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
;
128 unsigned short ioprio
;
132 void *special
; /* opaque pointer available for LLD use */
133 char *buffer
; /* kaddr of the current segment if available */
139 * when request is used as a packet command carrier
141 unsigned char __cmd
[BLK_MAX_CDB
];
143 unsigned short cmd_len
;
145 unsigned int extra_len
; /* length of alignment and padding */
146 unsigned int sense_len
;
147 unsigned int resid_len
; /* residual count */
150 unsigned long deadline
;
151 struct list_head timeout_list
;
152 unsigned int timeout
;
156 * completion callback.
158 rq_end_io_fn
*end_io
;
162 struct request
*next_rq
;
165 static inline unsigned short req_get_ioprio(struct request
*req
)
171 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
172 * requests. Some step values could eventually be made generic.
174 struct request_pm_state
176 /* PM state machine step value, currently driver specific */
178 /* requested PM state value (S1, S2, S3, S4, ...) */
180 void* data
; /* for driver use */
183 #include <linux/elevator.h>
185 typedef void (request_fn_proc
) (struct request_queue
*q
);
186 typedef int (make_request_fn
) (struct request_queue
*q
, struct bio
*bio
);
187 typedef int (prep_rq_fn
) (struct request_queue
*, struct request
*);
188 typedef void (unprep_rq_fn
) (struct request_queue
*, struct request
*);
189 typedef void (unplug_fn
) (struct request_queue
*);
192 struct bvec_merge_data
{
193 struct block_device
*bi_bdev
;
198 typedef int (merge_bvec_fn
) (struct request_queue
*, struct bvec_merge_data
*,
200 typedef void (softirq_done_fn
)(struct request
*);
201 typedef int (dma_drain_needed_fn
)(struct request
*);
202 typedef int (lld_busy_fn
) (struct request_queue
*q
);
204 enum blk_eh_timer_return
{
210 typedef enum blk_eh_timer_return (rq_timed_out_fn
)(struct request
*);
212 enum blk_queue_state
{
217 struct blk_queue_tag
{
218 struct request
**tag_index
; /* map of busy tags */
219 unsigned long *tag_map
; /* bit map of free/busy tags */
220 int busy
; /* current depth */
221 int max_depth
; /* what we will send to device */
222 int real_max_depth
; /* what the array can hold */
223 atomic_t refcnt
; /* map can be shared */
226 #define BLK_SCSI_MAX_CMDS (256)
227 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
229 struct queue_limits
{
230 unsigned long bounce_pfn
;
231 unsigned long seg_boundary_mask
;
233 unsigned int max_hw_sectors
;
234 unsigned int max_sectors
;
235 unsigned int max_segment_size
;
236 unsigned int physical_block_size
;
237 unsigned int alignment_offset
;
240 unsigned int max_discard_sectors
;
241 unsigned int discard_granularity
;
242 unsigned int discard_alignment
;
244 unsigned short logical_block_size
;
245 unsigned short max_segments
;
247 unsigned char misaligned
;
248 unsigned char discard_misaligned
;
249 unsigned char no_cluster
;
250 signed char discard_zeroes_data
;
256 * Together with queue_head for cacheline sharing
258 struct list_head queue_head
;
259 struct request
*last_merge
;
260 struct elevator_queue
*elevator
;
263 * the queue request freelist, one for reads and one for writes
265 struct request_list rq
;
267 request_fn_proc
*request_fn
;
268 make_request_fn
*make_request_fn
;
269 prep_rq_fn
*prep_rq_fn
;
270 unprep_rq_fn
*unprep_rq_fn
;
271 unplug_fn
*unplug_fn
;
272 merge_bvec_fn
*merge_bvec_fn
;
273 softirq_done_fn
*softirq_done_fn
;
274 rq_timed_out_fn
*rq_timed_out_fn
;
275 dma_drain_needed_fn
*dma_drain_needed
;
276 lld_busy_fn
*lld_busy_fn
;
279 * Dispatch queue sorting
282 struct request
*boundary_rq
;
285 * Auto-unplugging state
287 struct timer_list unplug_timer
;
288 int unplug_thresh
; /* After this many requests */
289 unsigned long unplug_delay
; /* After this many jiffies */
290 struct work_struct unplug_work
;
292 struct backing_dev_info backing_dev_info
;
295 * The queue owner gets to use this for whatever they like.
296 * ll_rw_blk doesn't touch it.
