1 #ifndef _LINUX_BLKDEV_H
2 #define _LINUX_BLKDEV_H
6 #include <linux/sched.h>
7 #include <linux/major.h>
8 #include <linux/genhd.h>
9 #include <linux/list.h>
10 #include <linux/timer.h>
11 #include <linux/workqueue.h>
12 #include <linux/pagemap.h>
13 #include <linux/backing-dev.h>
14 #include <linux/wait.h>
15 #include <linux/mempool.h>
16 #include <linux/bio.h>
17 #include <linux/module.h>
18 #include <linux/stringify.h>
19 #include <linux/gfp.h>
20 #include <linux/bsg.h>
21 #include <linux/smp.h>
23 #include <asm/scatterlist.h>
25 struct scsi_ioctl_command
;
28 struct elevator_queue
;
29 struct request_pm_state
;
34 #define BLKDEV_MIN_RQ 4
35 #define BLKDEV_MAX_RQ 128 /* Default maximum */
38 typedef void (rq_end_io_fn
)(struct request
*, int);
42 * count[], starved[], and wait[] are indexed by
43 * BLK_RW_SYNC/BLK_RW_ASYNC
49 wait_queue_head_t wait
[2];
53 * request command types
55 enum rq_cmd_type_bits
{
56 REQ_TYPE_FS
= 1, /* fs request */
57 REQ_TYPE_BLOCK_PC
, /* scsi command */
58 REQ_TYPE_SENSE
, /* sense request */
59 REQ_TYPE_PM_SUSPEND
, /* suspend request */
60 REQ_TYPE_PM_RESUME
, /* resume request */
61 REQ_TYPE_PM_SHUTDOWN
, /* shutdown request */
62 REQ_TYPE_SPECIAL
, /* driver defined type */
64 * for ATA/ATAPI devices. this really doesn't belong here, ide should
65 * use REQ_TYPE_SPECIAL and use rq->cmd[0] with the range of driver
66 * private REQ_LB opcodes to differentiate what type of request this is
68 REQ_TYPE_ATA_TASKFILE
,
72 #define BLK_MAX_CDB 16
75 * try to put the fields that are referenced together in the same cacheline.
76 * if you modify this structure, be sure to check block/blk-core.c:rq_init()
80 struct list_head queuelist
;
81 struct call_single_data csd
;
83 struct request_queue
*q
;
85 unsigned int cmd_flags
;
86 enum rq_cmd_type_bits cmd_type
;
87 unsigned long atomic_flags
;
91 /* the following two fields are internal, NEVER access directly */
92 unsigned int __data_len
; /* total data len */
93 sector_t __sector
; /* sector cursor */
98 struct hlist_node hash
; /* merge hash */
100 * The rb_node is only used inside the io scheduler, requests
101 * are pruned when moved to the dispatch queue. So let the
102 * completion_data share space with the rb_node.
105 struct rb_node rb_node
; /* sort/lookup */
106 void *completion_data
;
110 * Three pointers are available for the IO schedulers, if they need
111 * more they have to dynamically allocate it.
113 void *elevator_private
;
114 void *elevator_private2
;
115 void *elevator_private3
;
117 struct gendisk
*rq_disk
;
118 unsigned long start_time
;
119 #ifdef CONFIG_BLK_CGROUP
120 unsigned long long start_time_ns
;
121 unsigned long long io_start_time_ns
; /* when passed to hardware */
123 /* Number of scatter-gather DMA addr+len pairs after
124 * physical address coalescing is performed.
126 unsigned short nr_phys_segments
;
127 #if defined(CONFIG_BLK_DEV_INTEGRITY)
128 unsigned short nr_integrity_segments
;
131 unsigned short ioprio
;
135 void *special
; /* opaque pointer available for LLD use */
136 char *buffer
; /* kaddr of the current segment if available */
142 * when request is used as a packet command carrier
144 unsigned char __cmd
[BLK_MAX_CDB
];
146 unsigned short cmd_len
;
148 unsigned int extra_len
; /* length of alignment and padding */
149 unsigned int sense_len
;
150 unsigned int resid_len
; /* residual count */
153 unsigned long deadline
;
154 struct list_head timeout_list
;
155 unsigned int timeout
;
159 * completion callback.
161 rq_end_io_fn
*end_io
;
165 struct request
*next_rq
;
168 static inline unsigned short req_get_ioprio(struct request
*req
)
174 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
175 * requests. Some step values could eventually be made generic.
