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
;
42 #define BLKDEV_MIN_RQ 4
43 #define BLKDEV_MAX_RQ 128 /* Default maximum */
46 * Maximum number of blkcg policies allowed to be registered concurrently.
47 * Defined here to simplify include dependency.
49 #define BLKCG_MAX_POLS 2
51 typedef void (rq_end_io_fn
)(struct request
*, int);
53 #define BLK_RL_SYNCFULL (1U << 0)
54 #define BLK_RL_ASYNCFULL (1U << 1)
57 struct request_queue
*q
; /* the queue this rl belongs to */
58 #ifdef CONFIG_BLK_CGROUP
59 struct blkcg_gq
*blkg
; /* blkg this request pool belongs to */
62 * count[], starved[], and wait[] are indexed by
63 * BLK_RW_SYNC/BLK_RW_ASYNC
68 wait_queue_head_t wait
[2];
73 * request command types
75 enum rq_cmd_type_bits
{
76 REQ_TYPE_FS
= 1, /* fs request */
77 REQ_TYPE_BLOCK_PC
, /* scsi command */
78 REQ_TYPE_DRV_PRIV
, /* driver defined types from here */
83 typedef __u32 __bitwise req_flags_t
;
85 /* elevator knows about this request */
86 #define RQF_SORTED ((__force req_flags_t)(1 << 0))
87 /* drive already may have started this one */
88 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
89 /* uses tagged queueing */
90 #define RQF_QUEUED ((__force req_flags_t)(1 << 2))
91 /* may not be passed by ioscheduler */
92 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
93 /* request for flush sequence */
94 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
95 /* merge of different types, fail separately */
96 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
97 /* track inflight for MQ */
98 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
99 /* don't call prep for this one */
100 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
101 /* set for "ide_preempt" requests and also for requests for which the SCSI
102 "quiesce" state must be ignored. */
103 #define RQF_PREEMPT ((__force req_flags_t)(1 << 8))
104 /* contains copies of user pages */
105 #define RQF_COPY_USER ((__force req_flags_t)(1 << 9))
106 /* vaguely specified driver internal error. Ignored by the block layer */
107 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
108 /* don't warn about errors */
109 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
110 /* elevator private data attached */
111 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
112 /* account I/O stat */
113 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
114 /* request came from our alloc pool */
115 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
116 /* runtime pm request */
117 #define RQF_PM ((__force req_flags_t)(1 << 15))
118 /* on IO scheduler merge hash */
119 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
120 /* IO stats tracking on */
121 #define RQF_STATS ((__force req_flags_t)(1 << 17))
123 /* flags that prevent us from merging requests: */
124 #define RQF_NOMERGE_FLAGS \
125 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ)
127 #define BLK_MAX_CDB 16
130 * Try to put the fields that are referenced together in the same cacheline.
132 * If you modify this structure, make sure to update blk_rq_init() and
133 * especially blk_mq_rq_ctx_init() to take care of the added fields.
136 struct list_head queuelist
;
138 struct call_single_data csd
;
142 struct request_queue
*q
;
143 struct blk_mq_ctx
*mq_ctx
;
147 unsigned int cmd_flags
; /* op and common flags */
148 req_flags_t rq_flags
;
149 unsigned long atomic_flags
;
151 /* the following two fields are internal, NEVER access directly */
152 unsigned int __data_len
; /* total data len */
153 sector_t __sector
; /* sector cursor */
159 * The hash is used inside the scheduler, and killed once the
160 * request reaches the dispatch list. The ipi_list is only used
161 * to queue the request for softirq completion, which is long
162 * after the request has been unhashed (and even removed from
163 * the dispatch list).
166 struct hlist_node hash
; /* merge hash */
167 struct list_head ipi_list
;
171 * The rb_node is only used inside the io scheduler, requests
172 * are pruned when moved to the dispatch queue. So let the
173 * completion_data share space with the rb_node.
176 struct rb_node rb_node
; /* sort/lookup */
177 void *completion_data
;
181 * Three pointers are available for the IO schedulers, if they need
182 * more they have to dynamically allocate it. Flush requests are
183 * never put on the IO scheduler. So let the flush fields share
184 * space with the elevator data.
