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/llist.h>
12 #include <linux/timer.h>
13 #include <linux/workqueue.h>
14 #include <linux/pagemap.h>
15 #include <linux/backing-dev-defs.h>
16 #include <linux/wait.h>
17 #include <linux/mempool.h>
18 #include <linux/pfn.h>
19 #include <linux/bio.h>
20 #include <linux/stringify.h>
21 #include <linux/gfp.h>
22 #include <linux/bsg.h>
23 #include <linux/smp.h>
24 #include <linux/rcupdate.h>
25 #include <linux/percpu-refcount.h>
26 #include <linux/scatterlist.h>
27 #include <linux/blkzoned.h>
30 struct scsi_ioctl_command
;
33 struct elevator_queue
;
39 struct blk_flush_queue
;
43 #define BLKDEV_MIN_RQ 4
44 #define BLKDEV_MAX_RQ 128 /* Default maximum */
47 * Maximum number of blkcg policies allowed to be registered concurrently.
48 * Defined here to simplify include dependency.
50 #define BLKCG_MAX_POLS 2
52 typedef void (rq_end_io_fn
)(struct request
*, int);
54 #define BLK_RL_SYNCFULL (1U << 0)
55 #define BLK_RL_ASYNCFULL (1U << 1)
58 struct request_queue
*q
; /* the queue this rl belongs to */
59 #ifdef CONFIG_BLK_CGROUP
60 struct blkcg_gq
*blkg
; /* blkg this request pool belongs to */
63 * count[], starved[], and wait[] are indexed by
64 * BLK_RW_SYNC/BLK_RW_ASYNC
69 wait_queue_head_t wait
[2];
74 * request command types
76 enum rq_cmd_type_bits
{
77 REQ_TYPE_FS
= 1, /* fs request */
78 REQ_TYPE_BLOCK_PC
, /* scsi command */
79 REQ_TYPE_DRV_PRIV
, /* driver defined types from here */
84 typedef __u32 __bitwise req_flags_t
;
86 /* elevator knows about this request */
87 #define RQF_SORTED ((__force req_flags_t)(1 << 0))
88 /* drive already may have started this one */
89 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
90 /* uses tagged queueing */
91 #define RQF_QUEUED ((__force req_flags_t)(1 << 2))
92 /* may not be passed by ioscheduler */
93 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
94 /* request for flush sequence */
95 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
96 /* merge of different types, fail separately */
97 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
98 /* track inflight for MQ */
99 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
100 /* don't call prep for this one */
101 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
102 /* set for "ide_preempt" requests and also for requests for which the SCSI
103 "quiesce" state must be ignored. */
104 #define RQF_PREEMPT ((__force req_flags_t)(1 << 8))
105 /* contains copies of user pages */
106 #define RQF_COPY_USER ((__force req_flags_t)(1 << 9))
107 /* vaguely specified driver internal error. Ignored by the block layer */
108 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
109 /* don't warn about errors */
110 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
111 /* elevator private data attached */
112 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
113 /* account I/O stat */
114 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
115 /* request came from our alloc pool */
116 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
117 /* runtime pm request */
118 #define RQF_PM ((__force req_flags_t)(1 << 15))
119 /* on IO scheduler merge hash */
120 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
121 /* IO stats tracking on */
122 #define RQF_STATS ((__force req_flags_t)(1 << 17))
123 /* Look at ->special_vec for the actual data payload instead of the
125 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
127 /* flags that prevent us from merging requests: */
128 #define RQF_NOMERGE_FLAGS \
129 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
132 * Try to put the fields that are referenced together in the same cacheline.
134 * If you modify this structure, make sure to update blk_rq_init() and
135 * especially blk_mq_rq_ctx_init() to take care of the added fields.
138 struct list_head queuelist
;
140 struct call_single_data csd
;
144 struct request_queue
*q
;
145 struct blk_mq_ctx
*mq_ctx
;
149 unsigned int cmd_flags
; /* op and common flags */
150 req_flags_t rq_flags
;
151 unsigned long atomic_flags
;
153 /* the following two fields are internal, NEVER access directly */
154 unsigned int __data_len
; /* total data len */
156 sector_t __sector
; /* sector cursor */
162 * The hash is used inside the scheduler, and killed once the
163 * request reaches the dispatch list. The ipi_list is only used
164 * to queue the request for softirq completion, which is long
165 * after the request has been unhashed (and even removed from
166 * the dispatch list).
169 struct hlist_node hash
; /* merge hash */
170 struct list_head ipi_list
;
174 * The rb_node is only used inside the io scheduler, requests
175 * are pruned when moved to the dispatch queue. So let the
176 * completion_data share space with the rb_node.
179 struct rb_node rb_node
; /* sort/lookup */
180 struct bio_vec special_vec
;
181 void *completion_data
;
185 * Three pointers are available for the IO schedulers, if they need
186 * more they have to dynamically allocate it. Flush requests are
187 * never put on the IO scheduler. So let the flush fields share
188 * space with the elevator data.
198 struct list_head list
;
199 rq_end_io_fn
*saved_end_io
;
203 struct gendisk
*rq_disk
;
204 struct hd_struct
*part
;
205 unsigned long start_time
;
206 struct blk_issue_stat issue_stat
;
207 #ifdef CONFIG_BLK_CGROUP
208 struct request_list
*rl
; /* rl this rq is alloced from */
209 unsigned long long start_time_ns
;
210 unsigned long long io_start_time_ns
; /* when passed to hardware */
212 /* Number of scatter-gather DMA addr+len pairs after
213 * physical address coalescing is performed.
215 unsigned short nr_phys_segments
;
216 #if defined(CONFIG_BLK_DEV_INTEGRITY)
217 unsigned short nr_integrity_segments
;
220 unsigned short ioprio
;
224 void *special
; /* opaque pointer available for LLD use */
228 unsigned int extra_len
; /* length of alignment and padding */
230 unsigned long deadline
;
231 struct list_head timeout_list
;
232 unsigned int timeout
;
236 * completion callback.
238 rq_end_io_fn
*end_io
;
242 struct request
*next_rq
;
245 static inline bool blk_rq_is_passthrough(struct request
*rq
)
247 return rq
->cmd_type
!= REQ_TYPE_FS
;
250 static inline unsigned short req_get_ioprio(struct request
*req
)
255 #include <linux/elevator.h>
257 struct blk_queue_ctx
;
259 typedef void (request_fn_proc
) (struct request_queue
*q
);
260 typedef blk_qc_t (make_request_fn
) (struct request_queue
*q
, struct bio
*bio
);
261 typedef int (prep_rq_fn
) (struct request_queue
*, struct request
*);
262 typedef void (unprep_rq_fn
) (struct request_queue
*, struct request
*);
265 typedef void (softirq_done_fn
)(struct request
*);
266 typedef int (dma_drain_needed_fn
)(struct request
*);
267 typedef int (lld_busy_fn
) (struct request_queue
*q
);
268 typedef int (bsg_job_fn
) (struct bsg_job
*);
269 typedef int (init_rq_fn
)(struct request_queue
*, struct request
*, gfp_t
);
270 typedef void (exit_rq_fn
)(struct request_queue
*, struct request
*);
272 enum blk_eh_timer_return
{
278 typedef enum blk_eh_timer_return (rq_timed_out_fn
)(struct request
*);
280 enum blk_queue_state
{
285 struct blk_queue_tag
{
286 struct request
**tag_index
; /* map of busy tags */
287 unsigned long *tag_map
; /* bit map of free/busy tags */
288 int max_depth
; /* what we will send to device */
289 int real_max_depth
; /* what the array can hold */
290 atomic_t refcnt
; /* map can be shared */
291 int alloc_policy
; /* tag allocation policy */
292 int next_tag
; /* next tag */
294 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
295 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
297 #define BLK_SCSI_MAX_CMDS (256)
298 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
301 * Zoned block device models (zoned limit).
