_queue *q, int sync)
551 if (sync)
552 return test_bit(QUEUE_FLAG_SYNCFULL, &q->queue_flags);
553 return test_bit(QUEUE_FLAG_ASYNCFULL, &q->queue_flags);
556 static inline void blk_set_queue_full(struct request_queue *q, int sync)
558 if (sync)
559 queue_flag_set(QUEUE_FLAG_SYNCFULL, q);
560 else
561 queue_flag_set(QUEUE_FLAG_ASYNCFULL, q);
564 static inline void blk_clear_queue_full(struct request_queue *q, int sync)
566 if (sync)
567 queue_flag_clear(QUEUE_FLAG_SYNCFULL, q);
568 else
569 queue_flag_clear(QUEUE_FLAG_ASYNCFULL, q);
574 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
575 * it already be started by driver.
577 #define RQ_NOMERGE_FLAGS \
578 (REQ_NOMERGE | REQ_STARTED | REQ_SOFTBARRIER | REQ_FLUSH | REQ_FUA)
579 #define rq_mergeable(rq) \
580 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
581 (((rq)->cmd_flags & REQ_DISCARD) || \
582 (rq)->cmd_type == REQ_TYPE_FS))
585 * q->prep_rq_fn return values
587 #define BLKPREP_OK 0 /* serve it */
588 #define BLKPREP_KILL 1 /* fatal error, kill */
589 #define BLKPREP_DEFER 2 /* leave on queue */
591 extern unsigned long blk_max_low_pfn, blk_max_pfn;
594 * standard bounce addresses:
596 * BLK_BOUNCE_HIGH : bounce all highmem pages
597 * BLK_BOUNCE_ANY : don't bounce anything
598 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
601 #if BITS_PER_LONG == 32
602 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
603 #else
604 #define BLK_BOUNCE_HIGH -1ULL
605 #endif
606 #define BLK_BOUNCE_ANY (-1ULL)
607 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
610 * default timeout for SG_IO if none specified
612 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
613 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
615 #ifdef CONFIG_BOUNCE
616 extern int init_emergency_isa_pool(void);
617 extern void blk_queue_bounce(struct request_queue *q, struct bio **bio);
618 #else
619 static inline int init_emergency_isa_pool(void)
621 return 0;
623 static inline void blk_queue_bounce(struct request_queue *q, struct bio **bio)
626 #endif /* CONFIG_MMU */
628 struct rq_map_data {
629 struct page **pages;
630 int page_order;
631 int nr_entries;
632 unsigned long offset;
633 int null_mapped;
634 int from_user;
637 struct req_iterator {
638 int i;
639 struct bio *bio;
642 /* This should not be used directly - use rq_for_each_segment */
643 #define for_each_bio(_bio) \
644 for (; _bio; _bio = _bio->bi_next)
645 #define __rq_for_each_bio(_bio, rq) \
646 if ((rq->bio)) \
647 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
649 #define rq_for_each_segment(bvl, _rq, _iter) \
650 __rq_for_each_bio(_iter.bio, _rq) \
651 bio_for_each_segment(bvl, _iter.bio, _iter.i)
653 #define rq_iter_last(rq, _iter) \
654 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
656 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
657 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
658 #endif
659 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
660 extern void rq_flush_dcache_pages(struct request *rq);
661 #else
662 static inline void rq_flush_dcache_pages(struct request *rq)
665 #endif
667 extern int blk_register_queue(struct gendisk *disk);
668 extern void blk_unregister_queue(struct gendisk *disk);
669 extern void generic_make_request(struct bio *bio);
670 extern void blk_rq_init(struct request_queue *q, struct request *rq);
671 extern void blk_put_request(struct request *);
672 extern void __blk_put_request(struct request_queue *, struct request *);
673 extern struct request *blk_get_request(struct request_queue *, int, gfp_t);
674 extern struct request *blk_make_request(struct request_queue *, struct bio *,
675 gfp_t);
676 extern void blk_requeue_request(struct request_queue *, struct request *);
677 extern void blk_add_request_payload(struct request *rq, struct page *page,
678 unsigned int len);
679 extern int blk_rq_check_limits(struct request_queue *q, struct request *rq);
680 extern int blk_lld_busy(struct request_queue *q);
681 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
682 struct bio_set *bs, gfp_t gfp_mask,
683 int (*bio_ctr)(struct bio *, struct bio *, void *),
684 void *data);
685 extern void blk_rq_unprep_clone(struct request *rq);
686 extern int blk_insert_cloned_request(struct request_queue *q,
687 struct request *rq);
688 extern void blk_delay_queue(struct request_queue *, unsigned long);
689 extern void blk_recount_segments(struct request_queue *, struct bio *);
690 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
691 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
692 unsigned int, void __user *);
693 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
694 unsigned int, void __user *);
695 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
696 struct scsi_ioctl_command __user *);
698 extern void blk_queue_bio(struct request_queue *q, struct bio *bio);
701 * A queue has just exitted congestion. Note this in the global counter of
702 * congested queues, and wake up anyone who was waiting for requests to be
703 * put back.