194 struct list_head list
;
195 rq_end_io_fn
*saved_end_io
;
199 struct gendisk
*rq_disk
;
200 struct hd_struct
*part
;
201 unsigned long start_time
;
202 struct blk_issue_stat issue_stat
;
203 #ifdef CONFIG_BLK_CGROUP
204 struct request_list
*rl
; /* rl this rq is alloced from */
205 unsigned long long start_time_ns
;
206 unsigned long long io_start_time_ns
; /* when passed to hardware */
208 /* Number of scatter-gather DMA addr+len pairs after
209 * physical address coalescing is performed.
211 unsigned short nr_phys_segments
;
212 #if defined(CONFIG_BLK_DEV_INTEGRITY)
213 unsigned short nr_integrity_segments
;
216 unsigned short ioprio
;
218 void *special
; /* opaque pointer available for LLD use */
224 * when request is used as a packet command carrier
226 unsigned char __cmd
[BLK_MAX_CDB
];
228 unsigned short cmd_len
;
230 unsigned int extra_len
; /* length of alignment and padding */
231 unsigned int sense_len
;
232 unsigned int resid_len
; /* residual count */
235 unsigned long deadline
;
236 struct list_head timeout_list
;
237 unsigned int timeout
;
241 * completion callback.
243 rq_end_io_fn
*end_io
;
247 struct request
*next_rq
;
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
*);
270 enum blk_eh_timer_return
{
276 typedef enum blk_eh_timer_return (rq_timed_out_fn
)(struct request
*);
278 enum blk_queue_state
{
283 struct blk_queue_tag
{
284 struct request
**tag_index
; /* map of busy tags */
285 unsigned long *tag_map
; /* bit map of free/busy tags */
286 int busy
; /* current depth */
287 int max_depth
; /* what we will send to device */
288 int real_max_depth
; /* what the array can hold */
289 atomic_t refcnt
; /* map can be shared */
290 int alloc_policy
; /* tag allocation policy */
291 int next_tag
; /* next tag */
293 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
294 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
296 #define BLK_SCSI_MAX_CMDS (256)
297 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
300 * Zoned block device models (zoned limit).
302 enum blk_zoned_model
{
303 BLK_ZONED_NONE
, /* Regular block device */
304 BLK_ZONED_HA
, /* Host-aware zoned block device */
305 BLK_ZONED_HM
, /* Host-managed zoned block device */
308 struct queue_limits
{
309 unsigned long bounce_pfn
;
310 unsigned long seg_boundary_mask
;
311 unsigned long virt_boundary_mask
;
313 unsigned int max_hw_sectors
;
314 unsigned int max_dev_sectors
;
315 unsigned int chunk_sectors
;
316 unsigned int max_sectors
;
317 unsigned int max_segment_size
;
318 unsigned int physical_block_size
;
319 unsigned int alignment_offset
;
322 unsigned int max_discard_sectors
;
323 unsigned int max_hw_discard_sectors
;
324 unsigned int max_write_same_sectors
;
325 unsigned int discard_granularity
;
326 unsigned int discard_alignment
;
328 unsigned short logical_block_size
;
329 unsigned short max_segments
;
330 unsigned short max_integrity_segments
;
332 unsigned char misaligned
;
333 unsigned char discard_misaligned
;
334 unsigned char cluster
;
335 unsigned char discard_zeroes_data
;
336 unsigned char raid_partial_stripes_expensive
;
337 enum blk_zoned_model zoned
;
340 #ifdef CONFIG_BLK_DEV_ZONED
342 struct blk_zone_report_hdr
{
343 unsigned int nr_zones
;
347 extern int blkdev_report_zones(struct block_device
*bdev
,
348 sector_t sector
, struct blk_zone
*zones
,
349 unsigned int *nr_zones
, gfp_t gfp_mask
);
350 extern int blkdev_reset_zones(struct block_device
*bdev
, sector_t sectors
,
351 sector_t nr_sectors
, gfp_t gfp_mask
);
353 extern int blkdev_report_zones_ioctl(struct block_device
*bdev
, fmode_t mode
,
354 unsigned int cmd
, unsigned long arg
);
355 extern int blkdev_reset_zones_ioctl(struct block_device
*bdev
, fmode_t mode
,
356 unsigned int cmd
, unsigned long arg
);
358 #else /* CONFIG_BLK_DEV_ZONED */
360 static inline int blkdev_report_zones_ioctl(struct block_device
*bdev
,
361 fmode_t mode
, unsigned int cmd
,
367 static inline int blkdev_reset_zones_ioctl(struct block_device
*bdev
,
368 fmode_t mode
, unsigned int cmd
,
374 #endif /* CONFIG_BLK_DEV_ZONED */
376 struct request_queue
{
378 * Together with queue_head for cacheline sharing
380 struct list_head queue_head
;
381 struct request
*last_merge
;
382 struct elevator_queue
*elevator
;
383 int nr_rqs
[2]; /* # allocated [a]sync rqs */
384 int nr_rqs_elvpriv
; /* # allocated rqs w/ elvpriv */
387 * If blkcg is not used, @q->root_rl serves all requests. If blkcg
388 * is used, root blkg allocates from @q->root_rl and all other
389 * blkgs from their own blkg->rl. Which one to use should be
390 * determined using bio_request_list().
