1 #ifndef _LINUX_BLKDEV_H
2 #define _LINUX_BLKDEV_H
4 #include <linux/sched.h>
8 #include <linux/major.h>
9 #include <linux/genhd.h>
10 #include <linux/list.h>
11 #include <linux/timer.h>
12 #include <linux/workqueue.h>
13 #include <linux/pagemap.h>
14 #include <linux/backing-dev.h>
15 #include <linux/wait.h>
16 #include <linux/mempool.h>
17 #include <linux/bio.h>
18 #include <linux/stringify.h>
19 #include <linux/gfp.h>
20 #include <linux/bsg.h>
21 #include <linux/smp.h>
23 #include <asm/scatterlist.h>
26 struct scsi_ioctl_command
;
29 struct elevator_queue
;
30 struct request_pm_state
;
37 #define BLKDEV_MIN_RQ 4
38 #define BLKDEV_MAX_RQ 128 /* Default maximum */
41 * Maximum number of blkcg policies allowed to be registered concurrently.
42 * Defined here to simplify include dependency.
44 #define BLKCG_MAX_POLS 2
47 typedef void (rq_end_io_fn
)(struct request
*, int);
51 * count[], starved[], and wait[] are indexed by
52 * BLK_RW_SYNC/BLK_RW_ASYNC
58 wait_queue_head_t wait
[2];
62 * request command types
64 enum rq_cmd_type_bits
{
65 REQ_TYPE_FS
= 1, /* fs request */
66 REQ_TYPE_BLOCK_PC
, /* scsi command */
67 REQ_TYPE_SENSE
, /* sense request */
68 REQ_TYPE_PM_SUSPEND
, /* suspend request */
69 REQ_TYPE_PM_RESUME
, /* resume request */
70 REQ_TYPE_PM_SHUTDOWN
, /* shutdown request */
71 REQ_TYPE_SPECIAL
, /* driver defined type */
73 * for ATA/ATAPI devices. this really doesn't belong here, ide should
74 * use REQ_TYPE_SPECIAL and use rq->cmd[0] with the range of driver
75 * private REQ_LB opcodes to differentiate what type of request this is
77 REQ_TYPE_ATA_TASKFILE
,
81 #define BLK_MAX_CDB 16
84 * try to put the fields that are referenced together in the same cacheline.
85 * if you modify this structure, be sure to check block/blk-core.c:blk_rq_init()
89 struct list_head queuelist
;
90 struct call_single_data csd
;
92 struct request_queue
*q
;
94 unsigned int cmd_flags
;
95 enum rq_cmd_type_bits cmd_type
;
96 unsigned long atomic_flags
;
100 /* the following two fields are internal, NEVER access directly */
101 unsigned int __data_len
; /* total data len */
102 sector_t __sector
; /* sector cursor */
107 struct hlist_node hash
; /* merge hash */
109 * The rb_node is only used inside the io scheduler, requests
110 * are pruned when moved to the dispatch queue. So let the
111 * completion_data share space with the rb_node.
114 struct rb_node rb_node
; /* sort/lookup */
115 void *completion_data
;
119 * Three pointers are available for the IO schedulers, if they need
120 * more they have to dynamically allocate it. Flush requests are
121 * never put on the IO scheduler. So let the flush fields share
122 * space with the elevator data.
132 struct list_head list
;
133 rq_end_io_fn
*saved_end_io
;
137 struct gendisk
*rq_disk
;
138 struct hd_struct
*part
;
139 unsigned long start_time
;
140 #ifdef CONFIG_BLK_CGROUP
141 unsigned long long start_time_ns
;
142 unsigned long long io_start_time_ns
; /* when passed to hardware */
144 /* Number of scatter-gather DMA addr+len pairs after
145 * physical address coalescing is performed.
147 unsigned short nr_phys_segments
;
148 #if defined(CONFIG_BLK_DEV_INTEGRITY)
149 unsigned short nr_integrity_segments
;
152 unsigned short ioprio
;
156 void *special
; /* opaque pointer available for LLD use */
157 char *buffer
; /* kaddr of the current segment if available */
163 * when request is used as a packet command carrier
165 unsigned char __cmd
[BLK_MAX_CDB
];
167 unsigned short cmd_len
;
169 unsigned int extra_len
; /* length of alignment and padding */
170 unsigned int sense_len
;
171 unsigned int resid_len
; /* residual count */
174 unsigned long deadline
;
175 struct list_head timeout_list
;
176 unsigned int timeout
;
180 * completion callback.
182 rq_end_io_fn
*end_io
;
186 struct request
*next_rq
;
189 static inline unsigned short req_get_ioprio(struct request
*req
)
195 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
196 * requests. Some step values could eventually be made generic.