303 enum blk_zoned_model
{
304 BLK_ZONED_NONE
, /* Regular block device */
305 BLK_ZONED_HA
, /* Host-aware zoned block device */
306 BLK_ZONED_HM
, /* Host-managed zoned block device */
309 struct queue_limits
{
310 unsigned long bounce_pfn
;
311 unsigned long seg_boundary_mask
;
312 unsigned long virt_boundary_mask
;
314 unsigned int max_hw_sectors
;
315 unsigned int max_dev_sectors
;
316 unsigned int chunk_sectors
;
317 unsigned int max_sectors
;
318 unsigned int max_segment_size
;
319 unsigned int physical_block_size
;
320 unsigned int alignment_offset
;
323 unsigned int max_discard_sectors
;
324 unsigned int max_hw_discard_sectors
;
325 unsigned int max_write_same_sectors
;
326 unsigned int max_write_zeroes_sectors
;
327 unsigned int discard_granularity
;
328 unsigned int discard_alignment
;
330 unsigned short logical_block_size
;
331 unsigned short max_segments
;
332 unsigned short max_integrity_segments
;
334 unsigned char misaligned
;
335 unsigned char discard_misaligned
;
336 unsigned char cluster
;
337 unsigned char discard_zeroes_data
;
338 unsigned char raid_partial_stripes_expensive
;
339 enum blk_zoned_model zoned
;
342 #ifdef CONFIG_BLK_DEV_ZONED
344 struct blk_zone_report_hdr
{
345 unsigned int nr_zones
;
349 extern int blkdev_report_zones(struct block_device
*bdev
,
350 sector_t sector
, struct blk_zone
*zones
,
351 unsigned int *nr_zones
, gfp_t gfp_mask
);
352 extern int blkdev_reset_zones(struct block_device
*bdev
, sector_t sectors
,
353 sector_t nr_sectors
, gfp_t gfp_mask
);
355 extern int blkdev_report_zones_ioctl(struct block_device
*bdev
, fmode_t mode
,
356 unsigned int cmd
, unsigned long arg
);
357 extern int blkdev_reset_zones_ioctl(struct block_device
*bdev
, fmode_t mode
,
358 unsigned int cmd
, unsigned long arg
);
360 #else /* CONFIG_BLK_DEV_ZONED */
362 static inline int blkdev_report_zones_ioctl(struct block_device
*bdev
,
363 fmode_t mode
, unsigned int cmd
,
369 static inline int blkdev_reset_zones_ioctl(struct block_device
*bdev
,
370 fmode_t mode
, unsigned int cmd
,
376 #endif /* CONFIG_BLK_DEV_ZONED */
378 struct request_queue
{
380 * Together with queue_head for cacheline sharing
382 struct list_head queue_head
;
383 struct request
*last_merge
;
384 struct elevator_queue
*elevator
;
385 int nr_rqs
[2]; /* # allocated [a]sync rqs */
386 int nr_rqs_elvpriv
; /* # allocated rqs w/ elvpriv */
391 * If blkcg is not used, @q->root_rl serves all requests. If blkcg
392 * is used, root blkg allocates from @q->root_rl and all other
393 * blkgs from their own blkg->rl. Which one to use should be
394 * determined using bio_request_list().
396 struct request_list root_rl
;
398 request_fn_proc
*request_fn
;
399 make_request_fn
*make_request_fn
;
400 prep_rq_fn
*prep_rq_fn
;
401 unprep_rq_fn
*unprep_rq_fn
;
402 softirq_done_fn
*softirq_done_fn
;
403 rq_timed_out_fn
*rq_timed_out_fn
;
404 dma_drain_needed_fn
*dma_drain_needed
;
405 lld_busy_fn
*lld_busy_fn
;
406 init_rq_fn
*init_rq_fn
;
407 exit_rq_fn
*exit_rq_fn
;
409 const struct blk_mq_ops
*mq_ops
;
411 unsigned int *mq_map
;
414 struct blk_mq_ctx __percpu
*queue_ctx
;
415 unsigned int nr_queues
;
417 unsigned int queue_depth
;
419 /* hw dispatch queues */
420 struct blk_mq_hw_ctx
**queue_hw_ctx
;
421 unsigned int nr_hw_queues
;
424 * Dispatch queue sorting
427 struct request
*boundary_rq
;
430 * Delayed queue handling
432 struct delayed_work delay_work
;
434 struct backing_dev_info backing_dev_info
;
437 * The queue owner gets to use this for whatever they like.
438 * ll_rw_blk doesn't touch it.
443 * various queue flags, see QUEUE_* below
445 unsigned long queue_flags
;
448 * ida allocated id for this queue. Used to index queues from
454 * queue needs bounce pages for pages above this limit
459 * protects queue structures from reentrancy. ->__queue_lock should
460 * _never_ be used directly, it is queue private. always use
463 spinlock_t __queue_lock
;
464 spinlock_t
*queue_lock
;
474 struct kobject mq_kobj
;
476 #ifdef CONFIG_BLK_DEV_INTEGRITY
477 struct blk_integrity integrity
;
478 #endif /* CONFIG_BLK_DEV_INTEGRITY */
483 unsigned int nr_pending
;
489 unsigned long nr_requests
; /* Max # of requests */
490 unsigned int nr_congestion_on
;
491 unsigned int nr_congestion_off
;
492 unsigned int nr_batching
;
494 unsigned int dma_drain_size
;
495 void *dma_drain_buffer
;
496 unsigned int dma_pad_mask
;
497 unsigned int dma_alignment
;
499 struct blk_queue_tag
*queue_tags
;
500 struct list_head tag_busy_list
;
502 unsigned int nr_sorted
;
503 unsigned int in_flight
[2];
505 struct blk_rq_stat rq_stats
[2];
508 * Number of active block driver functions for which blk_drain_queue()
509 * must wait. Must be incremented around functions that unlock the
510 * queue_lock internally, e.g. scsi_request_fn().