177 struct request_pm_state
179 /* PM state machine step value, currently driver specific */
181 /* requested PM state value (S1, S2, S3, S4, ...) */
183 void* data
; /* for driver use */
186 #include <linux/elevator.h>
188 typedef void (request_fn_proc
) (struct request_queue
*q
);
189 typedef int (make_request_fn
) (struct request_queue
*q
, struct bio
*bio
);
190 typedef int (prep_rq_fn
) (struct request_queue
*, struct request
*);
191 typedef void (unprep_rq_fn
) (struct request_queue
*, struct request
*);
192 typedef void (unplug_fn
) (struct request_queue
*);
195 struct bvec_merge_data
{
196 struct block_device
*bi_bdev
;
201 typedef int (merge_bvec_fn
) (struct request_queue
*, struct bvec_merge_data
*,
203 typedef void (softirq_done_fn
)(struct request
*);
204 typedef int (dma_drain_needed_fn
)(struct request
*);
205 typedef int (lld_busy_fn
) (struct request_queue
*q
);
207 enum blk_eh_timer_return
{
213 typedef enum blk_eh_timer_return (rq_timed_out_fn
)(struct request
*);
215 enum blk_queue_state
{
220 struct blk_queue_tag
{
221 struct request
**tag_index
; /* map of busy tags */
222 unsigned long *tag_map
; /* bit map of free/busy tags */
223 int busy
; /* current depth */
224 int max_depth
; /* what we will send to device */
225 int real_max_depth
; /* what the array can hold */
226 atomic_t refcnt
; /* map can be shared */
229 #define BLK_SCSI_MAX_CMDS (256)
230 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
232 struct queue_limits
{
233 unsigned long bounce_pfn
;
234 unsigned long seg_boundary_mask
;
236 unsigned int max_hw_sectors
;
237 unsigned int max_sectors
;
238 unsigned int max_segment_size
;
239 unsigned int physical_block_size
;
240 unsigned int alignment_offset
;
243 unsigned int max_discard_sectors
;
244 unsigned int discard_granularity
;
245 unsigned int discard_alignment
;
247 unsigned short logical_block_size
;
248 unsigned short max_segments
;
249 unsigned short max_integrity_segments
;
251 unsigned char misaligned
;
252 unsigned char discard_misaligned
;
253 unsigned char no_cluster
;
254 signed char discard_zeroes_data
;
260 * Together with queue_head for cacheline sharing
262 struct list_head queue_head
;
263 struct request
*last_merge
;
264 struct elevator_queue
*elevator
;
267 * the queue request freelist, one for reads and one for writes
269 struct request_list rq
;
271 request_fn_proc
*request_fn
;
272 make_request_fn
*make_request_fn
;
273 prep_rq_fn
*prep_rq_fn
;
274 unprep_rq_fn
*unprep_rq_fn
;
275 unplug_fn
*unplug_fn
;
276 merge_bvec_fn
*merge_bvec_fn
;
277 softirq_done_fn
*softirq_done_fn
;
278 rq_timed_out_fn
*rq_timed_out_fn
;
279 dma_drain_needed_fn
*dma_drain_needed
;
280 lld_busy_fn
*lld_busy_fn
;
283 * Dispatch queue sorting
286 struct request
*boundary_rq
;
289 * Auto-unplugging state
291 struct timer_list unplug_timer
;
292 int unplug_thresh
; /* After this many requests */
293 unsigned long unplug_delay
; /* After this many jiffies */
294 struct work_struct unplug_work
;
296 struct backing_dev_info backing_dev_info
;
299 * The queue owner gets to use this for whatever they like.
300 * ll_rw_blk doesn't touch it.
305 * queue needs bounce pages for pages above this limit
310 * various queue flags, see QUEUE_* below
312 unsigned long queue_flags
;
315 * protects queue structures from reentrancy. ->__queue_lock should
316 * _never_ be used directly, it is queue private. always use
319 spinlock_t __queue_lock
;
320 spinlock_t
*queue_lock
;
330 unsigned long nr_requests
; /* Max # of requests */
331 unsigned int nr_congestion_on
;
332 unsigned int nr_congestion_off
;
333 unsigned int nr_batching
;
335 void *dma_drain_buffer
;
336 unsigned int dma_drain_size
;
337 unsigned int dma_pad_mask
;
338 unsigned int dma_alignment
;
340 struct blk_queue_tag
*queue_tags
;
341 struct list_head tag_busy_list
;
343 unsigned int nr_sorted
;
344 unsigned int in_flight
[2];
346 unsigned int rq_timeout
;
347 struct timer_list timeout
;
348 struct list_head timeout_list
;
350 struct queue_limits limits
;
355 unsigned int sg_timeout
;
356 unsigned int sg_reserved_size
;
358 #ifdef CONFIG_BLK_DEV_IO_TRACE
359 struct blk_trace
*blk_trace
;
362 * reserved for flush operations
364 unsigned int ordered
, next_ordered
, ordseq
;
365 int orderr
, ordcolor
;
366 struct request pre_flush_rq
, bar_rq
, post_flush_rq
;
367 struct request
*orig_bar_rq
;
369 struct mutex sysfs_lock
;
371 #if defined(CONFIG_BLK_DEV_BSG)
372 struct bsg_class_device bsg_dev
;
376 #define QUEUE_FLAG_CLUSTER 0 /* cluster several segments into 1 */
377 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
378 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
379 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
380 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
381 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
382 #define QUEUE_FLAG_REENTER 6 /* Re-entrancy avoidance */
383 #define QUEUE_FLAG_PLUGGED 7 /* queue is plugged */
384 #define QUEUE_FLAG_ELVSWITCH 8 /* don't use elevator, just do FIFO */
385 #define QUEUE_FLAG_BIDI 9 /* queue supports bidi requests */
386 #define QUEUE_FLAG_NOMERGES 10 /* disable merge attempts */
387 #define QUEUE_FLAG_SAME_COMP 11 /* force complete on same CPU */
388 #define QUEUE_FLAG_FAIL_IO 12 /* fake timeout */
389 #define QUEUE_FLAG_STACKABLE 13 /* supports request stacking */