198 struct request_pm_state
200 /* PM state machine step value, currently driver specific */
202 /* requested PM state value (S1, S2, S3, S4, ...) */
204 void* data
; /* for driver use */
207 #include <linux/elevator.h>
209 typedef void (request_fn_proc
) (struct request_queue
*q
);
210 typedef void (make_request_fn
) (struct request_queue
*q
, struct bio
*bio
);
211 typedef int (prep_rq_fn
) (struct request_queue
*, struct request
*);
212 typedef void (unprep_rq_fn
) (struct request_queue
*, struct request
*);
215 struct bvec_merge_data
{
216 struct block_device
*bi_bdev
;
221 typedef int (merge_bvec_fn
) (struct request_queue
*, struct bvec_merge_data
*,
223 typedef void (softirq_done_fn
)(struct request
*);
224 typedef int (dma_drain_needed_fn
)(struct request
*);
225 typedef int (lld_busy_fn
) (struct request_queue
*q
);
226 typedef int (bsg_job_fn
) (struct bsg_job
*);
228 enum blk_eh_timer_return
{
234 typedef enum blk_eh_timer_return (rq_timed_out_fn
)(struct request
*);
236 enum blk_queue_state
{
241 struct blk_queue_tag
{
242 struct request
**tag_index
; /* map of busy tags */
243 unsigned long *tag_map
; /* bit map of free/busy tags */
244 int busy
; /* current depth */
245 int max_depth
; /* what we will send to device */
246 int real_max_depth
; /* what the array can hold */
247 atomic_t refcnt
; /* map can be shared */
250 #define BLK_SCSI_MAX_CMDS (256)
251 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
253 struct queue_limits
{
254 unsigned long bounce_pfn
;
255 unsigned long seg_boundary_mask
;
257 unsigned int max_hw_sectors
;
258 unsigned int max_sectors
;
259 unsigned int max_segment_size
;
260 unsigned int physical_block_size
;
261 unsigned int alignment_offset
;
264 unsigned int max_discard_sectors
;
265 unsigned int discard_granularity
;
266 unsigned int discard_alignment
;
268 unsigned short logical_block_size
;
269 unsigned short max_segments
;
270 unsigned short max_integrity_segments
;
272 unsigned char misaligned
;
273 unsigned char discard_misaligned
;
274 unsigned char cluster
;
275 unsigned char discard_zeroes_data
;
278 struct request_queue
{
280 * Together with queue_head for cacheline sharing
282 struct list_head queue_head
;
283 struct request
*last_merge
;
284 struct elevator_queue
*elevator
;
287 * the queue request freelist, one for reads and one for writes
289 struct request_list rq
;
291 request_fn_proc
*request_fn
;
292 make_request_fn
*make_request_fn
;
293 prep_rq_fn
*prep_rq_fn
;
294 unprep_rq_fn
*unprep_rq_fn
;
295 merge_bvec_fn
*merge_bvec_fn
;
296 softirq_done_fn
*softirq_done_fn
;
297 rq_timed_out_fn
*rq_timed_out_fn
;
298 dma_drain_needed_fn
*dma_drain_needed
;
299 lld_busy_fn
*lld_busy_fn
;
302 * Dispatch queue sorting
305 struct request
*boundary_rq
;
308 * Delayed queue handling
310 struct delayed_work delay_work
;
312 struct backing_dev_info backing_dev_info
;
315 * The queue owner gets to use this for whatever they like.
316 * ll_rw_blk doesn't touch it.
321 * various queue flags, see QUEUE_* below
323 unsigned long queue_flags
;
326 * ida allocated id for this queue. Used to index queues from
332 * queue needs bounce pages for pages above this limit
337 * protects queue structures from reentrancy. ->__queue_lock should
338 * _never_ be used directly, it is queue private. always use
341 spinlock_t __queue_lock
;
342 spinlock_t
*queue_lock
;
352 unsigned long nr_requests
; /* Max # of requests */
353 unsigned int nr_congestion_on
;
354 unsigned int nr_congestion_off
;
355 unsigned int nr_batching
;
357 unsigned int dma_drain_size
;
358 void *dma_drain_buffer
;
359 unsigned int dma_pad_mask
;
360 unsigned int dma_alignment
;
362 struct blk_queue_tag
*queue_tags
;
363 struct list_head tag_busy_list
;
365 unsigned int nr_sorted
;
366 unsigned int in_flight
[2];
368 unsigned int rq_timeout
;
369 struct timer_list timeout
;
370 struct list_head timeout_list
;
372 struct list_head icq_list
;
373 #ifdef CONFIG_BLK_CGROUP
374 DECLARE_BITMAP (blkcg_pols
, BLKCG_MAX_POLS
);
375 struct blkcg_gq
*root_blkg
;
376 struct list_head blkg_list
;
379 struct queue_limits limits
;
384 unsigned int sg_timeout
;
385 unsigned int sg_reserved_size
;
387 #ifdef CONFIG_BLK_DEV_IO_TRACE
388 struct blk_trace
*blk_trace
;
391 * for flush operations
393 unsigned int flush_flags
;
394 unsigned int flush_not_queueable
:1;
395 unsigned int flush_queue_delayed
:1;
396 unsigned int flush_pending_idx
:1;
397 unsigned int flush_running_idx
:1;
398 unsigned long flush_pending_since
;
399 struct list_head flush_queue
[2];
400 struct list_head flush_data_in_flight
;
401 struct request flush_rq
;
403 struct mutex sysfs_lock
;
407 #if defined(CONFIG_BLK_DEV_BSG)
408 bsg_job_fn
*bsg_job_fn
;
410 struct bsg_class_device bsg_dev
;
413 #ifdef CONFIG_BLK_CGROUP
414 struct list_head all_q_node
;
416 #ifdef CONFIG_BLK_DEV_THROTTLING
418 struct throtl_data
*td
;
422 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
423 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
424 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
425 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