512 unsigned int request_fn_active
;
514 unsigned int rq_timeout
;
516 struct timer_list timeout
;
517 struct work_struct timeout_work
;
518 struct list_head timeout_list
;
520 struct list_head icq_list
;
521 #ifdef CONFIG_BLK_CGROUP
522 DECLARE_BITMAP (blkcg_pols
, BLKCG_MAX_POLS
);
523 struct blkcg_gq
*root_blkg
;
524 struct list_head blkg_list
;
527 struct queue_limits limits
;
532 unsigned int sg_timeout
;
533 unsigned int sg_reserved_size
;
535 #ifdef CONFIG_BLK_DEV_IO_TRACE
536 struct blk_trace
*blk_trace
;
539 * for flush operations
541 struct blk_flush_queue
*fq
;
543 struct list_head requeue_list
;
544 spinlock_t requeue_lock
;
545 struct delayed_work requeue_work
;
547 struct mutex sysfs_lock
;
550 atomic_t mq_freeze_depth
;
552 #if defined(CONFIG_BLK_DEV_BSG)
553 bsg_job_fn
*bsg_job_fn
;
555 struct bsg_class_device bsg_dev
;
558 #ifdef CONFIG_BLK_DEV_THROTTLING
560 struct throtl_data
*td
;
562 struct rcu_head rcu_head
;
563 wait_queue_head_t mq_freeze_wq
;
564 struct percpu_ref q_usage_counter
;
565 struct list_head all_q_node
;
567 struct blk_mq_tag_set
*tag_set
;
568 struct list_head tag_set_list
;
569 struct bio_set
*bio_split
;
571 #ifdef CONFIG_DEBUG_FS
572 struct dentry
*debugfs_dir
;
573 struct dentry
*mq_debugfs_dir
;
576 bool mq_sysfs_init_done
;
582 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
583 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
584 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
585 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
586 #define QUEUE_FLAG_DYING 5 /* queue being torn down */
587 #define QUEUE_FLAG_BYPASS 6 /* act as dumb FIFO queue */
588 #define QUEUE_FLAG_BIDI 7 /* queue supports bidi requests */
589 #define QUEUE_FLAG_NOMERGES 8 /* disable merge attempts */
590 #define QUEUE_FLAG_SAME_COMP 9 /* complete on same CPU-group */
591 #define QUEUE_FLAG_FAIL_IO 10 /* fake timeout */
592 #define QUEUE_FLAG_STACKABLE 11 /* supports request stacking */
593 #define QUEUE_FLAG_NONROT 12 /* non-rotational device (SSD) */
594 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
595 #define QUEUE_FLAG_IO_STAT 13 /* do IO stats */
596 #define QUEUE_FLAG_DISCARD 14 /* supports DISCARD */
597 #define QUEUE_FLAG_NOXMERGES 15 /* No extended merges */
598 #define QUEUE_FLAG_ADD_RANDOM 16 /* Contributes to random pool */
599 #define QUEUE_FLAG_SECERASE 17 /* supports secure erase */
600 #define QUEUE_FLAG_SAME_FORCE 18 /* force complete on same CPU */
601 #define QUEUE_FLAG_DEAD 19 /* queue tear-down finished */
602 #define QUEUE_FLAG_INIT_DONE 20 /* queue is initialized */
603 #define QUEUE_FLAG_NO_SG_MERGE 21 /* don't attempt to merge SG segments*/
604 #define QUEUE_FLAG_POLL 22 /* IO polling enabled if set */
605 #define QUEUE_FLAG_WC 23 /* Write back caching */
606 #define QUEUE_FLAG_FUA 24 /* device supports FUA writes */
607 #define QUEUE_FLAG_FLUSH_NQ 25 /* flush not queueuable */
608 #define QUEUE_FLAG_DAX 26 /* device supports DAX */
609 #define QUEUE_FLAG_STATS 27 /* track rq completion times */
610 #define QUEUE_FLAG_RESTART 28 /* queue needs restart at completion */
612 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
613 (1 << QUEUE_FLAG_STACKABLE) | \
614 (1 << QUEUE_FLAG_SAME_COMP) | \
615 (1 << QUEUE_FLAG_ADD_RANDOM))
617 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
618 (1 << QUEUE_FLAG_STACKABLE) | \
619 (1 << QUEUE_FLAG_SAME_COMP) | \
620 (1 << QUEUE_FLAG_POLL))
622 static inline void queue_lockdep_assert_held(struct request_queue
*q
)
625 lockdep_assert_held(q
->queue_lock
);
628 static inline void queue_flag_set_unlocked(unsigned int flag
,
629 struct request_queue
*q
)
631 __set_bit(flag
, &q
->queue_flags
);
634 static inline int queue_flag_test_and_clear(unsigned int flag
,
635 struct request_queue
*q
)
637 queue_lockdep_assert_held(q
);
639 if (test_bit(flag
, &q
->queue_flags
)) {
640 __clear_bit(flag
, &q
->queue_flags
);
647 static inline int queue_flag_test_and_set(unsigned int flag
,
648 struct request_queue
*q
)
650 queue_lockdep_assert_held(q
);
652 if (!test_bit(flag
, &q
->queue_flags
)) {
653 __set_bit(flag
, &q
->queue_flags
);
660 static inline void queue_flag_set(unsigned int flag
, struct request_queue
*q
)
662 queue_lockdep_assert_held(q
);
663 __set_bit(flag
, &q
->queue_flags
);
666 static inline void queue_flag_clear_unlocked(unsigned int flag
,
667 struct request_queue
*q
)
669 __clear_bit(flag
, &q
->queue_flags
);
672 static inline int queue_in_flight(struct request_queue
*q
)
674 return q
->in_flight
[0] + q
->in_flight
[1];
677 static inline void queue_flag_clear(unsigned int flag
, struct request_queue
*q
)
679 queue_lockdep_assert_held(q
);
680 __clear_bit(flag
, &q
->queue_flags
);
683 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
684 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
685 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
686 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
687 #define blk_queue_bypass(q) test_bit(QUEUE_FLAG_BYPASS, &(q)->queue_flags)
688 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
689 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
690 #define blk_queue_noxmerges(q) \
691 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
692 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
693 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
694 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
695 #define blk_queue_stackable(q) \
696 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
697 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
698 #define blk_queue_secure_erase(q) \
699 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
700 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
702 #define blk_noretry_request(rq) \
703 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
704 REQ_FAILFAST_DRIVER))
706 static inline bool blk_account_rq(struct request
*rq
)
708 return (rq
->rq_flags
& RQF_STARTED
) && !blk_rq_is_passthrough(rq
);
711 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
712 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
713 /* rq->queuelist of dequeued request must be list_empty() */
714 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
716 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
718 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
721 * Driver can handle struct request, if it either has an old style
722 * request_fn defined, or is blk-mq based.
724 static inline bool queue_is_rq_based(struct request_queue
*q
)
726 return q
->request_fn
|| q
->mq_ops
;
729 static inline unsigned int blk_queue_cluster(struct request_queue
*q
)
731 return q
->limits
.cluster
;
734 static inline enum blk_zoned_model
735 blk_queue_zoned_model(struct request_queue
*q
)
737 return q
->limits
.zoned
;
740 static inline bool blk_queue_is_zoned(struct request_queue
*q
)
742 switch (blk_queue_zoned_model(q
)) {
751 static inline unsigned int blk_queue_zone_sectors(struct request_queue
*q
)
753 return blk_queue_is_zoned(q
) ? q
->limits
.chunk_sectors
: 0;
756 static inline bool rq_is_sync(struct request
*rq
)
758 return op_is_sync(rq
->cmd_flags
);
761 static inline bool blk_rl_full(struct request_list
*rl
, bool sync
)
763 unsigned int flag
= sync
? BLK_RL_SYNCFULL
: BLK_RL_ASYNCFULL
;
765 return rl
->flags
& flag
;
768 static inline void blk_set_rl_full(struct request_list
*rl
, bool sync
)
770 unsigned int flag
= sync
? BLK_RL_SYNCFULL
: BLK_RL_ASYNCFULL
;
775 static inline void blk_clear_rl_full(struct request_list
*rl
, bool sync
)
777 unsigned int flag
= sync
? BLK_RL_SYNCFULL
: BLK_RL_ASYNCFULL
;
782 static inline bool rq_mergeable(struct request
*rq
)
784 if (blk_rq_is_passthrough(rq
))
787 if (req_op(rq
) == REQ_OP_FLUSH