705 static inline void blk_clear_queue_congested(struct request_queue *q, int sync)
707 clear_bdi_congested(&q->backing_dev_info, sync);
711 * A queue has just entered congestion. Flag that in the queue's VM-visible
712 * state flags and increment the global gounter of congested queues.
714 static inline void blk_set_queue_congested(struct request_queue *q, int sync)
716 set_bdi_congested(&q->backing_dev_info, sync);
719 extern void blk_start_queue(struct request_queue *q);
720 extern void blk_stop_queue(struct request_queue *q);
721 extern void blk_sync_queue(struct request_queue *q);
722 extern void __blk_stop_queue(struct request_queue *q);
723 extern void __blk_run_queue(struct request_queue *q);
724 extern void blk_run_queue(struct request_queue *);
725 extern void blk_run_queue_async(struct request_queue *q);
726 extern int blk_rq_map_user(struct request_queue *, struct request *,
727 struct rq_map_data *, void __user *, unsigned long,
728 gfp_t);
729 extern int blk_rq_unmap_user(struct bio *);
730 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
731 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
732 struct rq_map_data *, struct sg_iovec *, int,
733 unsigned int, gfp_t);
734 extern int blk_execute_rq(struct request_queue *, struct gendisk *,
735 struct request *, int);
736 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
737 struct request *, int, rq_end_io_fn *);
739 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
741 return bdev->bd_disk->queue;
745 * blk_rq_pos() : the current sector
746 * blk_rq_bytes() : bytes left in the entire request
747 * blk_rq_cur_bytes() : bytes left in the current segment
748 * blk_rq_err_bytes() : bytes left till the next error boundary
749 * blk_rq_sectors() : sectors left in the entire request
750 * blk_rq_cur_sectors() : sectors left in the current segment
752 static inline sector_t blk_rq_pos(const struct request *rq)
754 return rq->__sector;
757 static inline unsigned int blk_rq_bytes(const struct request *rq)
759 return rq->__data_len;
762 static inline int blk_rq_cur_bytes(const struct request *rq)
764 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
767 extern unsigned int blk_rq_err_bytes(const struct request *rq);
769 static inline unsigned int blk_rq_sectors(const struct request *rq)
771 return blk_rq_bytes(rq) >> 9;
774 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
776 return blk_rq_cur_bytes(rq) >> 9;
780 * Request issue related functions.
782 extern struct request *blk_peek_request(struct request_queue *q);
783 extern void blk_start_request(struct request *rq);
784 extern struct request *blk_fetch_request(struct request_queue *q);
787 * Request completion related functions.
789 * blk_update_request() completes given number of bytes and updates
790 * the request without completing it.
792 * blk_end_request() and friends. __blk_end_request() must be called
793 * with the request queue spinlock acquired.
795 * Several drivers define their own end_request and call
796 * blk_end_request() for parts of the original function.
797 * This prevents code duplication in drivers.