426 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
427 #define QUEUE_FLAG_BYPASS 6 /* act as dumb FIFO queue */
428 #define QUEUE_FLAG_BIDI 7 /* queue supports bidi requests */
429 #define QUEUE_FLAG_NOMERGES 8 /* disable merge attempts */
430 #define QUEUE_FLAG_SAME_COMP 9 /* complete on same CPU-group */
431 #define QUEUE_FLAG_FAIL_IO 10 /* fake timeout */
432 #define QUEUE_FLAG_STACKABLE 11 /* supports request stacking */
433 #define QUEUE_FLAG_NONROT 12 /* non-rotational device (SSD) */
434 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
435 #define QUEUE_FLAG_IO_STAT 13 /* do IO stats */
436 #define QUEUE_FLAG_DISCARD 14 /* supports DISCARD */
437 #define QUEUE_FLAG_NOXMERGES 15 /* No extended merges */
438 #define QUEUE_FLAG_ADD_RANDOM 16 /* Contributes to random pool */
439 #define QUEUE_FLAG_SECDISCARD 17 /* supports SECDISCARD */
440 #define QUEUE_FLAG_SAME_FORCE 18 /* force complete on same CPU */
442 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
443 (1 << QUEUE_FLAG_STACKABLE) | \
444 (1 << QUEUE_FLAG_SAME_COMP) | \
445 (1 << QUEUE_FLAG_ADD_RANDOM))
447 static inline void queue_lockdep_assert_held(struct request_queue
*q
)
450 lockdep_assert_held(q
->queue_lock
);
453 static inline void queue_flag_set_unlocked(unsigned int flag
,
454 struct request_queue
*q
)
456 __set_bit(flag
, &q
->queue_flags
);
459 static inline int queue_flag_test_and_clear(unsigned int flag
,
460 struct request_queue
*q
)
462 queue_lockdep_assert_held(q
);
464 if (test_bit(flag
, &q
->queue_flags
)) {
465 __clear_bit(flag
, &q
->queue_flags
);
472 static inline int queue_flag_test_and_set(unsigned int flag
,
473 struct request_queue
*q
)
475 queue_lockdep_assert_held(q
);
477 if (!test_bit(flag
, &q
->queue_flags
)) {
478 __set_bit(flag
, &q
->queue_flags
);
485 static inline void queue_flag_set(unsigned int flag
, struct request_queue
*q
)
487 queue_lockdep_assert_held(q
);
488 __set_bit(flag
, &q
->queue_flags
);
491 static inline void queue_flag_clear_unlocked(unsigned int flag
,
492 struct request_queue
*q
)
494 __clear_bit(flag
, &q
->queue_flags
);
497 static inline int queue_in_flight(struct request_queue
*q
)
499 return q
->in_flight
[0] + q
->in_flight
[1];
502 static inline void queue_flag_clear(unsigned int flag
, struct request_queue
*q
)
504 queue_lockdep_assert_held(q
);
505 __clear_bit(flag
, &q
->queue_flags
);
508 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
509 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
510 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
511 #define blk_queue_bypass(q) test_bit(QUEUE_FLAG_BYPASS, &(q)->queue_flags)
512 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
513 #define blk_queue_noxmerges(q) \
514 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
515 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
516 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
517 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
518 #define blk_queue_stackable(q) \
519 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
520 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
521 #define blk_queue_secdiscard(q) (blk_queue_discard(q) && \
522 test_bit(QUEUE_FLAG_SECDISCARD, &(q)->queue_flags))
524 #define blk_noretry_request(rq) \
525 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
526 REQ_FAILFAST_DRIVER))
528 #define blk_account_rq(rq) \
529 (((rq)->cmd_flags & REQ_STARTED) && \
530 ((rq)->cmd_type == REQ_TYPE_FS || \
531 ((rq)->cmd_flags & REQ_DISCARD)))
533 #define blk_pm_request(rq) \
534 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
535 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
537 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
538 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
539 /* rq->queuelist of dequeued request must be list_empty() */
540 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
542 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
544 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
546 static inline unsigned int blk_queue_cluster(struct request_queue
*q
)
548 return q
->limits
.cluster
;
552 * We regard a request as sync, if either a read or a sync write
554 static inline bool rw_is_sync(unsigned int rw_flags
)
556 return !(rw_flags
& REQ_WRITE
) || (rw_flags
& REQ_SYNC
);
559 static inline bool rq_is_sync(struct request
*rq
)
561 return rw_is_sync(rq
->cmd_flags
);
564 static inline int blk_queue_full(struct request_queue
*q
, int sync
)
567 return test_bit(QUEUE_FLAG_SYNCFULL
, &q
->queue_flags
);
568 return test_bit(QUEUE_FLAG_ASYNCFULL
, &q
->queue_flags
);
571 static inline void blk_set_queue_full(struct request_queue
*q
, int sync
)
574 queue_flag_set(QUEUE_FLAG_SYNCFULL
, q
);
576 queue_flag_set(QUEUE_FLAG_ASYNCFULL
, q
);
579 static inline void blk_clear_queue_full(struct request_queue
*q
, int sync
)
582 queue_flag_clear(QUEUE_FLAG_SYNCFULL
, q
);
584 queue_flag_clear(QUEUE_FLAG_ASYNCFULL
, q
);
589 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
590 * it already be started by driver.