301 * queue needs bounce pages for pages above this limit
306 * various queue flags, see QUEUE_* below
308 unsigned long queue_flags
;
311 * protects queue structures from reentrancy. ->__queue_lock should
312 * _never_ be used directly, it is queue private. always use
315 spinlock_t __queue_lock
;
316 spinlock_t
*queue_lock
;
326 unsigned long nr_requests
; /* Max # of requests */
327 unsigned int nr_congestion_on
;
328 unsigned int nr_congestion_off
;
329 unsigned int nr_batching
;
331 void *dma_drain_buffer
;
332 unsigned int dma_drain_size
;
333 unsigned int dma_pad_mask
;
334 unsigned int dma_alignment
;
336 struct blk_queue_tag
*queue_tags
;
337 struct list_head tag_busy_list
;
339 unsigned int nr_sorted
;
340 unsigned int in_flight
[2];
342 unsigned int rq_timeout
;
343 struct timer_list timeout
;
344 struct list_head timeout_list
;
346 struct queue_limits limits
;
351 unsigned int sg_timeout
;
352 unsigned int sg_reserved_size
;
354 #ifdef CONFIG_BLK_DEV_IO_TRACE
355 struct blk_trace
*blk_trace
;
358 * reserved for flush operations
360 unsigned int ordered
, next_ordered
, ordseq
;
361 int orderr
, ordcolor
;
362 struct request pre_flush_rq
, bar_rq
, post_flush_rq
;
363 struct request
*orig_bar_rq
;
365 struct mutex sysfs_lock
;
367 #if defined(CONFIG_BLK_DEV_BSG)
368 struct bsg_class_device bsg_dev
;
372 #define QUEUE_FLAG_CLUSTER 0 /* cluster several segments into 1 */
373 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
374 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
375 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
376 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
377 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
378 #define QUEUE_FLAG_REENTER 6 /* Re-entrancy avoidance */
379 #define QUEUE_FLAG_PLUGGED 7 /* queue is plugged */
380 #define QUEUE_FLAG_ELVSWITCH 8 /* don't use elevator, just do FIFO */
381 #define QUEUE_FLAG_BIDI 9 /* queue supports bidi requests */
382 #define QUEUE_FLAG_NOMERGES 10 /* disable merge attempts */
383 #define QUEUE_FLAG_SAME_COMP 11 /* force complete on same CPU */
384 #define QUEUE_FLAG_FAIL_IO 12 /* fake timeout */
385 #define QUEUE_FLAG_STACKABLE 13 /* supports request stacking */
386 #define QUEUE_FLAG_NONROT 14 /* non-rotational device (SSD) */
387 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
388 #define QUEUE_FLAG_IO_STAT 15 /* do IO stats */
389 #define QUEUE_FLAG_DISCARD 16 /* supports DISCARD */
390 #define QUEUE_FLAG_NOXMERGES 17 /* No extended merges */
391 #define QUEUE_FLAG_ADD_RANDOM 18 /* Contributes to random pool */
393 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
394 (1 << QUEUE_FLAG_CLUSTER) | \
395 (1 << QUEUE_FLAG_STACKABLE) | \
396 (1 << QUEUE_FLAG_SAME_COMP) | \
397 (1 << QUEUE_FLAG_ADD_RANDOM))
399 static inline int queue_is_locked(struct request_queue
*q
)
402 spinlock_t
*lock
= q
->queue_lock
;
403 return lock
&& spin_is_locked(lock
);
409 static inline void queue_flag_set_unlocked(unsigned int flag
,
410 struct request_queue
*q
)
412 __set_bit(flag
, &q
->queue_flags
);
415 static inline int queue_flag_test_and_clear(unsigned int flag
,
416 struct request_queue
*q
)
418 WARN_ON_ONCE(!queue_is_locked(q
));
420 if (test_bit(flag
, &q
->queue_flags
)) {
421 __clear_bit(flag
, &q
->queue_flags
);
428 static inline int queue_flag_test_and_set(unsigned int flag
,
429 struct request_queue
*q
)
431 WARN_ON_ONCE(!queue_is_locked(q
));
433 if (!test_bit(flag
, &q
->queue_flags
)) {
434 __set_bit(flag
, &q
->queue_flags
);
441 static inline void queue_flag_set(unsigned int flag
, struct request_queue
*q
)
443 WARN_ON_ONCE(!queue_is_locked(q
));
444 __set_bit(flag
, &q
->queue_flags
);
447 static inline void queue_flag_clear_unlocked(unsigned int flag
,
448 struct request_queue
*q
)
450 __clear_bit(flag
, &q
->queue_flags
);
453 static inline int queue_in_flight(struct request_queue
*q
)
455 return q
->in_flight
[0] + q
->in_flight
[1];
458 static inline void queue_flag_clear(unsigned int flag
, struct request_queue
*q
)
460 WARN_ON_ONCE(!queue_is_locked(q
));
461 __clear_bit(flag
, &q
->queue_flags
);
466 * Hardbarrier is supported with one of the following methods.