)
790 if (req_op(rq
) == REQ_OP_WRITE_ZEROES
)
793 if (rq
->cmd_flags
& REQ_NOMERGE_FLAGS
)
795 if (rq
->rq_flags
& RQF_NOMERGE_FLAGS
)
801 static inline bool blk_write_same_mergeable(struct bio
*a
, struct bio
*b
)
803 if (bio_data(a
) == bio_data(b
))
809 static inline unsigned int blk_queue_depth(struct request_queue
*q
)
812 return q
->queue_depth
;
814 return q
->nr_requests
;
818 * q->prep_rq_fn return values
821 BLKPREP_OK
, /* serve it */
822 BLKPREP_KILL
, /* fatal error, kill, return -EIO */
823 BLKPREP_DEFER
, /* leave on queue */
824 BLKPREP_INVALID
, /* invalid command, kill, return -EREMOTEIO */
827 extern unsigned long blk_max_low_pfn
, blk_max_pfn
;
830 * standard bounce addresses:
832 * BLK_BOUNCE_HIGH : bounce all highmem pages
833 * BLK_BOUNCE_ANY : don't bounce anything
834 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
837 #if BITS_PER_LONG == 32
838 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
840 #define BLK_BOUNCE_HIGH -1ULL
842 #define BLK_BOUNCE_ANY (-1ULL)
843 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
846 * default timeout for SG_IO if none specified
848 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
849 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
852 extern int init_emergency_isa_pool(void);
853 extern void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
);
855 static inline int init_emergency_isa_pool(void)
859 static inline void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
)
862 #endif /* CONFIG_MMU */
868 unsigned long offset
;
873 struct req_iterator
{
874 struct bvec_iter iter
;
878 /* This should not be used directly - use rq_for_each_segment */
879 #define for_each_bio(_bio) \
880 for (; _bio; _bio = _bio->bi_next)
881 #define __rq_for_each_bio(_bio, rq) \
883 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
885 #define rq_for_each_segment(bvl, _rq, _iter) \
886 __rq_for_each_bio(_iter.bio, _rq) \
887 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
889 #define rq_iter_last(bvec, _iter) \
890 (_iter.bio->bi_next == NULL && \
891 bio_iter_last(bvec, _iter.iter))
893 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
894 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
896 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
897 extern void rq_flush_dcache_pages(struct request
*rq
);
899 static inline void rq_flush_dcache_pages(struct request
*rq
)
905 #define vfs_msg(sb, level, fmt, ...) \
906 __vfs_msg(sb, level, fmt, ##__VA_ARGS__)
908 #define vfs_msg(sb, level, fmt, ...) \
910 no_printk(fmt, ##__VA_ARGS__); \
911 __vfs_msg(sb, "", " "); \
915 extern int blk_register_queue(struct gendisk
*disk
);
916 extern void blk_unregister_queue(struct gendisk
*disk
);
917 extern blk_qc_t
generic_make_request(struct bio
*bio
);
918 extern void blk_rq_init(struct request_queue
*q
, struct request
*rq
);
919 extern void blk_put_request(struct request
*);
920 extern void __blk_put_request(struct request_queue
*, struct request
*);
921 extern struct request
*blk_get_request(struct request_queue
*, int, gfp_t
);
922 extern void blk_requeue_request(struct request_queue
*, struct request
*);
923 extern int blk_lld_busy(struct request_queue
*q
);
924 extern int blk_rq_prep_clone(struct request
*rq
, struct request
*rq_src
,
925 struct bio_set
*bs
, gfp_t gfp_mask
,
926 int (*bio_ctr
)(struct bio
*, struct bio
*, void *),
928 extern void blk_rq_unprep_clone(struct request
*rq
);
929 extern int blk_insert_cloned_request(struct request_queue
*q
,
931 extern int blk_rq_append_bio(struct request
*rq
, struct bio
*bio
);
932 extern void blk_delay_queue(struct request_queue
*, unsigned long);
933 extern void blk_queue_split(struct request_queue
*, struct bio
**,
935 extern void blk_recount_segments(struct request_queue
*, struct bio
*);
936 extern int scsi_verify_blk_ioctl(struct block_device
*, unsigned int);
937 extern int scsi_cmd_blk_ioctl(struct block_device
*, fmode_t
,
938 unsigned int, void __user
*);
939 extern int scsi_cmd_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
940 unsigned int, void __user
*);
941 extern int sg_scsi_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
942 struct scsi_ioctl_command __user
*);
944 extern int blk_queue_enter(struct request_queue
*q
, bool nowait
);
945 extern void blk_queue_exit(struct request_queue
*q
);
946 extern void blk_start_queue(struct request_queue
*q
);
947 extern void blk_start_queue_async(struct request_queue
*q
);
948 extern void blk_stop_queue(struct request_queue
*q
);
949 extern void blk_sync_queue(struct request_queue
*q
);
950 extern void __blk_stop_queue(struct request_queue
*q
);
951 extern void __blk_run_queue(struct request_queue
*q
);
952 extern void __blk_run_queue_uncond(struct request_queue
*q
);
953 extern void blk_run_queue(struct request_queue
*);
954 extern void blk_run_queue_async(struct request_queue
*q
);
955 extern void blk_mq_quiesce_queue(struct request_queue
*q
);
956 extern int blk_rq_map_user(struct request_queue
*, struct request
*,
957 struct rq_map_data
*, void __user
*, unsigned long,
959 extern int blk_rq_unmap_user(struct bio
*);
960 extern int blk_rq_map_kern(struct request_queue
*, struct request
*, void *, unsigned int, gfp_t
);
961 extern int blk_rq_map_user_iov(struct request_queue
*, struct request
*,
962 struct rq_map_data
*, const struct iov_iter
*,
964 extern int blk_execute_rq(struct request_queue
*, struct gendisk
*,
965 struct request
*, int);
966 extern void blk_execute_rq_nowait(struct request_queue
*, struct gendisk
*,
967 struct request
*, int, rq_end_io_fn
*);
969 bool blk_mq_poll(struct request_queue
*q
, blk_qc_t cookie
);
971 static inline struct request_queue
*bdev_get_queue(struct block_device
*bdev
)
973 return bdev
->bd_disk
->queue
; /* this is never NULL */
977 * blk_rq_pos() : the current sector
978 * blk_rq_bytes() : bytes left in the entire request
979 * blk_rq_cur_bytes() : bytes left in the current segment
980 * blk_rq_err_bytes() : bytes left till the next error boundary
981 * blk_rq_sectors() : sectors left in the entire request
982 * blk_rq_cur_sectors() : sectors left in the current segment
984 static inline sector_t
blk_rq_pos(const struct request
*rq
)
989 static inline unsigned int blk_rq_bytes(const struct request
*rq
)
991 return rq
->__data_len
;
994 static inline int blk_rq_cur_bytes(const struct request
*rq
)
996 return rq
->bio
? bio_cur_bytes(rq
->bio
) : 0;
999 extern unsigned int blk_rq_err_bytes(const struct request
*rq
);
1001 static inline unsigned int blk_rq_sectors(const struct request
*rq
)
1003 return blk_rq_bytes(rq
) >> 9;
1006 static inline unsigned int blk_rq_cur_sectors(const struct request
*rq
)
1008 return blk_rq_cur_bytes(rq
) >> 9;
1012 * Some commands like WRITE SAME have a payload or data transfer size which
1013 * is different from the size of the request. Any driver that supports such
1014 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1015 * calculate the data transfer size.
1017 static inline unsigned int blk_rq_payload_bytes(struct request
*rq
)
1019 if (rq
->rq_flags
& RQF_SPECIAL_PAYLOAD
)
1020 return rq
->special_vec
.bv_len
;
1021 return blk_rq_bytes(rq
);
1024 static inline unsigned int blk_queue_get_max_sectors(struct request_queue
*q
,
1027 if (unlikely(op
== REQ_OP_DISCARD
|| op
== REQ_OP_SECURE_ERASE
))
1028 return min(q
->limits
.max_discard_sectors
, UINT_MAX
>> 9);
1030 if (unlikely(op
== REQ_OP_WRITE_SAME
))
1031 return q
->limits
.max_write_same_sectors
;
1033 if (unlikely(op
== REQ_OP_WRITE_ZEROES
))
1034 return q
->limits
.max_write_zeroes_sectors
;
1036 return q
->limits
.max_sectors
;
1040 * Return maximum size of a request at given offset. Only valid for
1041 * file system requests.