592 #define RQ_NOMERGE_FLAGS \
593 (REQ_NOMERGE | REQ_STARTED | REQ_SOFTBARRIER | REQ_FLUSH | REQ_FUA)
594 #define rq_mergeable(rq) \
595 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
596 (((rq)->cmd_flags & REQ_DISCARD) || \
597 (rq)->cmd_type == REQ_TYPE_FS))
600 * q->prep_rq_fn return values
602 #define BLKPREP_OK 0 /* serve it */
603 #define BLKPREP_KILL 1 /* fatal error, kill */
604 #define BLKPREP_DEFER 2 /* leave on queue */
606 extern unsigned long blk_max_low_pfn
, blk_max_pfn
;
609 * standard bounce addresses:
611 * BLK_BOUNCE_HIGH : bounce all highmem pages
612 * BLK_BOUNCE_ANY : don't bounce anything
613 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
616 #if BITS_PER_LONG == 32
617 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
619 #define BLK_BOUNCE_HIGH -1ULL
621 #define BLK_BOUNCE_ANY (-1ULL)
622 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
625 * default timeout for SG_IO if none specified
627 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
628 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
631 extern int init_emergency_isa_pool(void);
632 extern void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
);
634 static inline int init_emergency_isa_pool(void)
638 static inline void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
)
641 #endif /* CONFIG_MMU */
647 unsigned long offset
;
652 struct req_iterator
{
657 /* This should not be used directly - use rq_for_each_segment */
658 #define for_each_bio(_bio) \
659 for (; _bio; _bio = _bio->bi_next)
660 #define __rq_for_each_bio(_bio, rq) \
662 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
664 #define rq_for_each_segment(bvl, _rq, _iter) \
665 __rq_for_each_bio(_iter.bio, _rq) \
666 bio_for_each_segment(bvl, _iter.bio, _iter.i)
668 #define rq_iter_last(rq, _iter) \
669 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
671 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
672 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
674 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
675 extern void rq_flush_dcache_pages(struct request
*rq
);
677 static inline void rq_flush_dcache_pages(struct request
*rq
)
682 extern int blk_register_queue(struct gendisk
*disk
);
683 extern void blk_unregister_queue(struct gendisk
*disk
);
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_requeue_request(struct request_queue
*, struct request
*);
692 extern void blk_add_request_payload(struct request
*rq
, struct page
*page
,
694 extern int blk_rq_check_limits(struct request_queue
*q
, struct request
*rq
);
695 extern int blk_lld_busy(struct request_queue
*q
);
696 extern int blk_rq_prep_clone(struct request
*rq
, struct request
*rq_src
,
697 struct bio_set
*bs
, gfp_t gfp_mask
,
698 int (*bio_ctr
)(struct bio
*, struct bio
*, void *),
700 extern void blk_rq_unprep_clone(struct request
*rq
);
701 extern int blk_insert_cloned_request(struct request_queue
*q
,
703 extern void blk_delay_queue(struct request_queue
*, unsigned long);
704 extern void blk_recount_segments(struct request_queue
*, struct bio
*);
705 extern int scsi_verify_blk_ioctl(struct block_device
*, unsigned int);
706 extern int scsi_cmd_blk_ioctl(struct block_device
*, fmode_t
,
707 unsigned int, void __user
*);
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
*);
713 extern void blk_queue_bio(struct request_queue
*q
, struct bio
*bio
);
716 * A queue has just exitted congestion. Note this in the global counter of
717 * congested queues, and wake up anyone who was waiting for requests to be
720 static inline void blk_clear_queue_congested(struct request_queue
*q
, int sync
)
722 clear_bdi_congested(&q
->backing_dev_info
, sync
);
726 * A queue has just entered congestion. Flag that in the queue's VM-visible
727 * state flags and increment the global gounter of congested queues.