468 * NONE : hardbarrier unsupported
469 * DRAIN : ordering by draining is enough
470 * DRAIN_FLUSH : ordering by draining w/ pre and post flushes
471 * DRAIN_FUA : ordering by draining w/ pre flush and FUA write
472 * TAG : ordering by tag is enough
473 * TAG_FLUSH : ordering by tag w/ pre and post flushes
474 * TAG_FUA : ordering by tag w/ pre flush and FUA write
476 QUEUE_ORDERED_BY_DRAIN
= 0x01,
477 QUEUE_ORDERED_BY_TAG
= 0x02,
478 QUEUE_ORDERED_DO_PREFLUSH
= 0x10,
479 QUEUE_ORDERED_DO_BAR
= 0x20,
480 QUEUE_ORDERED_DO_POSTFLUSH
= 0x40,
481 QUEUE_ORDERED_DO_FUA
= 0x80,
483 QUEUE_ORDERED_NONE
= 0x00,
485 QUEUE_ORDERED_DRAIN
= QUEUE_ORDERED_BY_DRAIN
|
486 QUEUE_ORDERED_DO_BAR
,
487 QUEUE_ORDERED_DRAIN_FLUSH
= QUEUE_ORDERED_DRAIN
|
488 QUEUE_ORDERED_DO_PREFLUSH
|
489 QUEUE_ORDERED_DO_POSTFLUSH
,
490 QUEUE_ORDERED_DRAIN_FUA
= QUEUE_ORDERED_DRAIN
|
491 QUEUE_ORDERED_DO_PREFLUSH
|
492 QUEUE_ORDERED_DO_FUA
,
494 QUEUE_ORDERED_TAG
= QUEUE_ORDERED_BY_TAG
|
495 QUEUE_ORDERED_DO_BAR
,
496 QUEUE_ORDERED_TAG_FLUSH
= QUEUE_ORDERED_TAG
|
497 QUEUE_ORDERED_DO_PREFLUSH
|
498 QUEUE_ORDERED_DO_POSTFLUSH
,
499 QUEUE_ORDERED_TAG_FUA
= QUEUE_ORDERED_TAG
|
500 QUEUE_ORDERED_DO_PREFLUSH
|
501 QUEUE_ORDERED_DO_FUA
,
504 * Ordered operation sequence
506 QUEUE_ORDSEQ_STARTED
= 0x01, /* flushing in progress */
507 QUEUE_ORDSEQ_DRAIN
= 0x02, /* waiting for the queue to be drained */
508 QUEUE_ORDSEQ_PREFLUSH
= 0x04, /* pre-flushing in progress */
509 QUEUE_ORDSEQ_BAR
= 0x08, /* original barrier req in progress */
510 QUEUE_ORDSEQ_POSTFLUSH
= 0x10, /* post-flushing in progress */
511 QUEUE_ORDSEQ_DONE
= 0x20,
514 #define blk_queue_plugged(q) test_bit(QUEUE_FLAG_PLUGGED, &(q)->queue_flags)
515 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
516 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
517 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
518 #define blk_queue_noxmerges(q) \
519 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
520 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
521 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
522 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
523 #define blk_queue_flushing(q) ((q)->ordseq)
524 #define blk_queue_stackable(q) \
525 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
526 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
528 #define blk_noretry_request(rq) \
529 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
530 REQ_FAILFAST_DRIVER))
532 #define blk_account_rq(rq) \
533 (((rq)->cmd_flags & REQ_STARTED) && \
534 ((rq)->cmd_type == REQ_TYPE_FS || \
535 ((rq)->cmd_flags & REQ_DISCARD)))
537 #define blk_pm_request(rq) \
538 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
539 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
541 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
542 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
543 /* rq->queuelist of dequeued request must be list_empty() */
544 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
546 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
548 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
551 * We regard a request as sync, if either a read or a sync write
553 static inline bool rw_is_sync(unsigned int rw_flags
)
555 return !(rw_flags
& REQ_WRITE
) || (rw_flags
& REQ_SYNC
);
558 static inline bool rq_is_sync(struct request
*rq
)
560 return rw_is_sync(rq
->cmd_flags
);
563 static inline int blk_queue_full(struct request_queue
*q
, int sync
)
566 return test_bit(QUEUE_FLAG_SYNCFULL
, &q
->queue_flags
);
567 return test_bit(QUEUE_FLAG_ASYNCFULL
, &q
->queue_flags
);
570 static inline void blk_set_queue_full(struct request_queue
*q
, int sync
)
573 queue_flag_set(QUEUE_FLAG_SYNCFULL
, q
);
575 queue_flag_set(QUEUE_FLAG_ASYNCFULL
, q
);
578 static inline void blk_clear_queue_full(struct request_queue
*q
, int sync
)
581 queue_flag_clear(QUEUE_FLAG_SYNCFULL
, q
);
583 queue_flag_clear(QUEUE_FLAG_ASYNCFULL
, q
);
588 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
589 * it already be started by driver.