392 struct request_list root_rl
;
394 request_fn_proc
*request_fn
;
395 make_request_fn
*make_request_fn
;
396 prep_rq_fn
*prep_rq_fn
;
397 unprep_rq_fn
*unprep_rq_fn
;
398 softirq_done_fn
*softirq_done_fn
;
399 rq_timed_out_fn
*rq_timed_out_fn
;
400 dma_drain_needed_fn
*dma_drain_needed
;
401 lld_busy_fn
*lld_busy_fn
;
403 struct blk_mq_ops
*mq_ops
;
405 unsigned int *mq_map
;
408 struct blk_mq_ctx __percpu
*queue_ctx
;
409 unsigned int nr_queues
;
411 unsigned int queue_depth
;
413 /* hw dispatch queues */
414 struct blk_mq_hw_ctx
**queue_hw_ctx
;
415 unsigned int nr_hw_queues
;
418 * Dispatch queue sorting
421 struct request
*boundary_rq
;
424 * Delayed queue handling
426 struct delayed_work delay_work
;
428 struct backing_dev_info backing_dev_info
;
431 * The queue owner gets to use this for whatever they like.
432 * ll_rw_blk doesn't touch it.
437 * various queue flags, see QUEUE_* below
439 unsigned long queue_flags
;
442 * ida allocated id for this queue. Used to index queues from
448 * queue needs bounce pages for pages above this limit
453 * protects queue structures from reentrancy. ->__queue_lock should
454 * _never_ be used directly, it is queue private. always use
457 spinlock_t __queue_lock
;
458 spinlock_t
*queue_lock
;
468 struct kobject mq_kobj
;
470 #ifdef CONFIG_BLK_DEV_INTEGRITY
471 struct blk_integrity integrity
;
472 #endif /* CONFIG_BLK_DEV_INTEGRITY */
477 unsigned int nr_pending
;
483 unsigned long nr_requests
; /* Max # of requests */
484 unsigned int nr_congestion_on
;
485 unsigned int nr_congestion_off
;
486 unsigned int nr_batching
;
488 unsigned int dma_drain_size
;
489 void *dma_drain_buffer
;
490 unsigned int dma_pad_mask
;
491 unsigned int dma_alignment
;
493 struct blk_queue_tag
*queue_tags
;
494 struct list_head tag_busy_list
;
496 unsigned int nr_sorted
;
497 unsigned int in_flight
[2];
499 struct blk_rq_stat rq_stats
[2];
502 * Number of active block driver functions for which blk_drain_queue()
503 * must wait. Must be incremented around functions that unlock the
504 * queue_lock internally, e.g. scsi_request_fn().