729 static inline void blk_set_queue_congested(struct request_queue
*q
, int sync
)
731 set_bdi_congested(&q
->backing_dev_info
, sync
);
734 extern void blk_start_queue(struct request_queue
*q
);
735 extern void blk_stop_queue(struct request_queue
*q
);
736 extern void blk_sync_queue(struct request_queue
*q
);
737 extern void __blk_stop_queue(struct request_queue
*q
);
738 extern void __blk_run_queue(struct request_queue
*q
);
739 extern void blk_run_queue(struct request_queue
*);
740 extern void blk_run_queue_async(struct request_queue
*q
);
741 extern int blk_rq_map_user(struct request_queue
*, struct request
*,
742 struct rq_map_data
*, void __user
*, unsigned long,
744 extern int blk_rq_unmap_user(struct bio
*);
745 extern int blk_rq_map_kern(struct request_queue
*, struct request
*, void *, unsigned int, gfp_t
);
746 extern int blk_rq_map_user_iov(struct request_queue
*, struct request
*,
747 struct rq_map_data
*, struct sg_iovec
*, int,
748 unsigned int, gfp_t
);
749 extern int blk_execute_rq(struct request_queue
*, struct gendisk
*,
750 struct request
*, int);
751 extern void blk_execute_rq_nowait(struct request_queue
*, struct gendisk
*,
752 struct request
*, int, rq_end_io_fn
*);
754 static inline struct request_queue
*bdev_get_queue(struct block_device
*bdev
)
756 return bdev
->bd_disk
->queue
;
760 * blk_rq_pos() : the current sector
761 * blk_rq_bytes() : bytes left in the entire request
762 * blk_rq_cur_bytes() : bytes left in the current segment
763 * blk_rq_err_bytes() : bytes left till the next error boundary
764 * blk_rq_sectors() : sectors left in the entire request
765 * blk_rq_cur_sectors() : sectors left in the current segment
767 static inline sector_t
blk_rq_pos(const struct request
*rq
)
772 static inline unsigned int blk_rq_bytes(const struct request
*rq
)
774 return rq
->__data_len
;
777 static inline int blk_rq_cur_bytes(const struct request
*rq
)
779 return rq
->bio
? bio_cur_bytes(rq
->bio
) : 0;
782 extern unsigned int blk_rq_err_bytes(const struct request
*rq
);
784 static inline unsigned int blk_rq_sectors(const struct request
*rq
)
786 return blk_rq_bytes(rq
) >> 9;
789 static inline unsigned int blk_rq_cur_sectors(const struct request
*rq
)
791 return blk_rq_cur_bytes(rq
) >> 9;
795 * Request issue related functions.
797 extern struct request
*blk_peek_request(struct request_queue
*q
);
798 extern void blk_start_request(struct request
*rq
);
799 extern struct request
*blk_fetch_request(struct request_queue
*q
);
802 * Request completion related functions.
804 * blk_update_request() completes given number of bytes and updates
805 * the request without completing it.
807 * blk_end_request() and friends. __blk_end_request() must be called
808 * with the request queue spinlock acquired.
810 * Several drivers define their own end_request and call
811 * blk_end_request() for parts of the original function.
812 * This prevents code duplication in drivers.
814 extern bool blk_update_request(struct request
*rq
, int error
,
815 unsigned int nr_bytes
);
816 extern bool blk_end_request(struct request
*rq
, int error
,
817 unsigned int nr_bytes
);
818 extern void blk_end_request_all(struct request
*rq
, int error
);
819 extern bool blk_end_request_cur(struct request
*rq
, int error
);
820 extern bool blk_end_request_err(struct request
*rq
, int error
);
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
);
827 extern void blk_complete_request(struct request
*);
828 extern void __blk_complete_request(struct request
*);
829 extern void blk_abort_request(struct request
*);
830 extern void blk_unprep_request(struct request
*);
833 * Access functions for manipulating queue properties
835 extern struct request_queue
*blk_init_queue_node(request_fn_proc
*rfn
,
836 spinlock_t
*lock
, int node_id
);
837 extern struct request_queue
*blk_init_queue(request_fn_proc
*, spinlock_t
*);
838 extern struct request_queue
*blk_init_allocated_queue(struct request_queue
*,
839 request_fn_proc
*, spinlock_t
*);
840 extern void blk_cleanup_queue(struct request_queue
*);
841 extern void blk_queue_make_request(struct request_queue
*, make_request_fn
*);
842 extern void blk_queue_bounce_limit(struct request_queue
*, u64
);
843 extern void blk_limits_max_hw_sectors(struct queue_limits
*, unsigned int);
844 extern void blk_queue_max_hw_sectors(struct request_queue
*, unsigned int);
845 extern void blk_queue_max_segments(struct request_queue
*, unsigned short);
846 extern void blk_queue_max_segment_size(struct request_queue
*, unsigned int);
847 extern void blk_queue_max_discard_sectors(struct request_queue
*q
,
848 unsigned int max_discard_sectors
);
849 extern void blk_queue_logical_block_size(struct request_queue
*, unsigned short);
850 extern void blk_queue_physical_block_size(struct request_queue
*, unsigned int);
851 extern void blk_queue_alignment_offset(struct request_queue
*q
,
852 unsigned int alignment
);
853 extern void blk_limits_io_min(struct queue_limits
*limits
, unsigned int min
);
854 extern void blk_queue_io_min(struct request_queue
*q
, unsigned int min
);
855 extern void blk_limits_io_opt(struct queue_limits
*limits
, unsigned int opt
);
856 extern void blk_queue_io_opt(struct request_queue
*q
, unsigned int opt
);
857 extern void blk_set_default_limits(struct queue_limits
*lim
);
858 extern void blk_set_stacking_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 void blk_queue_flush(struct request_queue
*q
, unsigned int flush
);
882 extern void blk_queue_flush_queueable(struct request_queue
*q
, bool queueable
);
883 extern struct backing_dev_info
*blk_get_backing_dev_info(struct block_device
*bdev
);
885 extern int blk_rq_map_sg(struct request_queue
*, struct request
*, struct scatterlist
*);
886 extern void blk_dump_rq_flags(struct request
*, char *);
887 extern long nr_blockdev_pages(void);
889 bool __must_check
blk_get_queue(struct request_queue
*);
890 struct request_queue
*blk_alloc_queue(gfp_t
);
891 struct request_queue
*blk_alloc_queue_node(gfp_t
, int);
892 extern void blk_put_queue(struct request_queue
*);
895 * blk_plug permits building a queue of related requests by holding the I/O
896 * fragments for a short period. This allows merging of sequential requests
897 * into single larger request. As the requests are moved from a per-task list to
898 * the device's request_queue in a batch, this results in improved scalability
899 * as the lock contention for request_queue lock is reduced.