591 #define RQ_NOMERGE_FLAGS \
592 (REQ_NOMERGE | REQ_STARTED | REQ_HARDBARRIER | REQ_SOFTBARRIER)
593 #define rq_mergeable(rq) \
594 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
595 (((rq)->cmd_flags & REQ_DISCARD) || \
596 (rq)->cmd_type == REQ_TYPE_FS))
599 * q->prep_rq_fn return values
601 #define BLKPREP_OK 0 /* serve it */
602 #define BLKPREP_KILL 1 /* fatal error, kill */
603 #define BLKPREP_DEFER 2 /* leave on queue */
605 extern unsigned long blk_max_low_pfn
, blk_max_pfn
;
608 * standard bounce addresses:
610 * BLK_BOUNCE_HIGH : bounce all highmem pages
611 * BLK_BOUNCE_ANY : don't bounce anything
612 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
615 #if BITS_PER_LONG == 32
616 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
618 #define BLK_BOUNCE_HIGH -1ULL
620 #define BLK_BOUNCE_ANY (-1ULL)
621 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
624 * default timeout for SG_IO if none specified
626 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
627 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
630 extern int init_emergency_isa_pool(void);
631 extern void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
);
633 static inline int init_emergency_isa_pool(void)
637 static inline void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
)
640 #endif /* CONFIG_MMU */
646 unsigned long offset
;
651 struct req_iterator
{
656 /* This should not be used directly - use rq_for_each_segment */
657 #define for_each_bio(_bio) \
658 for (; _bio; _bio = _bio->bi_next)
659 #define __rq_for_each_bio(_bio, rq) \
661 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
663 #define rq_for_each_segment(bvl, _rq, _iter) \
664 __rq_for_each_bio(_iter.bio, _rq) \
665 bio_for_each_segment(bvl, _iter.bio, _iter.i)
667 #define rq_iter_last(rq, _iter) \
668 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
670 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
671 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
673 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
674 extern void rq_flush_dcache_pages(struct request
*rq
);
676 static inline void rq_flush_dcache_pages(struct request
*rq
)
681 extern int blk_register_queue(struct gendisk
*disk
);
682 extern void blk_unregister_queue(struct gendisk
*disk
);
683 extern void register_disk(struct gendisk
*dev
);
684 extern void generic_make_request(struct bio
*bio
);
685 extern void blk_rq_init(struct request_queue
*q
, struct request
*rq
);
686 extern void blk_put_request(struct request
*);
687 extern void __blk_put_request(struct request_queue
*, struct request
*);
688 extern struct request
*blk_get_request(struct request_queue
*, int, gfp_t
);
689 extern struct request
*blk_make_request(struct request_queue
*, struct bio
*,
691 extern void blk_insert_request(struct request_queue
*, struct request
*, int, void *);
692 extern void blk_requeue_request(struct request_queue
*, struct request
*);
693 extern void blk_add_request_payload(struct request
*rq
, struct page
*page
,
695 extern int blk_rq_check_limits(struct request_queue
*q
, struct request
*rq
);
696 extern int blk_lld_busy(struct request_queue
*q
);
697 extern int blk_rq_prep_clone(struct request
*rq
, struct request
*rq_src
,
698 struct bio_set
*bs
, gfp_t gfp_mask
,
699 int (*bio_ctr
)(struct bio
*, struct bio
*, void *),
701 extern void blk_rq_unprep_clone(struct request
*rq
);
702 extern int blk_insert_cloned_request(struct request_queue
*q
,
704 extern void blk_plug_device(struct request_queue
*);
705 extern void blk_plug_device_unlocked(struct request_queue
*);
706 extern int blk_remove_plug(struct request_queue
*);
707 extern void blk_recount_segments(struct request_queue
*, struct bio
*);
708 extern int scsi_cmd_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
709 unsigned int, void __user
*);
710 extern int sg_scsi_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
711 struct scsi_ioctl_command __user
*);
714 * A queue has just exitted congestion. Note this in the global counter of
715 * congested queues, and wake up anyone who was waiting for requests to be
718 static inline void blk_clear_queue_congested(struct request_queue
*q
, int sync
)
720 clear_bdi_congested(&q
->backing_dev_info
, sync
);
724 * A queue has just entered congestion. Flag that in the queue's VM-visible
725 * state flags and increment the global gounter of congested queues.