390 #define QUEUE_FLAG_NONROT 14 /* non-rotational device (SSD) */
391 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
392 #define QUEUE_FLAG_IO_STAT 15 /* do IO stats */
393 #define QUEUE_FLAG_DISCARD 16 /* supports DISCARD */
394 #define QUEUE_FLAG_NOXMERGES 17 /* No extended merges */
395 #define QUEUE_FLAG_ADD_RANDOM 18 /* Contributes to random pool */
396 #define QUEUE_FLAG_SECDISCARD 19 /* supports SECDISCARD */
398 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
399 (1 << QUEUE_FLAG_CLUSTER) | \
400 (1 << QUEUE_FLAG_STACKABLE) | \
401 (1 << QUEUE_FLAG_SAME_COMP) | \
402 (1 << QUEUE_FLAG_ADD_RANDOM))
404 static inline int queue_is_locked(struct request_queue
*q
)
407 spinlock_t
*lock
= q
->queue_lock
;
408 return lock
&& spin_is_locked(lock
);
414 static inline void queue_flag_set_unlocked(unsigned int flag
,
415 struct request_queue
*q
)
417 __set_bit(flag
, &q
->queue_flags
);
420 static inline int queue_flag_test_and_clear(unsigned int flag
,
421 struct request_queue
*q
)
423 WARN_ON_ONCE(!queue_is_locked(q
));
425 if (test_bit(flag
, &q
->queue_flags
)) {
426 __clear_bit(flag
, &q
->queue_flags
);
433 static inline int queue_flag_test_and_set(unsigned int flag
,
434 struct request_queue
*q
)
436 WARN_ON_ONCE(!queue_is_locked(q
));
438 if (!test_bit(flag
, &q
->queue_flags
)) {
439 __set_bit(flag
, &q
->queue_flags
);
446 static inline void queue_flag_set(unsigned int flag
, struct request_queue
*q
)
448 WARN_ON_ONCE(!queue_is_locked(q
));
449 __set_bit(flag
, &q
->queue_flags
);
452 static inline void queue_flag_clear_unlocked(unsigned int flag
,
453 struct request_queue
*q
)
455 __clear_bit(flag
, &q
->queue_flags
);
458 static inline int queue_in_flight(struct request_queue
*q
)
460 return q
->in_flight
[0] + q
->in_flight
[1];
463 static inline void queue_flag_clear(unsigned int flag
, struct request_queue
*q
)
465 WARN_ON_ONCE(!queue_is_locked(q
));
466 __clear_bit(flag
, &q
->queue_flags
);
471 * Hardbarrier is supported with one of the following methods.
473 * NONE : hardbarrier unsupported
474 * DRAIN : ordering by draining is enough
475 * DRAIN_FLUSH : ordering by draining w/ pre and post flushes
476 * DRAIN_FUA : ordering by draining w/ pre flush and FUA write
477 * TAG : ordering by tag is enough
478 * TAG_FLUSH : ordering by tag w/ pre and post flushes
479 * TAG_FUA : ordering by tag w/ pre flush and FUA write
481 QUEUE_ORDERED_BY_DRAIN
= 0x01,
482 QUEUE_ORDERED_BY_TAG
= 0x02,
483 QUEUE_ORDERED_DO_PREFLUSH
= 0x10,
484 QUEUE_ORDERED_DO_BAR
= 0x20,
485 QUEUE_ORDERED_DO_POSTFLUSH
= 0x40,
486 QUEUE_ORDERED_DO_FUA
= 0x80,
488 QUEUE_ORDERED_NONE
= 0x00,
490 QUEUE_ORDERED_DRAIN
= QUEUE_ORDERED_BY_DRAIN
|
491 QUEUE_ORDERED_DO_BAR
,
492 QUEUE_ORDERED_DRAIN_FLUSH
= QUEUE_ORDERED_DRAIN
|
493 QUEUE_ORDERED_DO_PREFLUSH
|
494 QUEUE_ORDERED_DO_POSTFLUSH
,
495 QUEUE_ORDERED_DRAIN_FUA
= QUEUE_ORDERED_DRAIN
|
496 QUEUE_ORDERED_DO_PREFLUSH
|
497 QUEUE_ORDERED_DO_FUA
,
499 QUEUE_ORDERED_TAG
= QUEUE_ORDERED_BY_TAG
|
500 QUEUE_ORDERED_DO_BAR
,
501 QUEUE_ORDERED_TAG_FLUSH
= QUEUE_ORDERED_TAG
|
502 QUEUE_ORDERED_DO_PREFLUSH
|
503 QUEUE_ORDERED_DO_POSTFLUSH
,
504 QUEUE_ORDERED_TAG_FUA
= QUEUE_ORDERED_TAG
|
505 QUEUE_ORDERED_DO_PREFLUSH
|
506 QUEUE_ORDERED_DO_FUA
,
509 * Ordered operation sequence
511 QUEUE_ORDSEQ_STARTED
= 0x01, /* flushing in progress */
512 QUEUE_ORDSEQ_DRAIN
= 0x02, /* waiting for the queue to be drained */
513 QUEUE_ORDSEQ_PREFLUSH
= 0x04, /* pre-flushing in progress */
514 QUEUE_ORDSEQ_BAR
= 0x08, /* original barrier req in progress */
515 QUEUE_ORDSEQ_POSTFLUSH
= 0x10, /* post-flushing in progress */
516 QUEUE_ORDSEQ_DONE
= 0x20,
519 #define blk_queue_plugged(q) test_bit(QUEUE_FLAG_PLUGGED, &(q)->queue_flags)
520 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
521 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
522 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
523 #define blk_queue_noxmerges(q) \
524 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
525 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
526 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
527 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
528 #define blk_queue_flushing(q) ((q)->ordseq)
529 #define blk_queue_stackable(q) \
530 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
531 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
532 #define blk_queue_secdiscard(q) (blk_queue_discard(q) && \
533 test_bit(QUEUE_FLAG_SECDISCARD, &(q)->queue_flags))
535 #define blk_noretry_request(rq) \
536 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
537 REQ_FAILFAST_DRIVER))
539 #define blk_account_rq(rq) \
540 (((rq)->cmd_flags & REQ_STARTED) && \
541 ((rq)->cmd_type == REQ_TYPE_FS || \
542 ((rq)->cmd_flags & REQ_DISCARD)))
544 #define blk_pm_request(rq) \
545 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
546 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
548 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
549 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
550 /* rq->queuelist of dequeued request must be list_empty() */
551 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
553 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
555 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