901 * It is ok not to disable preemption when adding the request to the plug list
902 * or when attempting a merge, because blk_schedule_flush_list() will only flush
903 * the plug list when the task sleeps by itself. For details, please see
904 * schedule() where blk_schedule_flush_plug() is called.
907 unsigned long magic
; /* detect uninitialized use-cases */
908 struct list_head list
; /* requests */
909 struct list_head cb_list
; /* md requires an unplug callback */
910 unsigned int should_sort
; /* list to be sorted before flushing? */
912 #define BLK_MAX_REQUEST_COUNT 16
915 struct list_head list
;
916 void (*callback
)(struct blk_plug_cb
*);
919 extern void blk_start_plug(struct blk_plug
*);
920 extern void blk_finish_plug(struct blk_plug
*);
921 extern void blk_flush_plug_list(struct blk_plug
*, bool);
923 static inline void blk_flush_plug(struct task_struct
*tsk
)
925 struct blk_plug
*plug
= tsk
->plug
;
928 blk_flush_plug_list(plug
, false);
931 static inline void blk_schedule_flush_plug(struct task_struct
*tsk
)
933 struct blk_plug
*plug
= tsk
->plug
;
936 blk_flush_plug_list(plug
, true);
939 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
941 struct blk_plug
*plug
= tsk
->plug
;
943 return plug
&& (!list_empty(&plug
->list
) || !list_empty(&plug
->cb_list
));
949 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
950 extern int blk_queue_start_tag(struct request_queue
*, struct request
*);
951 extern struct request
*blk_queue_find_tag(struct request_queue
*, int);
952 extern void blk_queue_end_tag(struct request_queue
*, struct request
*);
953 extern int blk_queue_init_tags(struct request_queue
*, int, struct blk_queue_tag
*);
954 extern void blk_queue_free_tags(struct request_queue
*);
955 extern int blk_queue_resize_tags(struct request_queue
*, int);
956 extern void blk_queue_invalidate_tags(struct request_queue
*);
957 extern struct blk_queue_tag
*blk_init_tags(int);
958 extern void blk_free_tags(struct blk_queue_tag
*);
960 static inline struct request
*blk_map_queue_find_tag(struct blk_queue_tag
*bqt
,
963 if (unlikely(bqt
== NULL
|| tag
>= bqt
->real_max_depth
))
965 return bqt
->tag_index
[tag
];
968 #define BLKDEV_DISCARD_SECURE 0x01 /* secure discard */
970 extern int blkdev_issue_flush(struct block_device
*, gfp_t
, sector_t
*);
971 extern int blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
972 sector_t nr_sects
, gfp_t gfp_mask
, unsigned long flags
);
973 extern int blkdev_issue_zeroout(struct block_device
*bdev
, sector_t sector
,
974 sector_t nr_sects
, gfp_t gfp_mask
);
975 static inline int sb_issue_discard(struct super_block
*sb
, sector_t block
,
976 sector_t nr_blocks
, gfp_t gfp_mask
, unsigned long flags
)
978 return blkdev_issue_discard(sb
->s_bdev
, block
<< (sb
->s_blocksize_bits
- 9),
979 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
982 static inline int sb_issue_zeroout(struct super_block
*sb
, sector_t block
,
983 sector_t nr_blocks
, gfp_t gfp_mask
)
985 return blkdev_issue_zeroout(sb
->s_bdev
,
986 block
<< (sb
->s_blocksize_bits
- 9),
987 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
991 extern int blk_verify_command(unsigned char *cmd
, fmode_t has_write_perm
);
993 enum blk_default_limits
{
994 BLK_MAX_SEGMENTS
= 128,
995 BLK_SAFE_MAX_SECTORS
= 255,
996 BLK_DEF_MAX_SECTORS
= 1024,
997 BLK_MAX_SEGMENT_SIZE
= 65536,
998 BLK_SEG_BOUNDARY_MASK
= 0xFFFFFFFFUL
,
1001 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
1003 static inline unsigned long queue_bounce_pfn(struct request_queue
*q
)
1005 return q
->limits
.bounce_pfn
;
1008 static inline unsigned long queue_segment_boundary(struct request_queue
*q
)
1010 return q
->limits
.seg_boundary_mask
;
1013 static inline unsigned int queue_max_sectors(struct request_queue
*q
)
1015 return q
->limits
.max_sectors
;
1018 static inline unsigned int queue_max_hw_sectors(struct request_queue
*q
)
1020 return q
->limits
.max_hw_sectors
;
1023 static inline unsigned short queue_max_segments(struct request_queue
*q
)