727 static inline void blk_set_queue_congested(struct request_queue
*q
, int sync
)
729 set_bdi_congested(&q
->backing_dev_info
, sync
);
732 extern void blk_start_queue(struct request_queue
*q
);
733 extern void blk_stop_queue(struct request_queue
*q
);
734 extern void blk_sync_queue(struct request_queue
*q
);
735 extern void __blk_stop_queue(struct request_queue
*q
);
736 extern void __blk_run_queue(struct request_queue
*);
737 extern void blk_run_queue(struct request_queue
*);
738 extern int blk_rq_map_user(struct request_queue
*, struct request
*,
739 struct rq_map_data
*, void __user
*, unsigned long,
741 extern int blk_rq_unmap_user(struct bio
*);
742 extern int blk_rq_map_kern(struct request_queue
*, struct request
*, void *, unsigned int, gfp_t
);
743 extern int blk_rq_map_user_iov(struct request_queue
*, struct request
*,
744 struct rq_map_data
*, struct sg_iovec
*, int,
745 unsigned int, gfp_t
);
746 extern int blk_execute_rq(struct request_queue
*, struct gendisk
*,
747 struct request
*, int);
748 extern void blk_execute_rq_nowait(struct request_queue
*, struct gendisk
*,
749 struct request
*, int, rq_end_io_fn
*);
750 extern void blk_unplug(struct request_queue
*q
);
752 static inline struct request_queue
*bdev_get_queue(struct block_device
*bdev
)
754 return bdev
->bd_disk
->queue
;
758 * blk_rq_pos() : the current sector
759 * blk_rq_bytes() : bytes left in the entire request
760 * blk_rq_cur_bytes() : bytes left in the current segment
761 * blk_rq_err_bytes() : bytes left till the next error boundary
762 * blk_rq_sectors() : sectors left in the entire request
763 * blk_rq_cur_sectors() : sectors left in the current segment
765 static inline sector_t
blk_rq_pos(const struct request
*rq
)
770 static inline unsigned int blk_rq_bytes(const struct request
*rq
)
772 return rq
->__data_len
;
775 static inline int blk_rq_cur_bytes(const struct request
*rq
)
777 return rq
->bio
? bio_cur_bytes(rq
->bio
) : 0;
780 extern unsigned int blk_rq_err_bytes(const struct request
*rq
);
782 static inline unsigned int blk_rq_sectors(const struct request
*rq
)
784 return blk_rq_bytes(rq
) >> 9;
787 static inline unsigned int blk_rq_cur_sectors(const struct request
*rq
)
789 return blk_rq_cur_bytes(rq
) >> 9;
793 * Request issue related functions.
795 extern struct request
*blk_peek_request(struct request_queue
*q
);
796 extern void blk_start_request(struct request
*rq
);
797 extern struct request
*blk_fetch_request(struct request_queue
*q
);
800 * Request completion related functions.
802 * blk_update_request() completes given number of bytes and updates
803 * the request without completing it.
805 * blk_end_request() and friends. __blk_end_request() must be called
806 * with the request queue spinlock acquired.
808 * Several drivers define their own end_request and call
809 * blk_end_request() for parts of the original function.
810 * This prevents code duplication in drivers.
812 extern bool blk_update_request(struct request
*rq
, int error
,
813 unsigned int nr_bytes
);
814 extern bool blk_end_request(struct request
*rq
, int error
,
815 unsigned int nr_bytes
);
816 extern void blk_end_request_all(struct request
*rq
, int error
);
817 extern bool blk_end_request_cur(struct request
*rq
, int error
);
818 extern bool blk_end_request_err(struct request
*rq
, int error
);
819 extern bool __blk_end_request(struct request
*rq
, int error
,
820 unsigned int nr_bytes
);
821 extern void __blk_end_request_all(struct request
*rq
, int error
);
822 extern bool __blk_end_request_cur(struct request
*rq
, int error
);
823 extern bool __blk_end_request_err(struct request
*rq
, int error
);
825 extern void blk_complete_request(struct request
*);
826 extern void __blk_complete_request(struct request
*);
827 extern void blk_abort_request(struct request
*);
828 extern void blk_abort_queue(struct request_queue
*);
829 extern void blk_unprep_request(struct request
*);
832 * Access functions for manipulating queue properties
834 extern struct request_queue
*blk_init_queue_node(request_fn_proc
*rfn
,
835 spinlock_t
*lock
, int node_id
);
836 extern struct request_queue
*blk_init_allocated_queue_node(struct request_queue
*,
838 spinlock_t
*, int node_id
);
839 extern struct request_queue
*blk_init_queue(request_fn_proc
*, spinlock_t
*);
840 extern struct request_queue
*blk_init_allocated_queue(struct request_queue
*,
841 request_fn_proc
*, spinlock_t
*);
842 extern void blk_cleanup_queue(struct request_queue
*);
843 extern void blk_queue_make_request(struct request_queue
*, make_request_fn
*);
844 extern void blk_queue_bounce_limit(struct request_queue
*, u64
);
845 extern void blk_queue_max_hw_sectors(struct request_queue
*, unsigned int);
846 extern void blk_queue_max_segments(struct request_queue
*, unsigned short);