1043 static inline unsigned int blk_max_size_offset(struct request_queue
*q
,
1046 if (!q
->limits
.chunk_sectors
)
1047 return q
->limits
.max_sectors
;
1049 return q
->limits
.chunk_sectors
-
1050 (offset
& (q
->limits
.chunk_sectors
- 1));
1053 static inline unsigned int blk_rq_get_max_sectors(struct request
*rq
,
1056 struct request_queue
*q
= rq
->q
;
1058 if (blk_rq_is_passthrough(rq
))
1059 return q
->limits
.max_hw_sectors
;
1061 if (!q
->limits
.chunk_sectors
||
1062 req_op(rq
) == REQ_OP_DISCARD
||
1063 req_op(rq
) == REQ_OP_SECURE_ERASE
)
1064 return blk_queue_get_max_sectors(q
, req_op(rq
));
1066 return min(blk_max_size_offset(q
, offset
),
1067 blk_queue_get_max_sectors(q
, req_op(rq
)));
1070 static inline unsigned int blk_rq_count_bios(struct request
*rq
)
1072 unsigned int nr_bios
= 0;
1075 __rq_for_each_bio(bio
, rq
)
1082 * blk_rq_set_prio - associate a request with prio from ioc
1083 * @rq: request of interest
1084 * @ioc: target iocontext
1086 * Assocate request prio with ioc prio so request based drivers
1087 * can leverage priority information.
1089 static inline void blk_rq_set_prio(struct request
*rq
, struct io_context
*ioc
)
1092 rq
->ioprio
= ioc
->ioprio
;
1096 * Request issue related functions.
1098 extern struct request
*blk_peek_request(struct request_queue
*q
);
1099 extern void blk_start_request(struct request
*rq
);
1100 extern struct request
*blk_fetch_request(struct request_queue
*q
);
1103 * Request completion related functions.
1105 * blk_update_request() completes given number of bytes and updates
1106 * the request without completing it.
1108 * blk_end_request() and friends. __blk_end_request() must be called
1109 * with the request queue spinlock acquired.
1111 * Several drivers define their own end_request and call
1112 * blk_end_request() for parts of the original function.
1113 * This prevents code duplication in drivers.
1115 extern bool blk_update_request(struct request
*rq
, int error
,
1116 unsigned int nr_bytes
);
1117 extern void blk_finish_request(struct request
*rq
, int error
);
1118 extern bool blk_end_request(struct request
*rq
, int error
,
1119 unsigned int nr_bytes
);
1120 extern void blk_end_request_all(struct request
*rq
, int error
);
1121 extern bool blk_end_request_cur(struct request
*rq
, int error
);
1122 extern bool blk_end_request_err(struct request
*rq
, int error
);
1123 extern bool __blk_end_request(struct request
*rq
, int error
,
1124 unsigned int nr_bytes
);
1125 extern void __blk_end_request_all(struct request
*rq
, int error
);
1126 extern bool __blk_end_request_cur(struct request
*rq
, int error
);
1127 extern bool __blk_end_request_err(struct request
*rq
, int error
);
1129 extern void blk_complete_request(struct request
*);
1130 extern void __blk_complete_request(struct request
*);
1131 extern void blk_abort_request(struct request
*);
1132 extern void blk_unprep_request(struct request
*);
1135 * Access functions for manipulating queue properties
1137 extern struct request_queue
*blk_init_queue_node(request_fn_proc
*rfn
,
1138 spinlock_t
*lock
, int node_id
);
1139 extern struct request_queue
*blk_init_queue(request_fn_proc
*, spinlock_t
*);
1140 extern int blk_init_allocated_queue(struct request_queue
*);
1141 extern void blk_cleanup_queue(struct request_queue
*);
1142 extern void blk_queue_make_request(struct request_queue
*, make_request_fn
*);
1143 extern void blk_queue_bounce_limit(struct request_queue
*, u64
);
1144 extern void blk_queue_max_hw_sectors(struct request_queue
*, unsigned int);
1145 extern void blk_queue_chunk_sectors(struct request_queue
*, unsigned int);
1146 extern void blk_queue_max_segments(struct request_queue
*, unsigned short);
1147 extern void blk_queue_max_segment_size(struct request_queue
*, unsigned int);
1148 extern void blk_queue_max_discard_sectors(struct request_queue
*q
,
1149 unsigned int max_discard_sectors
);
1150 extern void blk_queue_max_write_same_sectors(struct request_queue
*q
,
1151 unsigned int max_write_same_sectors
);
1152 extern void blk_queue_max_write_zeroes_sectors(struct request_queue
*q
,
1153 unsigned int max_write_same_sectors
);
1154 extern void blk_queue_logical_block_size(struct request_queue
*, unsigned short);
1155 extern void blk_queue_physical_block_size(struct request_queue
*, unsigned int);
1156 extern void blk_queue_alignment_offset(struct request_queue
*q
,
1157 unsigned int alignment
);
1158 extern void blk_limits_io_min(struct queue_limits
*limits
, unsigned int min
);
1159 extern void blk_queue_io_min(struct request_queue
*q
, unsigned int min
);
1160 extern void blk_limits_io_opt(struct queue_limits
*limits
, unsigned int opt
);
1161 extern void blk_queue_io_opt(struct request_queue
*q
, unsigned int opt
);
1162 extern void blk_set_queue_depth(struct request_queue
*q
, unsigned int depth
);
1163 extern void blk_set_default_limits(struct queue_limits
*lim
);
1164 extern void blk_set_stacking_limits(struct queue_limits
*lim
);
1165 extern int blk_stack_limits(struct queue_limits
*t
, struct queue_limits
*b
,
1167 extern int bdev_stack_limits(struct queue_limits
*t
, struct block_device
*bdev
,
1169 extern void disk_stack_limits(struct gendisk
*disk
, struct block_device
*bdev
,
1171 extern void blk_queue_stack_limits(struct request_queue
*t
, struct request_queue
*b
);
1172 extern void blk_queue_dma_pad(struct request_queue
*, unsigned int);
1173 extern void blk_queue_update_dma_pad(struct request_queue
*, unsigned int);
1174 extern int blk_queue_dma_drain(struct request_queue
*q
,
1175 dma_drain_needed_fn
*dma_drain_needed
,
1176 void *buf
, unsigned int size
);
1177 extern void blk_queue_lld_busy(struct request_queue
*q
, lld_busy_fn
*fn
);
1178 extern void blk_queue_segment_boundary(struct request_queue
*, unsigned long);
1179 extern void blk_queue_virt_boundary(struct request_queue
*, unsigned long);
1180 extern void blk_queue_prep_rq(struct request_queue
*, prep_rq_fn
*pfn
);
1181 extern void blk_queue_unprep_rq(struct request_queue
*, unprep_rq_fn
*ufn
);
1182 extern void blk_queue_dma_alignment(struct request_queue
*, int);
1183 extern void blk_queue_update_dma_alignment(struct request_queue
*, int);
1184 extern void blk_queue_softirq_done(struct request_queue
*, softirq_done_fn
*);
1185 extern void blk_queue_rq_timed_out(struct request_queue
*, rq_timed_out_fn
*);
1186 extern void blk_queue_rq_timeout(struct request_queue
*, unsigned int);
1187 extern void blk_queue_flush_queueable(struct request_queue
*q
, bool queueable
);
1188 extern void blk_queue_write_cache(struct request_queue
*q
, bool enabled
, bool fua
);
1189 extern struct backing_dev_info
*blk_get_backing_dev_info(struct block_device
*bdev
);
1191 static inline unsigned short blk_rq_nr_phys_segments(struct request
*rq
)
1193 if (rq
->rq_flags
& RQF_SPECIAL_PAYLOAD
)
1195 return rq
->nr_phys_segments
;
1198 extern int blk_rq_map_sg(struct request_queue
*, struct request
*, struct scatterlist
*);
1199 extern void blk_dump_rq_flags(struct request
*, char *);
1200 extern long nr_blockdev_pages(void);
1202 bool __must_check
blk_get_queue(struct request_queue
*);
1203 struct request_queue
*blk_alloc_queue(gfp_t
);
1204 struct request_queue
*blk_alloc_queue_node(gfp_t
, int);
1205 extern void blk_put_queue(struct request_queue
*);
1206 extern void blk_set_queue_dying(struct request_queue
*);
1209 * block layer runtime pm functions
1212 extern void blk_pm_runtime_init(struct request_queue
*q
, struct device
*dev
);
1213 extern int blk_pre_runtime_suspend(struct request_queue
*q
);
1214 extern void blk_post_runtime_suspend(struct request_queue
*q
, int err
);
1215 extern void blk_pre_runtime_resume(struct request_queue
*q
);
1216 extern void blk_post_runtime_resume(struct request_queue
*q
, int err
);
1217 extern void blk_set_runtime_active(struct request_queue
*q
);
1219 static inline void blk_pm_runtime_init(struct request_queue
*q
,
1220 struct device
*dev
) {}
1221 static inline int blk_pre_runtime_suspend(struct request_queue
*q
)
1225 static inline void blk_post_runtime_suspend(struct request_queue
*q
, int err
) {}
1226 static inline void blk_pre_runtime_resume(struct request_queue
*q
) {}
1227 static inline void blk_post_runtime_resume(struct request_queue
*q
, int err
) {}
1228 static inline void blk_set_runtime_active(struct request_queue
*q
) {}
1232 * blk_plug permits building a queue of related requests by holding the I/O
1233 * fragments for a short period. This allows merging of sequential requests
1234 * into single larger request. As the requests are moved from a per-task list to
1235 * the device's request_queue in a batch, this results in improved scalability
1236 * as the lock contention for request_queue lock is reduced.