558 * We regard a request as sync, if either a read or a sync write
560 static inline bool rw_is_sync(unsigned int rw_flags
)
562 return !(rw_flags
& REQ_WRITE
) || (rw_flags
& REQ_SYNC
);
565 static inline bool rq_is_sync(struct request
*rq
)
567 return rw_is_sync(rq
->cmd_flags
);
570 static inline int blk_queue_full(struct request_queue
*q
, int sync
)
573 return test_bit(QUEUE_FLAG_SYNCFULL
, &q
->queue_flags
);
574 return test_bit(QUEUE_FLAG_ASYNCFULL
, &q
->queue_flags
);
577 static inline void blk_set_queue_full(struct request_queue
*q
, int sync
)
580 queue_flag_set(QUEUE_FLAG_SYNCFULL
, q
);
582 queue_flag_set(QUEUE_FLAG_ASYNCFULL
, q
);
585 static inline void blk_clear_queue_full(struct request_queue
*q
, int sync
)
588 queue_flag_clear(QUEUE_FLAG_SYNCFULL
, q
);
590 queue_flag_clear(QUEUE_FLAG_ASYNCFULL
, q
);
595 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
596 * it already be started by driver.
598 #define RQ_NOMERGE_FLAGS \
599 (REQ_NOMERGE | REQ_STARTED | REQ_HARDBARRIER | REQ_SOFTBARRIER)
600 #define rq_mergeable(rq) \
601 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
602 (((rq)->cmd_flags & REQ_DISCARD) || \
603 (rq)->cmd_type == REQ_TYPE_FS))
606 * q->prep_rq_fn return values
608 #define BLKPREP_OK 0 /* serve it */
609 #define BLKPREP_KILL 1 /* fatal error, kill */
610 #define BLKPREP_DEFER 2 /* leave on queue */
612 extern unsigned long blk_max_low_pfn
, blk_max_pfn
;
615 * standard bounce addresses:
617 * BLK_BOUNCE_HIGH : bounce all highmem pages
618 * BLK_BOUNCE_ANY : don't bounce anything
619 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
622 #if BITS_PER_LONG == 32
623 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
625 #define BLK_BOUNCE_HIGH -1ULL
627 #define BLK_BOUNCE_ANY (-1ULL)
628 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
631 * default timeout for SG_IO if none specified
633 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
634 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
637 extern int init_emergency_isa_pool(void);
638 extern void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
);
640 static inline int init_emergency_isa_pool(void)
644 static inline void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
)
647 #endif /* CONFIG_MMU */
653 unsigned long offset
;
658 struct req_iterator
{
663 /* This should not be used directly - use rq_for_each_segment */
664 #define for_each_bio(_bio) \
665 for (; _bio; _bio = _bio->bi_next)
666 #define __rq_for_each_bio(_bio, rq) \
668 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
670 #define rq_for_each_segment(bvl, _rq, _iter) \
671 __rq_for_each_bio(_iter.bio, _rq) \
672 bio_for_each_segment(bvl, _iter.bio, _iter.i)
674 #define rq_iter_last(rq, _iter) \
675 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
677 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
678 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
680 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
681 extern void rq_flush_dcache_pages(struct request
*rq
);
683 static inline void rq_flush_dcache_pages(struct request
*rq
)
688 extern int blk_register_queue(struct gendisk
*disk
);
689 extern void blk_unregister_queue(struct gendisk
*disk
);
690 extern void register_disk(struct gendisk
*dev
);
691 extern void generic_make_request(struct bio
*bio
);
692 extern void blk_rq_init(struct request_queue
*q
, struct request
*rq
);
693 extern void blk_put_request(struct request
*);
694 extern void __blk_put_request(struct request_queue
*, struct request
*);
695 extern struct request
*blk_get_request(struct request_queue
*, int, gfp_t
);
696 extern struct request
*blk_make_request(struct request_queue
*, struct bio
*,
698 extern void blk_insert_request(struct request_queue
*, struct request
*, int, void *);
699 extern void blk_requeue_request(struct request_queue
*, struct request
*);
700 extern void blk_add_request_payload(struct request
*rq
, struct page
*page
,
702 extern int blk_rq_check_limits(struct request_queue
*q
, struct request
*rq
);
703 extern int blk_lld_busy(struct request_queue
*q
);
704 extern int blk_rq_prep_clone(struct request
*rq
, struct request
*rq_src
,
705 struct bio_set
*bs
, gfp_t gfp_mask
,
706 int (*bio_ctr
)(struct bio
*, struct bio
*, void *),
708 extern void blk_rq_unprep_clone(struct request
*rq
);
709 extern int blk_insert_cloned_request(struct request_queue
*q
,
711 extern void blk_plug_device(struct request_queue
*);
712 extern void blk_plug_device_unlocked(struct request_queue
*);
713 extern int blk_remove_plug(struct request_queue
*);
714 extern void blk_recount_segments(struct request_queue
*, struct bio
*);
715 extern int scsi_cmd_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
716 unsigned int, void __user
*);
717 extern int sg_scsi_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
718 struct scsi_ioctl_command __user
*);
721 * A queue has just exitted congestion. Note this in the global counter of
722 * congested queues, and wake up anyone who was waiting for requests to be
725 static inline void blk_clear_queue_congested(struct request_queue
*q
, int sync
)
727 clear_bdi_congested(&q
->backing_dev_info
, sync
);
731 * A queue has just entered congestion. Flag that in the queue's VM-visible
732 * state flags and increment the global gounter of congested queues.