506 unsigned int request_fn_active
;
508 unsigned int rq_timeout
;
509 struct timer_list timeout
;
510 struct work_struct timeout_work
;
511 struct list_head timeout_list
;
513 struct list_head icq_list
;
514 #ifdef CONFIG_BLK_CGROUP
515 DECLARE_BITMAP (blkcg_pols
, BLKCG_MAX_POLS
);
516 struct blkcg_gq
*root_blkg
;
517 struct list_head blkg_list
;
520 struct queue_limits limits
;
525 unsigned int sg_timeout
;
526 unsigned int sg_reserved_size
;
528 #ifdef CONFIG_BLK_DEV_IO_TRACE
529 struct blk_trace
*blk_trace
;
532 * for flush operations
534 struct blk_flush_queue
*fq
;
536 struct list_head requeue_list
;
537 spinlock_t requeue_lock
;
538 struct delayed_work requeue_work
;
540 struct mutex sysfs_lock
;
543 atomic_t mq_freeze_depth
;
545 #if defined(CONFIG_BLK_DEV_BSG)
546 bsg_job_fn
*bsg_job_fn
;
548 struct bsg_class_device bsg_dev
;
551 #ifdef CONFIG_BLK_DEV_THROTTLING
553 struct throtl_data
*td
;
555 struct rcu_head rcu_head
;
556 wait_queue_head_t mq_freeze_wq
;
557 struct percpu_ref q_usage_counter
;
558 struct list_head all_q_node
;
560 struct blk_mq_tag_set
*tag_set
;
561 struct list_head tag_set_list
;
562 struct bio_set
*bio_split
;
564 bool mq_sysfs_init_done
;
567 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
568 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
569 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
570 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
571 #define QUEUE_FLAG_DYING 5 /* queue being torn down */
572 #define QUEUE_FLAG_BYPASS 6 /* act as dumb FIFO queue */
573 #define QUEUE_FLAG_BIDI 7 /* queue supports bidi requests */
574 #define QUEUE_FLAG_NOMERGES 8 /* disable merge attempts */
575 #define QUEUE_FLAG_SAME_COMP 9 /* complete on same CPU-group */
576 #define QUEUE_FLAG_FAIL_IO 10 /* fake timeout */
577 #define QUEUE_FLAG_STACKABLE 11 /* supports request stacking */
578 #define QUEUE_FLAG_NONROT 12 /* non-rotational device (SSD) */
579 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
580 #define QUEUE_FLAG_IO_STAT 13 /* do IO stats */
581 #define QUEUE_FLAG_DISCARD 14 /* supports DISCARD */
582 #define QUEUE_FLAG_NOXMERGES 15 /* No extended merges */
583 #define QUEUE_FLAG_ADD_RANDOM 16 /* Contributes to random pool */
584 #define QUEUE_FLAG_SECERASE 17 /* supports secure erase */
585 #define QUEUE_FLAG_SAME_FORCE 18 /* force complete on same CPU */
586 #define QUEUE_FLAG_DEAD 19 /* queue tear-down finished */
587 #define QUEUE_FLAG_INIT_DONE 20 /* queue is initialized */
588 #define QUEUE_FLAG_NO_SG_MERGE 21 /* don't attempt to merge SG segments*/
589 #define QUEUE_FLAG_POLL 22 /* IO polling enabled if set */
590 #define QUEUE_FLAG_WC 23 /* Write back caching */
591 #define QUEUE_FLAG_FUA 24 /* device supports FUA writes */
592 #define QUEUE_FLAG_FLUSH_NQ 25 /* flush not queueuable */
593 #define QUEUE_FLAG_DAX 26 /* device supports DAX */
594 #define QUEUE_FLAG_STATS 27 /* track rq completion times */
596 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
597 (1 << QUEUE_FLAG_STACKABLE) | \
598 (1 << QUEUE_FLAG_SAME_COMP) | \
599 (1 << QUEUE_FLAG_ADD_RANDOM))
601 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
602 (1 << QUEUE_FLAG_STACKABLE) | \
603 (1 << QUEUE_FLAG_SAME_COMP) | \
604 (1 << QUEUE_FLAG_POLL))
606 static inline void queue_lockdep_assert_held(struct request_queue
*q
)
609 lockdep_assert_held(q
->queue_lock
);
612 static inline void queue_flag_set_unlocked(unsigned int flag
,
613 struct request_queue
*q
)
615 __set_bit(flag
, &q
->queue_flags
);
618 static inline int queue_flag_test_and_clear(unsigned int flag
,
619 struct request_queue
*q
)
621 queue_lockdep_assert_held(q
);
623 if (test_bit(flag
, &q
->queue_flags
)) {
624 __clear_bit(flag
, &q
->queue_flags
);
631 static inline int queue_flag_test_and_set(unsigned int flag
,
632 struct request_queue
*q
)
634 queue_lockdep_assert_held(q
);
636 if (!test_bit(flag
, &q
->queue_flags
)) {
637 __set_bit(flag
, &q
->queue_flags
);
644 static inline void queue_flag_set(unsigned int flag
, struct request_queue
*q
)
646 queue_lockdep_assert_held(q
);
647 __set_bit(flag
, &q
->queue_flags
);
650 static inline void queue_flag_clear_unlocked(unsigned int flag
,
651 struct request_queue
*q
)
653 __clear_bit(flag
, &q
->queue_flags
);
656 static inline int queue_in_flight(struct request_queue
*q
)
658 return q
->in_flight
[0] + q
->in_flight
[1];
661 static inline void queue_flag_clear(unsigned int flag
, struct request_queue
*q
)
663 queue_lockdep_assert_held(q
);
664 __clear_bit(flag
, &q
->queue_flags
);
667 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
668 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
669 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
670 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
671 #define blk_queue_bypass(q) test_bit(QUEUE_FLAG_BYPASS, &(q)->queue_flags)
672 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
673 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
674 #define blk_queue_noxmerges(q) \
675 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
676 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
677 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
678 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
679 #define blk_queue_stackable(q) \
680 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
681 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
682 #define blk_queue_secure_erase(q) \
683 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
684 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
686 #define blk_noretry_request(rq) \
687 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
688 REQ_FAILFAST_DRIVER))
690 #define blk_account_rq(rq) \
691 (((rq)->rq_flags & RQF_STARTED) && \
692 ((rq)->cmd_type == REQ_TYPE_FS))
694 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
695 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
696 /* rq->queuelist of dequeued request must be list_empty() */
697 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
699 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
701 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
704 * Driver can handle struct request, if it either has an old style
705 * request_fn defined, or is blk-mq based.