799 extern bool blk_update_request(struct request *rq, int error,
800 unsigned int nr_bytes);
801 extern bool blk_end_request(struct request *rq, int error,
802 unsigned int nr_bytes);
803 extern void blk_end_request_all(struct request *rq, int error);
804 extern bool blk_end_request_cur(struct request *rq, int error);
805 extern bool blk_end_request_err(struct request *rq, int error);
806 extern bool __blk_end_request(struct request *rq, int error,
807 unsigned int nr_bytes);
808 extern void __blk_end_request_all(struct request *rq, int error);
809 extern bool __blk_end_request_cur(struct request *rq, int error);
810 extern bool __blk_end_request_err(struct request *rq, int error);
812 extern void blk_complete_request(struct request *);
813 extern void __blk_complete_request(struct request *);
814 extern void blk_abort_request(struct request *);
815 extern void blk_abort_queue(struct request_queue *);
816 extern void blk_unprep_request(struct request *);
819 * Access functions for manipulating queue properties
821 extern struct request_queue *blk_init_queue_node(request_fn_proc *rfn,
822 spinlock_t *lock, int node_id);
823 extern struct request_queue *blk_init_queue(request_fn_proc *, spinlock_t *);
824 extern struct request_queue *blk_init_allocated_queue(struct request_queue *,
825 request_fn_proc *, spinlock_t *);
826 extern void blk_cleanup_queue(struct request_queue *);
827 extern void blk_queue_make_request(struct request_queue *, make_request_fn *);
828 extern void blk_queue_bounce_limit(struct request_queue *, u64);
829 extern void blk_limits_max_hw_sectors(struct queue_limits *, unsigned int);
830 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
831 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
832 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
833 extern void blk_queue_max_discard_sectors(struct request_queue *q,
834 unsigned int max_discard_sectors);
835 extern void blk_queue_logical_block_size(struct request_queue *, unsigned short);
836 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
837 extern void blk_queue_alignment_offset(struct request_queue *q,
838 unsigned int alignment);
839 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
840 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
841 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
842 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
843 extern void blk_set_default_limits(struct queue_limits *lim);
844 extern void blk_set_stacking_limits(struct queue_limits *lim);
845 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
846 sector_t offset);
847 extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev,
848 sector_t offset);
849 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
850 sector_t offset);
851 extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b);
852 extern void blk_queue_dma_pad(struct request_queue *, unsigned int);
853 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
854 extern int blk_queue_dma_drain(struct request_queue *q,
855 dma_drain_needed_fn *dma_drain_needed,
856 void *buf, unsigned int size);
857 extern void blk_queue_lld_busy(struct request_queue *q, lld_busy_fn *fn);
858 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
859 extern void blk_queue_prep_rq(struct request_queue *, prep_rq_fn *pfn);
860 extern void blk_queue_unprep_rq(struct request_queue *, unprep_rq_fn *ufn);
861 extern void blk_queue_merge_bvec(struct request_queue *, merge_bvec_fn *);
862 extern void blk_queue_dma_alignment(struct request_queue *, int);
863 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
864 extern void blk_queue_softirq_done(struct request_queue *, softirq_done_fn *);
865 extern void blk_queue_rq_timed_out(struct request_queue *, rq_timed_out_fn *);
866 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
867 extern void blk_queue_flush(struct request_queue *q, unsigned int flush);
868 extern void blk_queue_flush_queueable(struct request_queue *q, bool queueable);
869 extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
871 extern int blk_rq_map_sg(struct request_queue *, struct request *, struct scatterlist *);
872 extern void blk_dump_rq_flags(struct request *, char *);
873 extern long nr_blockdev_pages(void);
875 bool __must_check blk_get_queue(struct request_queue *);
876 struct request_queue *blk_alloc_queue(gfp_t);
877 struct request_queue *blk_alloc_queue_node(gfp_t, int);
878 extern void blk_put_queue(struct request_queue *);
881 * blk_plug permits building a queue of related requests by holding the I/O
882 * fragments for a short period. This allows merging of sequential requests
883 * into single larger request. As the requests are moved from a per-task list to
884 * the device's request_queue in a batch, this results in improved scalability
885 * as the lock contention for request_queue lock is reduced.
887 * It is ok not to disable preemption when adding the request to the plug list
888 * or when attempting a merge, because blk_schedule_flush_list() will only flush
889 * the plug list when the task sleeps by itself. For details, please see
890 * schedule() where blk_schedule_flush_plug() is called.