734 static inline void blk_set_queue_congested(struct request_queue
*q
, int sync
)
736 set_bdi_congested(&q
->backing_dev_info
, sync
);
739 extern void blk_start_queue(struct request_queue
*q
);
740 extern void blk_stop_queue(struct request_queue
*q
);
741 extern void blk_sync_queue(struct request_queue
*q
);
742 extern void __blk_stop_queue(struct request_queue
*q
);
743 extern void __blk_run_queue(struct request_queue
*);
744 extern void blk_run_queue(struct request_queue
*);
745 extern int blk_rq_map_user(struct request_queue
*, struct request
*,
746 struct rq_map_data
*, void __user
*, unsigned long,
748 extern int blk_rq_unmap_user(struct bio
*);
749 extern int blk_rq_map_kern(struct request_queue
*, struct request
*, void *, unsigned int, gfp_t
);
750 extern int blk_rq_map_user_iov(struct request_queue
*, struct request
*,
751 struct rq_map_data
*, struct sg_iovec
*, int,
752 unsigned int, gfp_t
);
753 extern int blk_execute_rq(struct request_queue
*, struct gendisk
*,
754 struct request
*, int);
755 extern void blk_execute_rq_nowait(struct request_queue
*, struct gendisk
*,
756 struct request
*, int, rq_end_io_fn
*);
757 extern void blk_unplug(struct request_queue
*q
);
759 static inline struct request_queue
*bdev_get_queue(struct block_device
*bdev
)
761 return bdev
->bd_disk
->queue
;
765 * blk_rq_pos() : the current sector
766 * blk_rq_bytes() : bytes left in the entire request
767 * blk_rq_cur_bytes() : bytes left in the current segment
768 * blk_rq_err_bytes() : bytes left till the next error boundary
769 * blk_rq_sectors() : sectors left in the entire request
770 * blk_rq_cur_sectors() : sectors left in the current segment
772 static inline sector_t
blk_rq_pos(const struct request
*rq
)
777 static inline unsigned int blk_rq_bytes(const struct request
*rq
)
779 return rq
->__data_len
;
782 static inline int blk_rq_cur_bytes(const struct request
*rq
)
784 return rq
->bio
? bio_cur_bytes(rq
->bio
) : 0;
787 extern unsigned int blk_rq_err_bytes(const struct request
*rq
);
789 static inline unsigned int blk_rq_sectors(const struct request
*rq
)
791 return blk_rq_bytes(rq
) >> 9;
794 static inline unsigned int blk_rq_cur_sectors(const struct request
*rq
)
796 return blk_rq_cur_bytes(rq
) >> 9;
800 * Request issue related functions.
802 extern struct request
*blk_peek_request(struct request_queue
*q
);
803 extern void blk_start_request(struct request
*rq
);
804 extern struct request
*blk_fetch_request(struct request_queue
*q
);
807 * Request completion related functions.
809 * blk_update_request() completes given number of bytes and updates
810 * the request without completing it.
812 * blk_end_request() and friends. __blk_end_request() must be called
813 * with the request queue spinlock acquired.
815 * Several drivers define their own end_request and call
816 * blk_end_request() for parts of the original function.
817 * This prevents code duplication in drivers.
819 extern bool blk_update_request(struct request
*rq
, int error
,
820 unsigned int nr_bytes
);
821 extern bool blk_end_request(struct request
*rq
, int error
,
822 unsigned int nr_bytes
);
823 extern void blk_end_request_all(struct request
*rq
, int error
);
824 extern bool blk_end_request_cur(struct request
*rq
, int error
);
825 extern bool blk_end_request_err(struct request
*rq
, int error
);
826 extern bool __blk_end_request(struct request
*rq
, int error
,
827 unsigned int nr_bytes
);
828 extern void __blk_end_request_all(struct request
*rq
, int error
);
829 extern bool __blk_end_request_cur(struct request
*rq
, int error
);
830 extern bool __blk_end_request_err(struct request
*rq
, int error
);
832 extern void blk_complete_request(struct request
*);
833 extern void __blk_complete_request(struct request
*);
834 extern void blk_abort_request(struct request
*);
835 extern void blk_abort_queue(struct request_queue
*);
836 extern void blk_unprep_request(struct request
*);
839 * Access functions for manipulating queue properties
841 extern struct request_queue
*blk_init_queue_node(request_fn_proc
*rfn
,
842 spinlock_t
*lock
, int node_id
);
843 extern struct request_queue
*blk_init_allocated_queue_node(struct request_queue
*,
845 spinlock_t
*, int node_id
);
846 extern struct request_queue
*blk_init_queue(request_fn_proc
*, spinlock_t
*);
847 extern struct request_queue
*blk_init_allocated_queue(struct request_queue
*,
848 request_fn_proc
*, spinlock_t
*);
849 extern void blk_cleanup_queue(struct request_queue
*);
850 extern void blk_queue_make_request(struct request_queue
*, make_request_fn
*);
851 extern void blk_queue_bounce_limit(struct request_queue
*, u64
);
852 extern void blk_queue_max_hw_sectors(struct request_queue
*, unsigned int);
853 extern void blk_queue_max_segments(struct request_queue
*, unsigned short);
854 extern void blk_queue_max_segment_size(struct request_queue