1025 return q
->limits
.max_segments
;
1028 static inline unsigned int queue_max_segment_size(struct request_queue
*q
)
1030 return q
->limits
.max_segment_size
;
1033 static inline unsigned short queue_logical_block_size(struct request_queue
*q
)
1037 if (q
&& q
->limits
.logical_block_size
)
1038 retval
= q
->limits
.logical_block_size
;
1043 static inline unsigned short bdev_logical_block_size(struct block_device
*bdev
)
1045 return queue_logical_block_size(bdev_get_queue(bdev
));
1048 static inline unsigned int queue_physical_block_size(struct request_queue
*q
)
1050 return q
->limits
.physical_block_size
;
1053 static inline unsigned int bdev_physical_block_size(struct block_device
*bdev
)
1055 return queue_physical_block_size(bdev_get_queue(bdev
));
1058 static inline unsigned int queue_io_min(struct request_queue
*q
)
1060 return q
->limits
.io_min
;
1063 static inline int bdev_io_min(struct block_device
*bdev
)
1065 return queue_io_min(bdev_get_queue(bdev
));
1068 static inline unsigned int queue_io_opt(struct request_queue
*q
)
1070 return q
->limits
.io_opt
;
1073 static inline int bdev_io_opt(struct block_device
*bdev
)
1075 return queue_io_opt(bdev_get_queue(bdev
));
1078 static inline int queue_alignment_offset(struct request_queue
*q
)
1080 if (q
->limits
.misaligned
)
1083 return q
->limits
.alignment_offset
;
1086 static inline int queue_limit_alignment_offset(struct queue_limits
*lim
, sector_t sector
)
1088 unsigned int granularity
= max(lim
->physical_block_size
, lim
->io_min
);
1089 unsigned int alignment
= (sector
<< 9) & (granularity
- 1);
1091 return (granularity
+ lim
->alignment_offset
- alignment
)
1092 & (granularity
- 1);
1095 static inline int bdev_alignment_offset(struct block_device
*bdev
)
1097 struct request_queue
*q
= bdev_get_queue(bdev
);
1099 if (q
->limits
.misaligned
)
1102 if (bdev
!= bdev
->bd_contains
)
1103 return bdev
->bd_part
->alignment_offset
;
1105 return q
->limits
.alignment_offset
;
1108 static inline int queue_discard_alignment(struct request_queue
*q
)
1110 if (q
->limits
.discard_misaligned
)
1113 return q
->limits
.discard_alignment
;
1116 static inline int queue_limit_discard_alignment(struct queue_limits
*lim
, sector_t sector
)
1118 unsigned int alignment
= (sector
<< 9) & (lim
->discard_granularity
- 1);
1120 if (!lim
->max_discard_sectors
)
1123 return (lim
->discard_granularity
+ lim
->discard_alignment
- alignment
)
1124 & (lim
->discard_granularity
- 1);
1127 static inline unsigned int queue_discard_zeroes_data(struct request_queue
*q
)
1129 if (q
->limits
.max_discard_sectors
&& q
->limits
.discard_zeroes_data
== 1)
1135 static inline unsigned int bdev_discard_zeroes_data(struct block_device
*bdev
)
1137 return queue_discard_zeroes_data(bdev_get_queue(bdev
));
1140 static inline int queue_dma_alignment(struct request_queue
*q
)
1142 return q
? q
->dma_alignment
: 511;
1145 static inline int blk_rq_aligned(struct request_queue
*q
, unsigned long addr
,
1148 unsigned int alignment
= queue_dma_alignment(q
) | q
->dma_pad_mask
;
1149 return !(addr
& alignment
) && !(len
& alignment
);
1152 /* assumes size > 256 */
1153 static inline unsigned int blksize_bits(unsigned int size
)
1155 unsigned int bits
= 8;
1159 } while (size
> 256);
1163 static inline unsigned int block_size(struct block_device
*bdev
)
1165 return bdev
->bd_block_size
;
1168 static inline bool queue_flush_queueable(struct request_queue
*q
)
1170 return !q
->flush_not_queueable
;
1173 typedef struct {struct page
*v
;} Sector
;
1175 unsigned char *read_dev_sector(struct block_device
*, sector_t
, Sector
*);
1177 static inline void put_dev_sector(Sector p
)
1179 page_cache_release(p
.v
);
1183 int kblockd_schedule_work(struct request_queue
*q
, struct work_struct
*work
);
1185 #ifdef CONFIG_BLK_CGROUP
1187 * This should not be using sched_clock(). A real patch is in progress
1188 * to fix this up, until that is in place we need to disable preemption
1189 * around sched_clock() in this function and set_io_start_time_ns().