707 static inline bool queue_is_rq_based(struct request_queue
*q
)
709 return q
->request_fn
|| q
->mq_ops
;
712 static inline unsigned int blk_queue_cluster(struct request_queue
*q
)
714 return q
->limits
.cluster
;
717 static inline enum blk_zoned_model
718 blk_queue_zoned_model(struct request_queue
*q
)
720 return q
->limits
.zoned
;
723 static inline bool blk_queue_is_zoned(struct request_queue
*q
)
725 switch (blk_queue_zoned_model(q
)) {
734 static inline unsigned int blk_queue_zone_size(struct request_queue
*q
)
736 return blk_queue_is_zoned(q
) ? q
->limits
.chunk_sectors
: 0;
739 static inline bool rq_is_sync(struct request
*rq
)
741 return op_is_sync(rq
->cmd_flags
);
744 static inline bool blk_rl_full(struct request_list
*rl
, bool sync
)
746 unsigned int flag
= sync
? BLK_RL_SYNCFULL
: BLK_RL_ASYNCFULL
;
748 return rl
->flags
& flag
;
751 static inline void blk_set_rl_full(struct request_list
*rl
, bool sync
)
753 unsigned int flag
= sync
? BLK_RL_SYNCFULL
: BLK_RL_ASYNCFULL
;
758 static inline void blk_clear_rl_full(struct request_list
*rl
, bool sync
)
760 unsigned int flag
= sync
? BLK_RL_SYNCFULL
: BLK_RL_ASYNCFULL
;
765 static inline bool rq_mergeable(struct request
*rq
)
767 if (rq
->cmd_type
!= REQ_TYPE_FS
)
770 if (req_op(rq
) == REQ_OP_FLUSH
)
773 if (rq
->cmd_flags
& REQ_NOMERGE_FLAGS
)
775 if (rq
->rq_flags
& RQF_NOMERGE_FLAGS
)
781 static inline bool blk_write_same_mergeable(struct bio
*a
, struct bio
*b
)
783 if (bio_data(a
) == bio_data(b
))
789 static inline unsigned int blk_queue_depth(struct request_queue
*q
)
792 return q
->queue_depth
;
794 return q
->nr_requests
;
798 * q->prep_rq_fn return values
801 BLKPREP_OK
, /* serve it */
802 BLKPREP_KILL
, /* fatal error, kill, return -EIO */
803 BLKPREP_DEFER
, /* leave on queue */
804 BLKPREP_INVALID
, /* invalid command, kill, return -EREMOTEIO */
807 extern unsigned long blk_max_low_pfn
, blk_max_pfn
;
810 * standard bounce addresses:
812 * BLK_BOUNCE_HIGH : bounce all highmem pages
813 * BLK_BOUNCE_ANY : don't bounce anything
814 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
817 #if BITS_PER_LONG == 32
818 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
820 #define BLK_BOUNCE_HIGH -1ULL
822 #define BLK_BOUNCE_ANY (-1ULL)
823 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
826 * default timeout for SG_IO if none specified
828 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
829 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
832 extern int init_emergency_isa_pool(void);
833 extern void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
);
835 static inline int init_emergency_isa_pool(void)
839 static inline void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
)
842 #endif /* CONFIG_MMU */
848 unsigned long offset
;
853 struct req_iterator
{
854 struct bvec_iter iter
;
858 /* This should not be used directly - use rq_for_each_segment */
859 #define for_each_bio(_bio) \
860 for (; _bio; _bio = _bio->bi_next)
861 #define __rq_for_each_bio(_bio, rq) \
863 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
865 #define rq_for_each_segment(bvl, _rq, _iter) \
866 __rq_for_each_bio(_iter.bio, _rq) \
867 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
869 #define rq_iter_last(bvec, _iter) \
870 (_iter.bio->bi_next == NULL && \
871 bio_iter_last(bvec, _iter.iter))
873 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