*, unsigned int);
855 extern void blk_queue_max_discard_sectors(struct request_queue
*q
,
856 unsigned int max_discard_sectors
);
857 extern void blk_queue_logical_block_size(struct request_queue
*, unsigned short);
858 extern void blk_queue_physical_block_size(struct request_queue
*, unsigned short);
859 extern void blk_queue_alignment_offset(struct request_queue
*q
,
860 unsigned int alignment
);
861 extern void blk_limits_io_min(struct queue_limits
*limits
, unsigned int min
);
862 extern void blk_queue_io_min(struct request_queue
*q
, unsigned int min
);
863 extern void blk_limits_io_opt(struct queue_limits
*limits
, unsigned int opt
);
864 extern void blk_queue_io_opt(struct request_queue
*q
, unsigned int opt
);
865 extern void blk_set_default_limits(struct queue_limits
*lim
);
866 extern int blk_stack_limits(struct queue_limits
*t
, struct queue_limits
*b
,
868 extern int bdev_stack_limits(struct queue_limits
*t
, struct block_device
*bdev
,
870 extern void disk_stack_limits(struct gendisk
*disk
, struct block_device
*bdev
,
872 extern void blk_queue_stack_limits(struct request_queue
*t
, struct request_queue
*b
);
873 extern void blk_queue_dma_pad(struct request_queue
*, unsigned int);
874 extern void blk_queue_update_dma_pad(struct request_queue
*, unsigned int);
875 extern int blk_queue_dma_drain(struct request_queue
*q
,
876 dma_drain_needed_fn
*dma_drain_needed
,
877 void *buf
, unsigned int size
);
878 extern void blk_queue_lld_busy(struct request_queue
*q
, lld_busy_fn
*fn
);
879 extern void blk_queue_segment_boundary(struct request_queue
*, unsigned long);
880 extern void blk_queue_prep_rq(struct request_queue
*, prep_rq_fn
*pfn
);
881 extern void blk_queue_unprep_rq(struct request_queue
*, unprep_rq_fn
*ufn
);
882 extern void blk_queue_merge_bvec(struct request_queue
*, merge_bvec_fn
*);
883 extern void blk_queue_dma_alignment(struct request_queue
*, int);
884 extern void blk_queue_update_dma_alignment(struct request_queue
*, int);
885 extern void blk_queue_softirq_done(struct request_queue
*, softirq_done_fn
*);
886 extern void blk_queue_rq_timed_out(struct request_queue
*, rq_timed_out_fn
*);
887 extern void blk_queue_rq_timeout(struct request_queue
*, unsigned int);
888 extern struct backing_dev_info
*blk_get_backing_dev_info(struct block_device
*bdev
);
889 extern int blk_queue_ordered(struct request_queue
*, unsigned);
890 extern bool blk_do_ordered(struct request_queue
*, struct request
**);
891 extern unsigned blk_ordered_cur_seq(struct request_queue
*);
892 extern unsigned blk_ordered_req_seq(struct request
*);
893 extern bool blk_ordered_complete_seq(struct request_queue
*, unsigned, int);
895 extern int blk_rq_map_sg(struct request_queue
*, struct request
*, struct scatterlist
*);
896 extern void blk_dump_rq_flags(struct request
*, char *);
897 extern void generic_unplug_device(struct request_queue
*);
898 extern long nr_blockdev_pages(void);
900 int blk_get_queue(struct request_queue
*);
901 struct request_queue
*blk_alloc_queue(gfp_t
);
902 struct request_queue
*blk_alloc_queue_node(gfp_t
, int);
903 extern void blk_put_queue(struct request_queue
*);
908 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
909 extern int blk_queue_start_tag(struct request_queue
*, struct request
*);
910 extern struct request
*blk_queue_find_tag(struct request_queue
*, int);
911 extern void blk_queue_end_tag(struct request_queue
*, struct request
*);
912 extern int blk_queue_init_tags(struct request_queue
*, int, struct blk_queue_tag
*);
913 extern void blk_queue_free_tags(struct request_queue
*);
914 extern int blk_queue_resize_tags(struct request_queue
*, int);
915 extern void blk_queue_invalidate_tags(struct request_queue
*);
916 extern struct blk_queue_tag
*blk_init_tags(int);
917 extern void blk_free_tags(struct blk_queue_tag
*);
919 static inline struct request
*blk_map_queue_find_tag(struct blk_queue_tag
*bqt
,
922 if (unlikely(bqt
== NULL
|| tag
>= bqt
->real_max_depth
))
924 return bqt
->tag_index
[tag
];
927 BLKDEV_WAIT
, /* wait for completion */
928 BLKDEV_BARRIER
, /* issue request with barrier */
929 BLKDEV_SECURE
, /* secure discard */
931 #define BLKDEV_IFL_WAIT (1 << BLKDEV_WAIT)
932 #define BLKDEV_IFL_BARRIER (1 << BLKDEV_BARRIER)
933 #define BLKDEV_IFL_SECURE (1 << BLKDEV_SECURE)
934 extern int blkdev_issue_flush(struct block_device
*, gfp_t
, sector_t
*,
936 extern int blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
937 sector_t nr_sects
, gfp_t gfp_mask
, unsigned long flags
);
938 extern int blkdev_issue_zeroout(struct block_device
*bdev
, sector_t sector
,
939 sector_t nr_sects
, gfp_t gfp_mask
, unsigned long flags
);
940 static inline int sb_issue_discard(struct super_block
*sb
,
941 sector_t block
, sector_t nr_blocks
)
943 block
<<= (sb
->s_blocksize_bits
- 9);
944 nr_blocks
<<= (sb
->s_blocksize_bits