847 extern void blk_queue_max_segment_size(struct request_queue
*, unsigned int);
848 extern void blk_queue_max_discard_sectors(struct request_queue
*q
,
849 unsigned int max_discard_sectors
);
850 extern void blk_queue_logical_block_size(struct request_queue
*, unsigned short);
851 extern void blk_queue_physical_block_size(struct request_queue
*, unsigned short);
852 extern void blk_queue_alignment_offset(struct request_queue
*q
,
853 unsigned int alignment
);
854 extern void blk_limits_io_min(struct queue_limits
*limits
, unsigned int min
);
855 extern void blk_queue_io_min(struct request_queue
*q
, unsigned int min
);
856 extern void blk_limits_io_opt(struct queue_limits
*limits
, unsigned int opt
);
857 extern void blk_queue_io_opt(struct request_queue
*q
, unsigned int opt
);
858 extern void blk_set_default_limits(struct queue_limits
*lim
);
859 extern int blk_stack_limits(struct queue_limits
*t
, struct queue_limits
*b
,
861 extern int bdev_stack_limits(struct queue_limits
*t
, struct block_device
*bdev
,
863 extern void disk_stack_limits(struct gendisk
*disk
, struct block_device
*bdev
,
865 extern void blk_queue_stack_limits(struct request_queue
*t
, struct request_queue
*b
);
866 extern void blk_queue_dma_pad(struct request_queue
*, unsigned int);
867 extern void blk_queue_update_dma_pad(struct request_queue
*, unsigned int);
868 extern int blk_queue_dma_drain(struct request_queue
*q
,
869 dma_drain_needed_fn
*dma_drain_needed
,
870 void *buf
, unsigned int size
);
871 extern void blk_queue_lld_busy(struct request_queue
*q
, lld_busy_fn
*fn
);
872 extern void blk_queue_segment_boundary(struct request_queue
*, unsigned long);
873 extern void blk_queue_prep_rq(struct request_queue
*, prep_rq_fn
*pfn
);
874 extern void blk_queue_unprep_rq(struct request_queue
*, unprep_rq_fn
*ufn
);
875 extern void blk_queue_merge_bvec(struct request_queue
*, merge_bvec_fn
*);
876 extern void blk_queue_dma_alignment(struct request_queue
*, int);
877 extern void blk_queue_update_dma_alignment(struct request_queue
*, int);
878 extern void blk_queue_softirq_done(struct request_queue
*, softirq_done_fn
*);
879 extern void blk_queue_rq_timed_out(struct request_queue
*, rq_timed_out_fn
*);
880 extern void blk_queue_rq_timeout(struct request_queue
*, unsigned int);
881 extern struct backing_dev_info
*blk_get_backing_dev_info(struct block_device
*bdev
);
882 extern int blk_queue_ordered(struct request_queue
*, unsigned);
883 extern bool blk_do_ordered(struct request_queue
*, struct request
**);
884 extern unsigned blk_ordered_cur_seq(struct request_queue
*);
885 extern unsigned blk_ordered_req_seq(struct request
*);
886 extern bool blk_ordered_complete_seq(struct request_queue
*, unsigned, int);
888 extern int blk_rq_map_sg(struct request_queue
*, struct request
*, struct scatterlist
*);
889 extern void blk_dump_rq_flags(struct request
*, char *);
890 extern void generic_unplug_device(struct request_queue
*);
891 extern long nr_blockdev_pages(void);
893 int blk_get_queue(struct request_queue
*);
894 struct request_queue
*blk_alloc_queue(gfp_t
);
895 struct request_queue
*blk_alloc_queue_node(gfp_t
, int);
896 extern void blk_put_queue(struct request_queue
*);
901 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
902 extern int blk_queue_start_tag(struct request_queue
*, struct request
*);
903 extern struct request
*blk_queue_find_tag(struct request_queue
*, int);
904 extern void blk_queue_end_tag(struct request_queue
*, struct request
*);
905 extern int blk_queue_init_tags(struct request_queue
*, int, struct blk_queue_tag
*);
906 extern void blk_queue_free_tags(struct request_queue
*);
907 extern int blk_queue_resize_tags(struct request_queue
*, int);
908 extern void blk_queue_invalidate_tags(struct request_queue
*);
909 extern struct blk_queue_tag
*blk_init_tags(int);
910 extern void blk_free_tags(struct blk_queue_tag
*);
912 static inline struct request
*blk_map_queue_find_tag(struct blk_queue_tag
*bqt
,
915 if (unlikely(bqt
== NULL
|| tag
>= bqt
->real_max_depth
))
917 return bqt
->tag_index
[tag
];
920 BLKDEV_WAIT
, /* wait for completion */
921 BLKDEV_BARRIER
, /*issue request with barrier */
923 #define BLKDEV_IFL_WAIT (1 << BLKDEV_WAIT)
924 #define BLKDEV_IFL_BARRIER (1 << BLKDEV_BARRIER)
925 extern int blkdev_issue_flush(struct block_device
*, gfp_t
, sector_t
*,
927 extern int blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
928 sector_t nr_sects
, gfp_t gfp_mask
, unsigned long flags
);
929 extern int blkdev_issue_zeroout(struct block_device
*bdev
, sector_t sector
,
930 sector_t nr_sects
, gfp_t gfp_mask
, unsigned long flags
);
931 static inline int sb_issue_discard(struct super_block
*sb
,
932 sector_t block
, sector_t nr_blocks
)
934 block
<<= (sb
->s_blocksize_bits
- 9);
935 nr_blocks
<<= (sb
->s_blocksize_bits
- 9);
936 return blkdev_issue_discard(sb