1238 * It is ok not to disable preemption when adding the request to the plug list
1239 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1240 * the plug list when the task sleeps by itself. For details, please see
1241 * schedule() where blk_schedule_flush_plug() is called.
1244 struct list_head list
; /* requests */
1245 struct list_head mq_list
; /* blk-mq requests */
1246 struct list_head cb_list
; /* md requires an unplug callback */
1248 #define BLK_MAX_REQUEST_COUNT 16
1249 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1252 typedef void (*blk_plug_cb_fn
)(struct blk_plug_cb
*, bool);
1253 struct blk_plug_cb
{
1254 struct list_head list
;
1255 blk_plug_cb_fn callback
;
1258 extern struct blk_plug_cb
*blk_check_plugged(blk_plug_cb_fn unplug
,
1259 void *data
, int size
);
1260 extern void blk_start_plug(struct blk_plug
*);
1261 extern void blk_finish_plug(struct blk_plug
*);
1262 extern void blk_flush_plug_list(struct blk_plug
*, bool);
1264 static inline void blk_flush_plug(struct task_struct
*tsk
)
1266 struct blk_plug
*plug
= tsk
->plug
;
1269 blk_flush_plug_list(plug
, false);
1272 static inline void blk_schedule_flush_plug(struct task_struct
*tsk
)
1274 struct blk_plug
*plug
= tsk
->plug
;
1277 blk_flush_plug_list(plug
, true);
1280 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
1282 struct blk_plug
*plug
= tsk
->plug
;
1285 (!list_empty(&plug
->list
) ||
1286 !list_empty(&plug
->mq_list
) ||
1287 !list_empty(&plug
->cb_list
));
1293 extern int blk_queue_start_tag(struct request_queue
*, struct request
*);
1294 extern struct request
*blk_queue_find_tag(struct request_queue
*, int);
1295 extern void blk_queue_end_tag(struct request_queue
*, struct request
*);
1296 extern int blk_queue_init_tags(struct request_queue
*, int, struct blk_queue_tag
*, int);
1297 extern void blk_queue_free_tags(struct request_queue
*);
1298 extern int blk_queue_resize_tags(struct request_queue
*, int);
1299 extern void blk_queue_invalidate_tags(struct request_queue
*);
1300 extern struct blk_queue_tag
*blk_init_tags(int, int);
1301 extern void blk_free_tags(struct blk_queue_tag
*);
1303 static inline struct request
*blk_map_queue_find_tag(struct blk_queue_tag
*bqt
,
1306 if (unlikely(bqt
== NULL
|| tag
>= bqt
->real_max_depth
))
1308 return bqt
->tag_index
[tag
];
1312 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1313 #define BLKDEV_DISCARD_ZERO (1 << 1) /* must reliably zero data */
1315 extern int blkdev_issue_flush(struct block_device
*, gfp_t
, sector_t
*);
1316 extern int blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
1317 sector_t nr_sects
, gfp_t gfp_mask
, unsigned long flags
);
1318 extern int __blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
1319 sector_t nr_sects
, gfp_t gfp_mask
, int flags
,
1321 extern int blkdev_issue_write_same(struct block_device
*bdev
, sector_t sector
,
1322 sector_t nr_sects
, gfp_t gfp_mask
, struct page
*page
);
1323 extern int __blkdev_issue_zeroout(struct block_device
*bdev
, sector_t sector
,
1324 sector_t nr_sects
, gfp_t gfp_mask
, struct bio
**biop
,
1326 extern int blkdev_issue_zeroout(struct block_device
*bdev
, sector_t sector
,
1327 sector_t nr_sects
, gfp_t gfp_mask
, bool discard
);
1328 static inline int sb_issue_discard(struct super_block
*sb
, sector_t block
,
1329 sector_t nr_blocks
, gfp_t gfp_mask
, unsigned long flags
)
1331 return blkdev_issue_discard(sb
->s_bdev
, block
<< (sb
->s_blocksize_bits
- 9),
1332 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
1335 static inline int sb_issue_zeroout(struct super_block
*sb
, sector_t block
,
1336 sector_t nr_blocks
, gfp_t gfp_mask
)
1338 return blkdev_issue_zeroout(sb
->s_bdev
,
1339 block
<< (sb
->s_blocksize_bits
- 9),
1340 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
1344 extern int blk_verify_command(unsigned char *cmd
, fmode_t has_write_perm
);
1346 enum blk_default_limits
{
1347 BLK_MAX_SEGMENTS
= 128,
1348 BLK_SAFE_MAX_SECTORS
= 255,
1349 BLK_DEF_MAX_SECTORS
= 2560,
1350 BLK_MAX_SEGMENT_SIZE
= 65536,
1351 BLK_SEG_BOUNDARY_MASK
= 0xFFFFFFFFUL
,
1354 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
1356 static inline unsigned long queue_bounce_pfn(struct request_queue
*q
)
1358 return q
->limits
.bounce_pfn
;
1361 static inline unsigned long queue_segment_boundary(struct request_queue
*q
)
1363 return q
->limits
.seg_boundary_mask
;
1366 static inline unsigned long queue_virt_boundary(struct request_queue
*q
)
1368 return q
->limits
.virt_boundary_mask
;
1371 static inline unsigned int queue_max_sectors(struct request_queue
*q
)
1373 return q
->limits
.max_sectors
;
1376 static inline unsigned int queue_max_hw_sectors(struct request_queue
*q
)
1378 return q
->limits
.max_hw_sectors
;
1381 static inline unsigned short queue_max_segments(struct request_queue
*q
)
1383 return q
->limits
.max_segments
;
1386 static inline unsigned int queue_max_segment_size(struct request_queue
*q
)
1388 return q
->limits
.max_segment_size
;
1391 static inline unsigned short queue_logical_block_size(struct request_queue
*q
)
1395 if (q
&& q
->limits
.logical_block_size
)
1396 retval
= q
->limits
.logical_block_size
;
1401 static inline unsigned short bdev_logical_block_size(struct block_device
*bdev
)
1403 return queue_logical_block_size(bdev_get_queue(bdev
));
1406 static inline unsigned int queue_physical_block_size(struct request_queue
*q
)
1408 return q
->limits
.physical_block_size
;
1411 static inline unsigned int bdev_physical_block_size(struct block_device
*bdev
)
1413 return queue_physical_block_size(bdev_get_queue(bdev
));
1416 static inline unsigned int queue_io_min(struct request_queue
*q
)
1418 return q
->limits
.io_min
;
1421 static inline int bdev_io_min(struct block_device
*bdev
)
1423 return queue_io_min(bdev_get_queue(bdev
));
1426 static inline unsigned int queue_io_opt(struct request_queue
*q
)
1428 return q
->limits
.io_opt
;
1431 static inline int bdev_io_opt(struct block_device
*bdev
)
1433 return queue_io_opt(bdev_get_queue(bdev
));
1436 static inline int queue_alignment_offset(struct request_queue
*q
)
1438 if (q
->limits
.misaligned
)
1441 return q
->limits