1191 static inline void set_start_time_ns(struct request
*req
)
1194 req
->start_time_ns
= sched_clock();
1198 static inline void set_io_start_time_ns(struct request
*req
)
1201 req
->io_start_time_ns
= sched_clock();
1205 static inline uint64_t rq_start_time_ns(struct request
*req
)
1207 return req
->start_time_ns
;
1210 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1212 return req
->io_start_time_ns
;
1215 static inline void set_start_time_ns(struct request
*req
) {}
1216 static inline void set_io_start_time_ns(struct request
*req
) {}
1217 static inline uint64_t rq_start_time_ns(struct request
*req
)
1221 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1227 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1228 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1229 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1230 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1232 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1234 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1235 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1237 struct blk_integrity_exchg
{
1241 unsigned int data_size
;
1242 unsigned short sector_size
;
1243 const char *disk_name
;
1246 typedef void (integrity_gen_fn
) (struct blk_integrity_exchg
*);
1247 typedef int (integrity_vrfy_fn
) (struct blk_integrity_exchg
*);
1248 typedef void (integrity_set_tag_fn
) (void *, void *, unsigned int);
1249 typedef void (integrity_get_tag_fn
) (void *, void *, unsigned int);
1251 struct blk_integrity
{
1252 integrity_gen_fn
*generate_fn
;
1253 integrity_vrfy_fn
*verify_fn
;
1254 integrity_set_tag_fn
*set_tag_fn
;
1255 integrity_get_tag_fn
*get_tag_fn
;
1257 unsigned short flags
;
1258 unsigned short tuple_size
;
1259 unsigned short sector_size
;
1260 unsigned short tag_size
;
1264 struct kobject kobj
;
1267 extern bool blk_integrity_is_initialized(struct gendisk
*);
1268 extern int blk_integrity_register(struct gendisk
*, struct blk_integrity
*);
1269 extern void blk_integrity_unregister(struct gendisk
*);
1270 extern int blk_integrity_compare(struct gendisk
*, struct gendisk
*);
1271 extern int blk_rq_map_integrity_sg(struct request_queue
*, struct bio
*,
1272 struct scatterlist
*);
1273 extern int blk_rq_count_integrity_sg(struct request_queue
*, struct bio
*);
1274 extern int blk_integrity_merge_rq(struct request_queue
*, struct request
*,
1276 extern int blk_integrity_merge_bio(struct request_queue
*, struct request
*,
1280 struct blk_integrity
*bdev_get_integrity(struct block_device
*bdev
)
1282 return bdev
->bd_disk
->integrity
;
1285 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1287 return disk
->integrity
;
1290 static inline int blk_integrity_rq(struct request
*rq
)
1292 if (rq
->bio
== NULL
)
1295 return bio_integrity(rq
->bio
);
1298 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1301 q
->limits
.max_integrity_segments
= segs
;
1304 static inline unsigned short
1305 queue_max_integrity_segments(struct request_queue
*q
)
1307 return q
->limits
.max_integrity_segments
;
1310 #else /* CONFIG_BLK_DEV_INTEGRITY */
1313 struct block_device
;
1315 struct blk_integrity
;
1317 static inline int blk_integrity_rq(struct request
*rq
)
1321 static inline int blk_rq_count_integrity_sg(struct request_queue
*q
,
1326 static inline int blk_rq_map_integrity_sg(struct request_queue
*q
,
1328 struct scatterlist
*s
)
1332 static inline struct blk_integrity
*bdev_get_integrity(struct block_device
*b
)
1336 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1340 static inline int blk_integrity_compare(struct gendisk
*a
, struct gendisk
*b
)
1344 static inline int blk_integrity_register(struct gendisk
*d
,
1345 struct blk_integrity
*b
)
1349 static inline void blk_integrity_unregister(struct gendisk
*d
)
1352 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1356 static inline unsigned short queue_max_integrity_segments(struct request_queue
*q
)
1360 static inline int blk_integrity_merge_rq(struct request_queue
*rq
,
1366 static inline int blk_integrity_merge_bio(struct request_queue
*rq
,
1372 static inline bool blk_integrity_is_initialized(struct gendisk
*g
)
1377 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1379 struct block_device_operations
{
1380 int (*open
) (struct block_device
*, fmode_t
);
1381 int (*release
) (struct gendisk
*, fmode_t
);
1382 int (*ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1383 int (*compat_ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1384 int (*direct_access
) (struct block_device
*, sector_t
,
1385 void **, unsigned long *);
1386 unsigned int (*check_events
) (struct gendisk
*disk
,
1387 unsigned int clearing
);
1388 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1389 int (*media_changed
) (struct gendisk
*);
1390 void (*unlock_native_capacity
) (struct gendisk
*);
1391 int (*revalidate_disk
) (struct gendisk
*);
1392 int (*getgeo
)(struct block_device
*, struct hd_geometry
*);
1393 /* this callback is with swap_lock and sometimes page table lock held */
1394 void (*swap_slot_free_notify
) (struct block_device
*, unsigned long);
1395 struct module
*owner
;
1398 extern int __blkdev_driver_ioctl(struct block_device
*, fmode_t
, unsigned int,
1400 #else /* CONFIG_BLOCK */
1402 * stubs for when the block layer is configured out
1404 #define buffer_heads_over_limit 0
1406 static inline long nr_blockdev_pages(void)
1414 static inline void blk_start_plug(struct blk_plug
*plug
)
1418 static inline void blk_finish_plug(struct blk_plug
*plug
)
1422 static inline void blk_flush_plug(struct task_struct
*task
)
1426 static inline void blk_schedule_flush_plug(struct task_struct
*task
)
1431 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
1436 #endif /* CONFIG_BLOCK */