892 struct blk_plug {
893 unsigned long magic; /* detect uninitialized use-cases */
894 struct list_head list; /* requests */
895 struct list_head cb_list; /* md requires an unplug callback */
896 unsigned int should_sort; /* list to be sorted before flushing? */
898 #define BLK_MAX_REQUEST_COUNT 16
900 struct blk_plug_cb {
901 struct list_head list;
902 void (*callback)(struct blk_plug_cb *);
905 extern void blk_start_plug(struct blk_plug *);
906 extern void blk_finish_plug(struct blk_plug *);
907 extern void blk_flush_plug_list(struct blk_plug *, bool);
909 static inline void blk_flush_plug(struct task_struct *tsk)
911 struct blk_plug *plug = tsk->plug;
913 if (plug)
914 blk_flush_plug_list(plug, false);
917 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
919 struct blk_plug *plug = tsk->plug;
921 if (plug)
922 blk_flush_plug_list(plug, true);
925 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
927 struct blk_plug *plug = tsk->plug;
929 return plug && (!list_empty(&plug->list) || !list_empty(&plug->cb_list));
933 * tag stuff
935 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
936 extern int blk_queue_start_tag(struct request_queue *, struct request *);
937 extern struct request *blk_queue_find_tag(struct request_queue *, int);
938 extern void blk_queue_end_tag(struct request_queue *, struct request *);
939 extern int blk_queue_init_tags(struct request_queue *, int, struct blk_queue_tag *);
940 extern void blk_queue_free_tags(struct request_queue *);
941 extern int blk_queue_resize_tags(struct request_queue *, int);
942 extern void blk_queue_invalidate_tags(struct request_queue *);
943 extern struct blk_queue_tag *blk_init_tags(int);
944 extern void blk_free_tags(struct blk_queue_tag *);
946 static inline struct request *blk_map_queue_find_tag(struct blk_queue_tag *bqt,
947 int tag)
949 if (unlikely(bqt == NULL || tag >= bqt->real_max_depth))
950 return NULL;
951 return bqt->tag_index[tag];
954 #define BLKDEV_DISCARD_SECURE 0x01 /* secure discard */
956 extern int blkdev_issue_flush(struct block_device *, gfp_t, sector_t *);
957 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
958 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
959 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
960 sector_t nr_sects, gfp_t gfp_mask);
961 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
962 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
964 return blkdev_issue_discard(sb->s_bdev, block << (sb->s_blocksize_bits - 9),
965 nr_blocks << (sb->s_blocksize_bits - 9),
966 gfp_mask, flags);
968 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
969 sector_t nr_blocks, gfp_t gfp_mask)
971 return blkdev_issue_zeroout(sb->s_bdev,
972 block << (sb->s_blocksize_bits - 9),
973 nr_blocks << (sb->s_blocksize_bits - 9),
974 gfp_mask);
977 extern int blk_verify_command(unsigned char *cmd, fmode_t has_write_perm);
979 enum blk_default_limits {
980 BLK_MAX_SEGMENTS = 128,
981 BLK_SAFE_MAX_SECTORS = 255,
982 BLK_DEF_MAX_SECTORS = 1024,
983 BLK_MAX_SEGMENT_SIZE = 65536,
984 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
987 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
989 static inline unsigned long queue_bounce_pfn(struct request_queue *q)
991 return q->limits.bounce_pfn;
994 static inline unsigned long queue_segment_boundary(struct request_queue *q)
996 return q->limits.seg_boundary_mask;
999 static inline unsigned int queue_max_sectors(struct request_queue *q)
1001 return q->limits.max_sectors;
1004 static inline unsigned int queue_max_hw_sectors(struct request_queue *q)
1006 return q->limits.max_hw_sectors;
1009 static inline unsigned short queue_max_segments(struct request_queue *q)
1011 return q->limits.max_segments;
1014 static inline unsigned int queue_max_segment_size(struct request_queue *q)
1016 return q->limits.max_segment_size;
1019 static inline unsigned short queue_logical_block_size(struct request_queue *q)
1021 int retval = 512;
1023 if (q && q->limits.logical_block_size)
1024 retval = q->limits.logical_block_size;
1026 return retval;
1029 static inline unsigned short bdev_logical_block_size(struct block_device *bdev)
1031 return queue_logical_block_size(bdev_get_queue(bdev));
1034 static inline unsigned int queue_physical_block_size(struct request_queue *q)
1036 return q->limits.physical_block_size;
1039 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1041 return queue_physical_block_size(bdev_get_queue(bdev));
1044 static inline unsigned int queue_io_min(struct request_queue *q)
1046 return q->limits.io_min;
1049 static inline int bdev_io_min(struct block_device *bdev)
1051 return queue_io_min(bdev_get_queue(bdev));