.alignment_offset
;
1444 static inline int queue_limit_alignment_offset(struct queue_limits
*lim
, sector_t sector
)
1446 unsigned int granularity
= max(lim
->physical_block_size
, lim
->io_min
);
1447 unsigned int alignment
= sector_div(sector
, granularity
>> 9) << 9;
1449 return (granularity
+ lim
->alignment_offset
- alignment
) % granularity
;
1452 static inline int bdev_alignment_offset(struct block_device
*bdev
)
1454 struct request_queue
*q
= bdev_get_queue(bdev
);
1456 if (q
->limits
.misaligned
)
1459 if (bdev
!= bdev
->bd_contains
)
1460 return bdev
->bd_part
->alignment_offset
;
1462 return q
->limits
.alignment_offset
;
1465 static inline int queue_discard_alignment(struct request_queue
*q
)
1467 if (q
->limits
.discard_misaligned
)
1470 return q
->limits
.discard_alignment
;
1473 static inline int queue_limit_discard_alignment(struct queue_limits
*lim
, sector_t sector
)
1475 unsigned int alignment
, granularity
, offset
;
1477 if (!lim
->max_discard_sectors
)
1480 /* Why are these in bytes, not sectors? */
1481 alignment
= lim
->discard_alignment
>> 9;
1482 granularity
= lim
->discard_granularity
>> 9;
1486 /* Offset of the partition start in 'granularity' sectors */
1487 offset
= sector_div(sector
, granularity
);
1489 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1490 offset
= (granularity
+ alignment
- offset
) % granularity
;
1492 /* Turn it back into bytes, gaah */
1496 static inline int bdev_discard_alignment(struct block_device
*bdev
)
1498 struct request_queue
*q
= bdev_get_queue(bdev
);
1500 if (bdev
!= bdev
->bd_contains
)
1501 return bdev
->bd_part
->discard_alignment
;
1503 return q
->limits
.discard_alignment
;
1506 static inline unsigned int queue_discard_zeroes_data(struct request_queue
*q
)
1508 if (q
->limits
.max_discard_sectors
&& q
->limits
.discard_zeroes_data
== 1)
1514 static inline unsigned int bdev_discard_zeroes_data(struct block_device
*bdev
)
1516 return queue_discard_zeroes_data(bdev_get_queue(bdev
));
1519 static inline unsigned int bdev_write_same(struct block_device
*bdev
)
1521 struct request_queue
*q
= bdev_get_queue(bdev
);
1524 return q
->limits
.max_write_same_sectors
;
1529 static inline unsigned int bdev_write_zeroes_sectors(struct block_device
*bdev
)
1531 struct request_queue
*q
= bdev_get_queue(bdev
);
1534 return q
->limits
.max_write_zeroes_sectors
;
1539 static inline enum blk_zoned_model
bdev_zoned_model(struct block_device
*bdev
)
1541 struct request_queue
*q
= bdev_get_queue(bdev
);
1544 return blk_queue_zoned_model(q
);
1546 return BLK_ZONED_NONE
;
1549 static inline bool bdev_is_zoned(struct block_device
*bdev
)
1551 struct request_queue
*q
= bdev_get_queue(bdev
);
1554 return blk_queue_is_zoned(q
);
1559 static inline unsigned int bdev_zone_sectors(struct block_device
*bdev
)
1561 struct request_queue
*q
= bdev_get_queue(bdev
);
1564 return blk_queue_zone_sectors(q
);
1569 static inline int queue_dma_alignment(struct request_queue
*q
)
1571 return q
? q
->dma_alignment
: 511;
1574 static inline int blk_rq_aligned(struct request_queue
*q
, unsigned long addr
,
1577 unsigned int alignment
= queue_dma_alignment(q
) | q
->dma_pad_mask
;
1578 return !(addr
& alignment
) && !(len
& alignment
);
1581 /* assumes size > 256 */
1582 static inline unsigned int blksize_bits(unsigned int size
)
1584 unsigned int bits
= 8;
1588 } while (size
> 256);
1592 static inline unsigned int block_size(struct block_device
*bdev
)
1594 return bdev
->bd_block_size
;
1597 static inline bool queue_flush_queueable(struct request_queue
*q
)
1599 return !test_bit(QUEUE_FLAG_FLUSH_NQ
, &q
->queue_flags
);
1602 typedef struct {struct page
*v
;} Sector
;
1604 unsigned char *read_dev_sector(struct block_device
*, sector_t
, Sector
*);
1606 static inline void put_dev_sector(Sector p
)
1611 static inline bool __bvec_gap_to_prev(struct request_queue
*q
,
1612 struct bio_vec
*bprv
, unsigned int offset
)
1615 ((bprv
->bv_offset
+ bprv
->bv_len
) & queue_virt_boundary(q
));
1619 * Check if adding a bio_vec after bprv with offset would create a gap in
1620 * the SG list. Most drivers don't care about this, but some do.
1622 static inline bool bvec_gap_to_prev(struct request_queue
*q
,
1623 struct bio_vec
*bprv
, unsigned int offset
)
1625 if (!queue_virt_boundary(q
))
1627 return __bvec_gap_to_prev(q
, bprv
, offset
);
1631 * Check if the two bvecs from two bios can be merged to one segment.
1632 * If yes, no need to check gap between the two bios since the 1st bio
1633 * and the 1st bvec in the 2nd bio can be handled in one segment.
1635 static inline bool bios_segs_mergeable(struct request_queue
*q
,
1636 struct bio
*prev
, struct bio_vec
*prev_last_bv
,
1637 struct bio_vec
*next_first_bv
)
1639 if (!BIOVEC_PHYS_MERGEABLE(prev_last_bv
, next_first_bv
))
1641 if (!BIOVEC_SEG_BOUNDARY(q
, prev_last_bv
, next_first_bv
))
1643 if (prev
->bi_seg_back_size
+ next_first_bv
->bv_len
>
1644 queue_max_segment_size(q
))
1649 static inline bool bio_will_gap(struct request_queue
*q
, struct bio
*prev
,
1652 if (bio_has_data(prev
) && queue_virt_boundary(q
)) {
1653 struct bio_vec pb
, nb
;
1655 bio_get_last_bvec(prev
, &pb
);
1656 bio_get_first_bvec(next
, &nb
);
1658 if (!bios_segs_mergeable(q
, prev
, &pb
, &nb
))
1659 return __bvec_gap_to_prev(q
, &pb
, nb
.bv_offset
);
1665 static inline bool req_gap_back_merge(struct request
*req
, struct bio
*bio
)
1667 return bio_will_gap(req
->q
, req
->biotail
, bio
);
1670 static inline bool req_gap_front_merge(struct request
*req
, struct bio
*bio
)
1672 return bio_will_gap(req
->q
, bio
, req
->bio
);
1675 int kblockd_schedule_work(struct work_struct
*work
);
1676 int kblockd_schedule_work_on(int cpu
, struct work_struct
*work
);
1677 int kblockd_schedule_delayed_work(struct delayed_work
*dwork
, unsigned long delay
);
1678 int kblockd_schedule_delayed_work_on(int cpu
, struct delayed_work
*dwork
, unsigned long delay
);
1680 #ifdef CONFIG_BLK_CGROUP
1682 * This should not be using sched_clock(). A real patch is in progress
1683 * to fix this up, until that is in place we need to disable preemption
1684 * around sched_clock() in this function and set_io_start_time_ns().