->s_bdev
, block
, nr_blocks
, GFP_NOFS
,
937 BLKDEV_IFL_WAIT
| BLKDEV_IFL_BARRIER
);
940 extern int blk_verify_command(unsigned char *cmd
, fmode_t has_write_perm
);
942 enum blk_default_limits
{
943 BLK_MAX_SEGMENTS
= 128,
944 BLK_SAFE_MAX_SECTORS
= 255,
945 BLK_DEF_MAX_SECTORS
= 1024,
946 BLK_MAX_SEGMENT_SIZE
= 65536,
947 BLK_SEG_BOUNDARY_MASK
= 0xFFFFFFFFUL
,
950 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
952 static inline unsigned long queue_bounce_pfn(struct request_queue
*q
)
954 return q
->limits
.bounce_pfn
;
957 static inline unsigned long queue_segment_boundary(struct request_queue
*q
)
959 return q
->limits
.seg_boundary_mask
;
962 static inline unsigned int queue_max_sectors(struct request_queue
*q
)
964 return q
->limits
.max_sectors
;
967 static inline unsigned int queue_max_hw_sectors(struct request_queue
*q
)
969 return q
->limits
.max_hw_sectors
;
972 static inline unsigned short queue_max_segments(struct request_queue
*q
)
974 return q
->limits
.max_segments
;
977 static inline unsigned int queue_max_segment_size(struct request_queue
*q
)
979 return q
->limits
.max_segment_size
;
982 static inline unsigned short queue_logical_block_size(struct request_queue
*q
)
986 if (q
&& q
->limits
.logical_block_size
)
987 retval
= q
->limits
.logical_block_size
;
992 static inline unsigned short bdev_logical_block_size(struct block_device
*bdev
)
994 return queue_logical_block_size(bdev_get_queue(bdev
));
997 static inline unsigned int queue_physical_block_size(struct request_queue
*q
)
999 return q
->limits
.physical_block_size
;
1002 static inline int bdev_physical_block_size(struct block_device
*bdev
)
1004 return queue_physical_block_size(bdev_get_queue(bdev
));
1007 static inline unsigned int queue_io_min(struct request_queue
*q
)
1009 return q
->limits
.io_min
;
1012 static inline int bdev_io_min(struct block_device
*bdev
)
1014 return queue_io_min(bdev_get_queue(bdev
));
1017 static inline unsigned int queue_io_opt(struct request_queue
*q
)
1019 return q
->limits
.io_opt
;
1022 static inline int bdev_io_opt(struct block_device
*bdev
)
1024 return queue_io_opt(bdev_get_queue(bdev
));
1027 static inline int queue_alignment_offset(struct request_queue
*q
)
1029 if (q
->limits
.misaligned
)
1032 return q
->limits
.alignment_offset
;
1035 static inline int queue_limit_alignment_offset(struct queue_limits
*lim
, sector_t sector
)
1037 unsigned int granularity
= max(lim
->physical_block_size
, lim
->io_min
);
1038 unsigned int alignment
= (sector
<< 9) & (granularity
- 1);
1040 return (granularity
+ lim
->alignment_offset
- alignment
)
1041 & (granularity
- 1);
1044 static inline int bdev_alignment_offset(struct block_device
*bdev
)
1046 struct request_queue
*q
= bdev_get_queue(bdev
);
1048 if (q
->limits
.misaligned
)
1051 if (bdev
!= bdev
->bd_contains
)
1052 return bdev
->bd_part
->alignment_offset
;
1054 return q
->limits
.alignment_offset
;
1057 static inline int queue_discard_alignment(struct request_queue
*q
)
1059 if (q
->limits
.discard_misaligned
)
1062 return q
->limits
.discard_alignment
;
1065 static inline int queue_limit_discard_alignment(struct queue_limits
*lim
, sector_t sector
)
1067 unsigned int alignment
= (sector
<< 9) & (lim
->discard_granularity
- 1);
1069 return (lim
->discard_granularity
+ lim
->discard_alignment
- alignment
)
1070 & (lim
->discard_granularity
- 1);
1073 static inline unsigned int queue_discard_zeroes_data(struct request_queue
*q
)
1075 if (q
->limits
.discard_zeroes_data
== 1)
1081 static inline unsigned int bdev_discard_zeroes_data(struct block_device
*bdev
)
1083 return queue_discard_zeroes_data(bdev_get_queue(bdev
));
1086 static inline int queue_dma_alignment(struct request_queue
*q
)
1088 return q
? q
->dma_alignment
: 511;
1091 static inline int blk_rq_aligned(struct request_queue
*q
, void *addr
,
1094 unsigned int alignment
= queue_dma_alignment(q
) | q
->dma_pad_mask
;
1095 return !((unsigned long)addr
& alignment
) && !(len
& alignment
);
1098 /* assumes size > 256 */
1099 static inline unsigned int blksize_bits(unsigned int size
)
1101 unsigned int bits
= 8;
1105 } while (size
> 256);
1109 static inline unsigned int block_size(struct block_device
*bdev
)
1111 return bdev
->bd_block_size
;
1114 typedef struct {struct page
*v
;} Sector
;
1116 unsigned char *read_dev_sector(struct block_device
*, sector_t
, Sector
*);
1118 static inline void put_dev_sector(Sector p
)
1120 page_cache_release(p
.v
);
1124 int kblockd_schedule_work(struct request_queue
*q
, struct work_struct
*work
);
1126 #ifdef CONFIG_BLK_CGROUP
1128 * This should not be using sched_clock(). A real patch is in progress
1129 * to fix this up, until that is in place we need to disable preemption
1130 * around sched_clock() in this function and set_io_start_time_ns().