1054 static inline unsigned int queue_io_opt(struct request_queue *q)
1056 return q->limits.io_opt;
1059 static inline int bdev_io_opt(struct block_device *bdev)
1061 return queue_io_opt(bdev_get_queue(bdev));
1064 static inline int queue_alignment_offset(struct request_queue *q)
1066 if (q->limits.misaligned)
1067 return -1;
1069 return q->limits.alignment_offset;
1072 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1074 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1075 unsigned int alignment = (sector << 9) & (granularity - 1);
1077 return (granularity + lim->alignment_offset - alignment)
1078 & (granularity - 1);
1081 static inline int bdev_alignment_offset(struct block_device *bdev)
1083 struct request_queue *q = bdev_get_queue(bdev);
1085 if (q->limits.misaligned)
1086 return -1;
1088 if (bdev != bdev->bd_contains)
1089 return bdev->bd_part->alignment_offset;
1091 return q->limits.alignment_offset;
1094 static inline int queue_discard_alignment(struct request_queue *q)
1096 if (q->limits.discard_misaligned)
1097 return -1;
1099 return q->limits.discard_alignment;
1102 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1104 unsigned int alignment = (sector << 9) & (lim->discard_granularity - 1);
1106 if (!lim->max_discard_sectors)
1107 return 0;
1109 return (lim->discard_granularity + lim->discard_alignment - alignment)
1110 & (lim->discard_granularity - 1);
1113 static inline unsigned int queue_discard_zeroes_data(struct request_queue *q)
1115 if (q->limits.max_discard_sectors && q->limits.discard_zeroes_data == 1)
1116 return 1;
1118 return 0;
1121 static inline unsigned int bdev_discard_zeroes_data(struct block_device *bdev)
1123 return queue_discard_zeroes_data(bdev_get_queue(bdev));
1126 static inline int queue_dma_alignment(struct request_queue *q)
1128 return q ? q->dma_alignment : 511;
1131 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1132 unsigned int len)
1134 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1135 return !(addr & alignment) && !(len & alignment);
1138 /* assumes size > 256 */
1139 static inline unsigned int blksize_bits(unsigned int size)
1141 unsigned int bits = 8;
1142 do {
1143 bits++;
1144 size >>= 1;
1145 } while (size > 256);
1146 return bits;
1149 static inline unsigned int block_size(struct block_device *bdev)
1151 return bdev->bd_block_size;
1154 static inline bool queue_flush_queueable(struct request_queue *q)
1156 return !q->flush_not_queueable;
1159 typedef struct {struct page *v;} Sector;
1161 unsigned char *read_dev_sector(struct block_device *, sector_t, Sector *);
1163 static inline void put_dev_sector(Sector p)
1165 page_cache_release(p.v);
1168 struct work_struct;
1169 int kblockd_schedule_work(struct request_queue *q, struct work_struct *work);
1171 #ifdef CONFIG_BLK_CGROUP
1173 * This should not be using sched_clock(). A real patch is in progress
1174 * to fix this up, until that is in place we need to disable preemption
1175 * around sched_clock() in this function and set_io_start_time_ns().
1177 static inline void set_start_time_ns(struct request *req)
1179 preempt_disable();
1180 req->start_time_ns = sched_clock();
1181 preempt_enable();
1184 static inline void set_io_start_time_ns(struct request *req)
1186 preempt_disable();
1187 req->io_start_time_ns = sched_clock();
1188 preempt_enable();
1191 static inline uint64_t rq_start_time_ns(struct request *req)
1193 return req->start_time_ns;
1196 static inline uint64_t rq_io_start_time_ns(struct request *req)
1198 return req->io_start_time_ns;
1200 #else
1201 static inline void set_start_time_ns(struct request *req) {}
1202 static inline void set_io_start_time_ns(struct request *req) {}
1203 static inline uint64_t rq_start_time_ns(struct request *req)
1205 return 0;
1207 static inline uint64_t rq_io_start_time_ns(struct request *req)
1209 return 0;
1211 #endif
1213 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1214 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1215 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1216 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1218 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1220 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1221 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1223 struct blk_integrity_exchg {
1224 void *prot_buf;
1225 void *data_buf;
1226 sector_t sector;
1227 unsigned int data_size;
1228 unsigned short sector_size;
1229 const char *disk_name;
1232 typedef void (integrity_gen_fn) (struct blk_integrity_exchg *);
1233 typedef int (integrity_vrfy_fn) (struct blk_integrity_exchg *);
1234 typedef void (integrity_set_tag_fn) (void *, void *, unsigned int);