1686 static inline void set_start_time_ns(struct request
*req
)
1689 req
->start_time_ns
= sched_clock();
1693 static inline void set_io_start_time_ns(struct request
*req
)
1696 req
->io_start_time_ns
= sched_clock();
1700 static inline uint64_t rq_start_time_ns(struct request
*req
)
1702 return req
->start_time_ns
;
1705 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1707 return req
->io_start_time_ns
;
1710 static inline void set_start_time_ns(struct request
*req
) {}
1711 static inline void set_io_start_time_ns(struct request
*req
) {}
1712 static inline uint64_t rq_start_time_ns(struct request
*req
)
1716 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1722 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1723 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1724 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1725 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1727 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1729 enum blk_integrity_flags
{
1730 BLK_INTEGRITY_VERIFY
= 1 << 0,
1731 BLK_INTEGRITY_GENERATE
= 1 << 1,
1732 BLK_INTEGRITY_DEVICE_CAPABLE
= 1 << 2,
1733 BLK_INTEGRITY_IP_CHECKSUM
= 1 << 3,
1736 struct blk_integrity_iter
{
1740 unsigned int data_size
;
1741 unsigned short interval
;
1742 const char *disk_name
;
1745 typedef int (integrity_processing_fn
) (struct blk_integrity_iter
*);
1747 struct blk_integrity_profile
{
1748 integrity_processing_fn
*generate_fn
;
1749 integrity_processing_fn
*verify_fn
;
1753 extern void blk_integrity_register(struct gendisk
*, struct blk_integrity
*);
1754 extern void blk_integrity_unregister(struct gendisk
*);
1755 extern int blk_integrity_compare(struct gendisk
*, struct gendisk
*);
1756 extern int blk_rq_map_integrity_sg(struct request_queue
*, struct bio
*,
1757 struct scatterlist
*);
1758 extern int blk_rq_count_integrity_sg(struct request_queue
*, struct bio
*);
1759 extern bool blk_integrity_merge_rq(struct request_queue
*, struct request
*,
1761 extern bool blk_integrity_merge_bio(struct request_queue
*, struct request
*,
1764 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1766 struct blk_integrity
*bi
= &disk
->queue
->integrity
;
1775 struct blk_integrity
*bdev_get_integrity(struct block_device
*bdev
)
1777 return blk_get_integrity(bdev
->bd_disk
);
1780 static inline bool blk_integrity_rq(struct request
*rq
)
1782 return rq
->cmd_flags
& REQ_INTEGRITY
;
1785 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1788 q
->limits
.max_integrity_segments
= segs
;
1791 static inline unsigned short
1792 queue_max_integrity_segments(struct request_queue
*q
)
1794 return q
->limits
.max_integrity_segments
;
1797 static inline bool integrity_req_gap_back_merge(struct request
*req
,
1800 struct bio_integrity_payload
*bip
= bio_integrity(req
->bio
);
1801 struct bio_integrity_payload
*bip_next
= bio_integrity(next
);
1803 return bvec_gap_to_prev(req
->q
, &bip
->bip_vec
[bip
->bip_vcnt
- 1],
1804 bip_next
->bip_vec
[0].bv_offset
);
1807 static inline bool integrity_req_gap_front_merge(struct request
*req
,
1810 struct bio_integrity_payload
*bip
= bio_integrity(bio
);
1811 struct bio_integrity_payload
*bip_next
= bio_integrity(req
->bio
);
1813 return bvec_gap_to_prev(req
->q
, &bip
->bip_vec
[bip
->bip_vcnt
- 1],
1814 bip_next
->bip_vec
[0].bv_offset
);
1817 #else /* CONFIG_BLK_DEV_INTEGRITY */
1820 struct block_device
;
1822 struct blk_integrity
;
1824 static inline int blk_integrity_rq(struct request
*rq
)
1828 static inline int blk_rq_count_integrity_sg(struct request_queue
*q
,
1833 static inline int blk_rq_map_integrity_sg(struct request_queue
*q
,
1835 struct scatterlist
*s
)
1839 static inline struct blk_integrity
*bdev_get_integrity(struct block_device
*b
)
1843 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1847 static inline int blk_integrity_compare(struct gendisk
*a
, struct gendisk
*b
)
1851 static inline void blk_integrity_register(struct gendisk
*d
,
1852 struct blk_integrity
*b
)
1855 static inline void blk_integrity_unregister(struct gendisk
*d
)
1858 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1862 static inline unsigned short queue_max_integrity_segments(struct request_queue
*q
)
1866 static inline bool blk_integrity_merge_rq(struct request_queue
*rq
,
1872 static inline bool blk_integrity_merge_bio(struct request_queue
*rq
,
1879 static inline bool integrity_req_gap_back_merge(struct request
*req
,
1884 static inline bool integrity_req_gap_front_merge(struct request
*req
,
1890 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1893 * struct blk_dax_ctl - control and output parameters for ->direct_access
1894 * @sector: (input) offset relative to a block_device
1895 * @addr: (output) kernel virtual address for @sector populated by driver
1896 * @pfn: (output) page frame number for @addr populated by driver
1897 * @size: (input) number of bytes requested
1899 struct blk_dax_ctl
{
1906 struct block_device_operations
{
1907 int (*open
) (struct block_device
*, fmode_t
);
1908 void (*release
) (struct gendisk
*, fmode_t
);
1909 int (*rw_page
)(struct block_device
*, sector_t
, struct page
*, bool);
1910 int (*ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1911 int (*compat_ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1912 long (*direct_access
)(struct block_device
*, sector_t
, void **, pfn_t
*,
1914 unsigned int (*check_events
) (struct gendisk
*disk
,
1915 unsigned int clearing
);
1916 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1917 int (*media_changed
) (struct gendisk
*);
1918 void (*unlock_native_capacity
) (struct gendisk
*);
1919 int (*revalidate_disk
) (struct gendisk
*);
1920 int (*getgeo
)(struct block_device
*, struct hd_geometry
*);
1921 /* this callback is with swap_lock and sometimes page table lock held */
1922 void (*swap_slot_free_notify
) (struct block_device
*, unsigned long);
1923 struct module
*owner
;
1924 const struct pr_ops
*pr_ops
;
1927 extern int __blkdev_driver_ioctl(struct block_device
*, fmode_t
, unsigned int,
1929 extern int bdev_read_page(struct block_device
*, sector_t
, struct page
*);
1930 extern int bdev_write_page(struct block_device
*, sector_t
, struct page
*,
1931 struct writeback_control
*);
1932 extern long bdev_direct_access(struct block_device
*, struct blk_dax_ctl
*);
1933 extern int bdev_dax_supported(struct super_block
*, int);
1934 extern bool bdev_dax_capable(struct block_device
*);
1935 #else /* CONFIG_BLOCK */
1937 struct block_device
;
1940 * stubs for when the block layer is configured out
1942 #define buffer_heads_over_limit 0
1944 static inline long nr_blockdev_pages(void)
1952 static inline void blk_start_plug(struct blk_plug
*plug
)
1956 static inline void blk_finish_plug(struct blk_plug
*plug
)
1960 static inline void blk_flush_plug(struct task_struct
*task
)
1964 static inline void blk_schedule_flush_plug(struct task_struct
*task
)
1969 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
1974 static inline int blkdev_issue_flush(struct block_device
*bdev
, gfp_t gfp_mask
,
1975 sector_t
*error_sector
)
1980 #endif /* CONFIG_BLOCK */