- 9);
945 return blkdev_issue_discard(sb
->s_bdev
, block
, nr_blocks
, GFP_NOFS
,
946 BLKDEV_IFL_WAIT
| BLKDEV_IFL_BARRIER
);
949 extern int blk_verify_command(unsigned char *cmd
, fmode_t has_write_perm
);
951 enum blk_default_limits
{
952 BLK_MAX_SEGMENTS
= 128,
953 BLK_SAFE_MAX_SECTORS
= 255,
954 BLK_DEF_MAX_SECTORS
= 1024,
955 BLK_MAX_SEGMENT_SIZE
= 65536,
956 BLK_SEG_BOUNDARY_MASK
= 0xFFFFFFFFUL
,
959 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
961 static inline unsigned long queue_bounce_pfn(struct request_queue
*q
)
963 return q
->limits
.bounce_pfn
;
966 static inline unsigned long queue_segment_boundary(struct request_queue
*q
)
968 return q
->limits
.seg_boundary_mask
;
971 static inline unsigned int queue_max_sectors(struct request_queue
*q
)
973 return q
->limits
.max_sectors
;
976 static inline unsigned int queue_max_hw_sectors(struct request_queue
*q
)
978 return q
->limits
.max_hw_sectors
;
981 static inline unsigned short queue_max_segments(struct request_queue
*q
)
983 return q
->limits
.max_segments
;
986 static inline unsigned int queue_max_segment_size(struct request_queue
*q
)
988 return q
->limits
.max_segment_size
;
991 static inline unsigned short queue_logical_block_size(struct request_queue
*q
)
995 if (q
&& q
->limits
.logical_block_size
)
996 retval
= q
->limits
.logical_block_size
;
1001 static inline unsigned short bdev_logical_block_size(struct block_device
*bdev
)
1003 return queue_logical_block_size(bdev_get_queue(bdev
));
1006 static inline unsigned int queue_physical_block_size(struct request_queue
*q
)
1008 return q
->limits
.physical_block_size
;
1011 static inline int bdev_physical_block_size(struct block_device
*bdev
)
1013 return queue_physical_block_size(bdev_get_queue(bdev
));
1016 static inline unsigned int queue_io_min(struct request_queue
*q
)
1018 return q
->limits
.io_min
;
1021 static inline int bdev_io_min(struct block_device
*bdev
)
1023 return queue_io_min(bdev_get_queue(bdev
));
1026 static inline unsigned int queue_io_opt(struct request_queue
*q
)
1028 return q
->limits
.io_opt
;
1031 static inline int bdev_io_opt(struct block_device
*bdev
)
1033 return queue_io_opt(bdev_get_queue(bdev
));
1036 static inline int queue_alignment_offset(struct request_queue
*q
)
1038 if (q
->limits
.misaligned
)
1041 return q
->limits
.alignment_offset
;
1044 static inline int queue_limit_alignment_offset(struct queue_limits
*lim
, sector_t sector
)
1046 unsigned int granularity
= max(lim
->physical_block_size
, lim
->io_min
);
1047 unsigned int alignment
= (sector
<< 9) & (granularity
- 1);
1049 return (granularity
+ lim
->alignment_offset
- alignment
)
1050 & (granularity
- 1);
1053 static inline int bdev_alignment_offset(struct block_device
*bdev
)
1055 struct request_queue
*q
= bdev_get_queue(bdev
);
1057 if (q
->limits
.misaligned
)
1060 if (bdev
!= bdev
->bd_contains
)
1061 return bdev
->bd_part
->alignment_offset
;
1063 return q
->limits
.alignment_offset
;
1066 static inline int queue_discard_alignment(struct request_queue
*q
)
1068 if (q
->limits
.discard_misaligned
)
1071 return q
->limits
.discard_alignment
;
1074 static inline int queue_limit_discard_alignment(struct queue_limits
*lim
, sector_t sector
)
1076 unsigned int alignment
= (sector
<< 9) & (lim
->discard_granularity
- 1);
1078 return (lim
->discard_granularity
+ lim
->discard_alignment
- alignment
)
1079 & (lim
->discard_granularity
- 1);
1082 static inline unsigned int queue_discard_zeroes_data(struct request_queue
*q
)
1084 if (q
->limits
.discard_zeroes_data
== 1)
1090 static inline unsigned int bdev_discard_zeroes_data(struct block_device
*bdev
)
1092 return queue_discard_zeroes_data(bdev_get_queue(bdev
));
1095 static inline int queue_dma_alignment(struct request_queue
*q
)
1097 return q
? q
->dma_alignment
: 511;
1100 static inline int blk_rq_aligned(struct request_queue
*q
, void *addr
,
1103 unsigned int alignment
= queue_dma_alignment(q
) | q
->dma_pad_mask
;
1104 return !((unsigned long)addr
& alignment
) && !(len
& alignment
);
1107 /* assumes size > 256 */
1108 static inline unsigned int blksize_bits(unsigned int size
)
1110 unsigned int bits
= 8;
1114 } while (size
> 256);
1118 static inline unsigned int block_size(struct block_device
*bdev
)
1120 return bdev
->bd_block_size
;
1123 typedef struct {struct page
*v
;} Sector
;
1125 unsigned char *read_dev_sector(struct block_device
*, sector_t
, Sector
*);
1127 static inline void put_dev_sector(Sector p
)
1129 page_cache_release(p
.v
);
1133 int kblockd_schedule_work(struct request_queue
*q
, struct work_struct
*work
);
1135 #ifdef CONFIG_BLK_CGROUP
1137 * This should not be using sched_clock(). A real patch is in progress
1138 * to fix this up, until that is in place we need to disable preemption
1139 * around sched_clock() in this function and set_io_start_time_ns().