1235 typedef void (integrity_get_tag_fn) (void *, void *, unsigned int);
1237 struct blk_integrity {
1238 integrity_gen_fn *generate_fn;
1239 integrity_vrfy_fn *verify_fn;
1240 integrity_set_tag_fn *set_tag_fn;
1241 integrity_get_tag_fn *get_tag_fn;
1243 unsigned short flags;
1244 unsigned short tuple_size;
1245 unsigned short sector_size;
1246 unsigned short tag_size;
1248 const char *name;
1250 struct kobject kobj;
1253 extern bool blk_integrity_is_initialized(struct gendisk *);
1254 extern int blk_integrity_register(struct gendisk *, struct blk_integrity *);
1255 extern void blk_integrity_unregister(struct gendisk *);
1256 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1257 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1258 struct scatterlist *);
1259 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1260 extern int blk_integrity_merge_rq(struct request_queue *, struct request *,
1261 struct request *);
1262 extern int blk_integrity_merge_bio(struct request_queue *, struct request *,
1263 struct bio *);
1265 static inline
1266 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1268 return bdev->bd_disk->integrity;
1271 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1273 return disk->integrity;
1276 static inline int blk_integrity_rq(struct request *rq)
1278 if (rq->bio == NULL)
1279 return 0;
1281 return bio_integrity(rq->bio);
1284 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1285 unsigned int segs)
1287 q->limits.max_integrity_segments = segs;
1290 static inline unsigned short
1291 queue_max_integrity_segments(struct request_queue *q)
1293 return q->limits.max_integrity_segments;
1296 #else /* CONFIG_BLK_DEV_INTEGRITY */
1298 struct bio;
1299 struct block_device;
1300 struct gendisk;
1301 struct blk_integrity;
1303 static inline int blk_integrity_rq(struct request *rq)
1305 return 0;
1307 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1308 struct bio *b)
1310 return 0;
1312 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1313 struct bio *b,
1314 struct scatterlist *s)
1316 return 0;
1318 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1320 return 0;
1322 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1324 return NULL;
1326 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1328 return 0;
1330 static inline int blk_integrity_register(struct gendisk *d,
1331 struct blk_integrity *b)
1333 return 0;
1335 static inline void blk_integrity_unregister(struct gendisk *d)
1338 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1339 unsigned int segs)
1342 static inline unsigned short queue_max_integrity_segments(struct request_queue *q)
1344 return 0;
1346 static inline int blk_integrity_merge_rq(struct request_queue *rq,
1347 struct request *r1,
1348 struct request *r2)
1350 return 0;
1352 static inline int blk_integrity_merge_bio(struct request_queue *rq,
1353 struct request *r,
1354 struct bio *b)
1356 return 0;
1358 static inline bool blk_integrity_is_initialized(struct gendisk *g)
1360 return 0;
1363 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1365 struct block_device_operations {
1366 int (*open) (struct block_device *, fmode_t);
1367 int (*release) (struct gendisk *, fmode_t);
1368 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1369 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1370 int (*direct_access) (struct block_device *, sector_t,
1371 void **, unsigned long *);
1372 unsigned int (*check_events) (struct gendisk *disk,
1373 unsigned int clearing);
1374 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1375 int (*media_changed) (struct gendisk *);
1376 void (*unlock_native_capacity) (struct gendisk *);
1377 int (*revalidate_disk) (struct gendisk *);
1378 int (*getgeo)(struct block_device *, struct hd_geometry *);
1379 /* this callback is with swap_lock and sometimes page table lock held */
1380 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1381 struct module *owner;
1384 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1385 unsigned long);
1386 #else /* CONFIG_BLOCK */
1388 * stubs for when the block layer is configured out
1390 #define buffer_heads_over_limit 0
1392 static inline long nr_blockdev_pages(void)
1394 return 0;
1397 struct blk_plug {
1400 static inline void blk_start_plug(struct blk_plug *plug)
1404 static inline void blk_finish_plug(struct blk_plug *plug)
1408 static inline void blk_flush_plug(struct task_struct *task)
1412 static inline void blk_schedule_flush_plug(struct task_struct *task)
1417 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1419 return false;
1422 #endif /* CONFIG_BLOCK */
1424 #endif