874 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
876 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
877 extern void rq_flush_dcache_pages(struct request
*rq
);
879 static inline void rq_flush_dcache_pages(struct request
*rq
)
885 #define vfs_msg(sb, level, fmt, ...) \
886 __vfs_msg(sb, level, fmt, ##__VA_ARGS__)
888 #define vfs_msg(sb, level, fmt, ...) \
890 no_printk(fmt, ##__VA_ARGS__); \
891 __vfs_msg(sb, "", " "); \
895 extern int blk_register_queue(struct gendisk
*disk
);
896 extern void blk_unregister_queue(struct gendisk
*disk
);
897 extern blk_qc_t
generic_make_request(struct bio
*bio
);
898 extern void blk_rq_init(struct request_queue
*q
, struct request
*rq
);
899 extern void blk_put_request(struct request
*);
900 extern void __blk_put_request(struct request_queue
*, struct request
*);
901 extern struct request
*blk_get_request(struct request_queue
*, int, gfp_t
);
902 extern void blk_rq_set_block_pc(struct request
*);
903 extern void blk_requeue_request(struct request_queue
*, struct request
*);
904 extern void blk_add_request_payload(struct request
*rq
, struct page
*page
,
905 int offset
, unsigned int len
);
906 extern int blk_lld_busy(struct request_queue
*q
);
907 extern int blk_rq_prep_clone(struct request
*rq
, struct request
*rq_src
,
908 struct bio_set
*bs
, gfp_t gfp_mask
,
909 int (*bio_ctr
)(struct bio
*, struct bio
*, void *),
911 extern void blk_rq_unprep_clone(struct request
*rq
);
912 extern int blk_insert_cloned_request(struct request_queue
*q
,
914 extern int blk_rq_append_bio(struct request
*rq
, struct bio
*bio
);
915 extern void blk_delay_queue(struct request_queue
*, unsigned long);
916 extern void blk_queue_split(struct request_queue
*, struct bio
**,
918 extern void blk_recount_segments(struct request_queue
*, struct bio
*);
919 extern int scsi_verify_blk_ioctl(struct block_device
*, unsigned int);
920 extern int scsi_cmd_blk_ioctl(struct block_device
*, fmode_t
,
921 unsigned int, void __user
*);
922 extern int scsi_cmd_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
923 unsigned int, void __user
*);
924 extern int sg_scsi_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
925 struct scsi_ioctl_command __user
*);
927 extern int blk_queue_enter(struct request_queue
*q
, bool nowait
);
928 extern void blk_queue_exit(struct request_queue
*q
);
929 extern void blk_start_queue(struct request_queue
*q
);
930 extern void blk_start_queue_async(struct request_queue
*q
);
931 extern void blk_stop_queue(struct request_queue
*q
);
932 extern void blk_sync_queue(struct request_queue
*q
);
933 extern void __blk_stop_queue(struct request_queue
*q
);
934 extern void __blk_run_queue(struct request_queue
*q
);
935 extern void __blk_run_queue_uncond(struct request_queue
*q
);
936 extern void blk_run_queue(struct request_queue
*);
937 extern void blk_run_queue_async(struct request_queue
*q
);
938 extern void blk_mq_quiesce_queue(struct request_queue
*q
);
939 extern int blk_rq_map_user(struct request_queue
*, struct request
*,
940 struct rq_map_data
*, void __user
*, unsigned long,
942 extern int blk_rq_unmap_user(struct bio
*);
943 extern int blk_rq_map_kern(struct request_queue
*, struct request
*, void *, unsigned int, gfp_t
);
944 extern int blk_rq_map_user_iov(struct request_queue
*, struct request
*,
945 struct rq_map_data
*, const struct iov_iter