1141 static inline void set_start_time_ns(struct request
*req
)
1144 req
->start_time_ns
= sched_clock();
1148 static inline void set_io_start_time_ns(struct request
*req
)
1151 req
->io_start_time_ns
= sched_clock();
1155 static inline uint64_t rq_start_time_ns(struct request
*req
)
1157 return req
->start_time_ns
;
1160 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1162 return req
->io_start_time_ns
;
1165 static inline void set_start_time_ns(struct request
*req
) {}
1166 static inline void set_io_start_time_ns(struct request
*req
) {}
1167 static inline uint64_t rq_start_time_ns(struct request
*req
)
1171 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1177 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1178 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1179 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1180 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1182 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1184 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1185 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1187 struct blk_integrity_exchg
{
1191 unsigned int data_size
;
1192 unsigned short sector_size
;
1193 const char *disk_name
;
1196 typedef void (integrity_gen_fn
) (struct blk_integrity_exchg
*);
1197 typedef int (integrity_vrfy_fn
) (struct blk_integrity_exchg
*);
1198 typedef void (integrity_set_tag_fn
) (void *, void *, unsigned int);
1199 typedef void (integrity_get_tag_fn
) (void *, void *, unsigned int);
1201 struct blk_integrity
{
1202 integrity_gen_fn
*generate_fn
;
1203 integrity_vrfy_fn
*verify_fn
;
1204 integrity_set_tag_fn
*set_tag_fn
;
1205 integrity_get_tag_fn
*get_tag_fn
;
1207 unsigned short flags
;
1208 unsigned short tuple_size
;
1209 unsigned short sector_size
;
1210 unsigned short tag_size
;
1214 struct kobject kobj
;
1217 extern int blk_integrity_register(struct gendisk
*, struct blk_integrity
*);
1218 extern void blk_integrity_unregister(struct gendisk
*);
1219 extern int blk_integrity_compare(struct gendisk
*, struct gendisk
*);
1220 extern int blk_rq_map_integrity_sg(struct request_queue
*, struct bio
*,
1221 struct scatterlist
*);
1222 extern int blk_rq_count_integrity_sg(struct request_queue
*, struct bio
*);
1223 extern int blk_integrity_merge_rq(struct request_queue
*, struct request
*,
1225 extern int blk_integrity_merge_bio(struct request_queue
*, struct request
*,
1229 struct blk_integrity
*bdev_get_integrity(struct block_device
*bdev
)
1231 return bdev
->bd_disk
->integrity
;
1234 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1236 return disk
->integrity
;
1239 static inline int blk_integrity_rq(struct request
*rq
)
1241 if (rq
->bio
== NULL
)
1244 return bio_integrity(rq
->bio
);
1247 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1250 q
->limits
.max_integrity_segments
= segs
;
1253 static inline unsigned short
1254 queue_max_integrity_segments(struct request_queue
*q
)
1256 return q
->limits
.max_integrity_segments
;
1259 #else /* CONFIG_BLK_DEV_INTEGRITY */
1261 #define blk_integrity_rq(rq) (0)
1262 #define blk_rq_count_integrity_sg(a, b) (0)
1263 #define blk_rq_map_integrity_sg(a, b, c) (0)
1264 #define bdev_get_integrity(a) (0)
1265 #define blk_get_integrity(a) (0)
1266 #define blk_integrity_compare(a, b) (0)
1267 #define blk_integrity_register(a, b) (0)
1268 #define blk_integrity_unregister(a) do { } while (0);
1269 #define blk_queue_max_integrity_segments(a, b) do { } while (0);
1270 #define queue_max_integrity_segments(a) (0)
1271 #define blk_integrity_merge_rq(a, b, c) (0)
1272 #define blk_integrity_merge_bio(a, b, c) (0)
1274 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1276 struct block_device_operations
{
1277 int (*open
) (struct block_device
*, fmode_t
);
1278 int (*release
) (struct gendisk
*, fmode_t
);
1279 int (*ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1280 int (*compat_ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1281 int (*direct_access
) (struct block_device
*, sector_t
,
1282 void **, unsigned long *);
1283 int (*media_changed
) (struct gendisk
*);
1284 void (*unlock_native_capacity
) (struct gendisk
*);
1285 int (*revalidate_disk
) (struct gendisk
*);
1286 int (*getgeo
)(struct block_device
*, struct hd_geometry
*);
1287 /* this callback is with swap_lock and sometimes page table lock held */
1288 void (*swap_slot_free_notify
) (struct block_device
*, unsigned long);
1289 struct module
*owner
;
1292 extern int __blkdev_driver_ioctl(struct block_device
*, fmode_t
, unsigned int,
1294 #else /* CONFIG_BLOCK */
1296 * stubs for when the block layer is configured out
1298 #define buffer_heads_over_limit 0
1300 static inline long nr_blockdev_pages(void)
1305 #endif /* CONFIG_BLOCK */