1132 static inline void set_start_time_ns(struct request
*req
)
1135 req
->start_time_ns
= sched_clock();
1139 static inline void set_io_start_time_ns(struct request
*req
)
1142 req
->io_start_time_ns
= sched_clock();
1146 static inline uint64_t rq_start_time_ns(struct request
*req
)
1148 return req
->start_time_ns
;
1151 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1153 return req
->io_start_time_ns
;
1156 static inline void set_start_time_ns(struct request
*req
) {}
1157 static inline void set_io_start_time_ns(struct request
*req
) {}
1158 static inline uint64_t rq_start_time_ns(struct request
*req
)
1162 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1168 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1169 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1170 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1171 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1173 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1175 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1176 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1178 struct blk_integrity_exchg
{
1182 unsigned int data_size
;
1183 unsigned short sector_size
;
1184 const char *disk_name
;
1187 typedef void (integrity_gen_fn
) (struct blk_integrity_exchg
*);
1188 typedef int (integrity_vrfy_fn
) (struct blk_integrity_exchg
*);
1189 typedef void (integrity_set_tag_fn
) (void *, void *, unsigned int);
1190 typedef void (integrity_get_tag_fn
) (void *, void *, unsigned int);
1192 struct blk_integrity
{
1193 integrity_gen_fn
*generate_fn
;
1194 integrity_vrfy_fn
*verify_fn
;
1195 integrity_set_tag_fn
*set_tag_fn
;
1196 integrity_get_tag_fn
*get_tag_fn
;
1198 unsigned short flags
;
1199 unsigned short tuple_size
;
1200 unsigned short sector_size
;
1201 unsigned short tag_size
;
1205 struct kobject kobj
;
1208 extern int blk_integrity_register(struct gendisk
*, struct blk_integrity
*);
1209 extern void blk_integrity_unregister(struct gendisk
*);
1210 extern int blk_integrity_compare(struct gendisk
*, struct gendisk
*);
1211 extern int blk_rq_map_integrity_sg(struct request
*, struct scatterlist
*);
1212 extern int blk_rq_count_integrity_sg(struct request
*);
1215 struct blk_integrity
*bdev_get_integrity(struct block_device
*bdev
)
1217 return bdev
->bd_disk
->integrity
;
1220 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1222 return disk
->integrity
;
1225 static inline int blk_integrity_rq(struct request
*rq
)
1227 if (rq
->bio
== NULL
)
1230 return bio_integrity(rq
->bio
);
1233 #else /* CONFIG_BLK_DEV_INTEGRITY */
1235 #define blk_integrity_rq(rq) (0)
1236 #define blk_rq_count_integrity_sg(a) (0)
1237 #define blk_rq_map_integrity_sg(a, b) (0)
1238 #define bdev_get_integrity(a) (0)
1239 #define blk_get_integrity(a) (0)
1240 #define blk_integrity_compare(a, b) (0)
1241 #define blk_integrity_register(a, b) (0)
1242 #define blk_integrity_unregister(a) do { } while (0);
1244 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1246 struct block_device_operations
{
1247 int (*open
) (struct block_device
*, fmode_t
);
1248 int (*release
) (struct gendisk
*, fmode_t
);
1249 int (*ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1250 int (*compat_ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1251 int (*direct_access
) (struct block_device
*, sector_t
,
1252 void **, unsigned long *);
1253 int (*media_changed
) (struct gendisk
*);
1254 void (*unlock_native_capacity
) (struct gendisk
*);
1255 int (*revalidate_disk
) (struct gendisk
*);
1256 int (*getgeo
)(struct block_device
*, struct hd_geometry
*);
1257 /* this callback is with swap_lock and sometimes page table lock held */
1258 void (*swap_slot_free_notify
) (struct block_device
*, unsigned long);
1259 struct module
*owner
;
1262 extern int __blkdev_driver_ioctl(struct block_device
*, fmode_t
, unsigned int,
1264 #else /* CONFIG_BLOCK */
1266 * stubs for when the block layer is configured out
1268 #define buffer_heads_over_limit 0
1270 static inline long nr_blockdev_pages(void)
1275 #endif /* CONFIG_BLOCK */