*,
947 extern int blk_execute_rq(struct request_queue
*, struct gendisk
*,
948 struct request
*, int);
949 extern void blk_execute_rq_nowait(struct request_queue
*, struct gendisk
*,
950 struct request
*, int, rq_end_io_fn
*);
952 bool blk_poll(struct request_queue
*q
, blk_qc_t cookie
);
954 static inline struct request_queue
*bdev_get_queue(struct block_device
*bdev
)
956 return bdev
->bd_disk
->queue
; /* this is never NULL */
960 * blk_rq_pos() : the current sector
961 * blk_rq_bytes() : bytes left in the entire request
962 * blk_rq_cur_bytes() : bytes left in the current segment
963 * blk_rq_err_bytes() : bytes left till the next error boundary
964 * blk_rq_sectors() : sectors left in the entire request
965 * blk_rq_cur_sectors() : sectors left in the current segment
967 static inline sector_t
blk_rq_pos(const struct request
*rq
)
972 static inline unsigned int blk_rq_bytes(const struct request
*rq
)
974 return rq
->__data_len
;
977 static inline int blk_rq_cur_bytes(const struct request
*rq
)
979 return rq
->bio
? bio_cur_bytes(rq
->bio
) : 0;
982 extern unsigned int blk_rq_err_bytes(const struct request
*rq
);
984 static inline unsigned int blk_rq_sectors(const struct request
*rq
)
986 return blk_rq_bytes(rq
) >> 9;
989 static inline unsigned int blk_rq_cur_sectors(const struct request
*rq
)
991 return blk_rq_cur_bytes(rq
) >> 9;
994 static inline unsigned int blk_queue_get_max_sectors(struct request_queue
*q
,
997 if (unlikely(op
== REQ_OP_DISCARD
|| op
== REQ_OP_SECURE_ERASE
))
998 return min(q
->limits
.max_discard_sectors
, UINT_MAX
>> 9);
1000 if (unlikely(op
== REQ_OP_WRITE_SAME
))
1001 return q
->limits
.max_write_same_sectors
;
1003 return q
->limits
.max_sectors
;
1007 * Return maximum size of a request at given offset. Only valid for
1008 * file system requests.
1010 static inline unsigned int blk_max_size_offset(struct request_queue
*q
,
1013 if (!q
->limits
.chunk_sectors
)
1014 return q
->limits
.max_sectors
;
1016 return q
->limits
.chunk_sectors
-
1017 (offset
& (q
->limits
.chunk_sectors
- 1));
1020 static inline unsigned int blk_rq_get_max_sectors(struct request
*rq
,
1023 struct request_queue
*q
= rq
->q
;
1025 if (unlikely(rq
->cmd_type
!= REQ_TYPE_FS
))
1026 return q
->limits
.max_hw_sectors
;
1028 if (!q
->limits
.chunk_sectors
||
1029 req_op(rq
) == REQ_OP_DISCARD
||
1030 req_op(rq
) == REQ_OP_SECURE_ERASE
)
1031 return blk_queue_get_max_sectors(q
, req_op(rq
));
1033 return min(blk_max_size_offset(q
, offset
),
1034 blk_queue_get_max_sectors(q
, req_op(rq
)));
1037 static inline unsigned int blk_rq_count_bios(struct request
*rq
)
1039 unsigned int nr_bios
= 0;
1042 __rq_for_each_bio(bio
, rq
)
1049 * Request issue related functions.
1051 extern struct request
*blk_peek_request(struct request_queue
*q
);
1052 extern void blk_start_request(struct request
*rq
);
1053 extern struct request
*blk_fetch_request(struct request_queue
*q
);
1056 * Request completion related functions.
1058 * blk_update_request() completes given number of bytes and updates
1059 * the request without completing it.
1061 * blk_end_request() and friends. __blk_end_request() must be called
1062 * with the request queue spinlock acquired.
1064 * Several drivers define their own end_request and call
1065 * blk_end_request() for parts of the original function.
1066 * This prevents code duplication in drivers.
1068 extern bool blk_update_request(struct request
*rq
, int error
,
1069 unsigned int nr_bytes
);
1070 extern void blk_finish_request(struct request
*rq
, int error
);
1071 extern bool blk_end_request(struct request
*rq
, int error
,
1072 unsigned int nr_bytes
);
1073 extern void blk_end_request_all(struct request
*rq
, int error
);
1074 extern bool blk_end_request_cur(struct request
*rq
, int error
);
1075 extern bool blk_end_request_err(struct request
*rq
, int error
);
1076 extern bool __blk_end_request(struct request
*rq
, int error
,
1077 unsigned int nr_bytes
);
1078 extern void __blk_end_request_all(struct request
*rq
, int error
);
1079 extern bool __blk_end_request_cur(struct request
*rq
, int error
);
1080 extern bool __blk_end_request_err(struct request
*rq
, int error
);
1082 extern void blk_complete_request(struct request
*);
1083 extern void __blk_complete_request(struct request
*);
1084 extern void blk_abort_request(struct request
*);
1085 extern void blk_unprep_request(struct request
*);
1088 * Access functions for manipulating queue properties
1090 extern struct request_queue
*blk_init_queue_node(request_fn_proc
*rfn
,
1091 spinlock_t
*lock
, int node_id
);
1092 extern struct request_queue
*blk_init_queue(request_fn_proc
*, spinlock_t
*);
1093 extern struct request_queue
*blk_init_allocated_queue(struct request_queue
*,
1094 request_fn_proc
*, spinlock_t
*);
1095 extern void blk_cleanup_queue(struct request_queue
*);
1096 extern void blk_queue_make_request(struct request_queue
*, make_request_fn
*);
1097 extern void blk_queue_bounce_limit(struct request_queue
*, u64
);
1098 extern void blk_queue_max_hw_sectors(struct request_queue
*, unsigned int);
1099 extern void blk_queue_chunk_sectors(struct request_queue
*, unsigned int);
1100 extern void blk_queue_max_segments(struct request_queue
*, unsigned short);
1101 extern void blk_queue_max_segment_size(struct request_queue
*, unsigned int);
1102 extern void blk_queue_max_discard_sectors(struct request_queue
*q
,
1103 unsigned int max_discard_sectors
);
1104 extern void blk_queue_max_write_same_sectors(struct request_queue
*q
,
1105 unsigned int max_write_same_sectors
);
1106 extern void blk_queue_logical_block_size(struct request_queue
*, unsigned short);
1107 extern void blk_queue_physical_block_size(struct request_queue
*, unsigned int);
1108 extern void blk_queue_alignment_offset(struct request_queue
*q
,
1109 unsigned int alignment
);
1110 extern void blk_limits_io_min(struct queue_limits
*limits
, unsigned int min
);
1111 extern void blk_queue_io_min(struct request_queue
*q
, unsigned int min
);
1112 extern void blk_limits_io_opt(struct queue_limits
*limits
, unsigned int opt
);
1113 extern void blk_queue_io_opt(struct request_queue
*q
, unsigned int opt
);
1114 extern void blk_set_queue_depth(struct request_queue
*q
, unsigned int depth
);
1115 extern void blk_set_default_limits(struct queue_limits
*lim
);
1116 extern void blk_set_stacking_limits(struct queue_limits
*lim
);
1117 extern int blk_stack_limits(struct queue_limits
*t
, struct queue_limits
*b
,
1119 extern int bdev_stack_limits(struct queue_limits
*t
, struct block_device
*bdev
,
1121 extern void disk_stack_limits(struct gendisk
*disk
, struct block_device
*bdev
,
1123 extern void blk_queue_stack_limits(struct request_queue
*t
, struct request_queue
*b
);
1124 extern void blk_queue_dma_pad(struct request_queue
*, unsigned int);
1125 extern void blk_queue_update_dma_pad(struct request_queue
*, unsigned int);
1126 extern int blk_queue_dma_drain(struct request_queue
*q
,
1127 dma_drain_needed_fn
*dma_drain_needed
,
1128 void *buf
, unsigned int size
);
1129 extern void blk_queue_lld_busy(struct request_queue
*q
, lld_busy_fn
*fn
);
1130 extern void blk_queue_segment_boundary(struct request_queue
*, unsigned long);
1131 extern void blk_queue_virt_boundary(struct request_queue
*, unsigned long);
1132 extern void blk_queue_prep_rq(struct request_queue
*, prep_rq_fn
*pfn
);
1133 extern void blk_queue_unprep_rq(struct request_queue
*, unprep_rq_fn
*ufn
);
1134 extern void blk_queue_dma_alignment(struct request_queue
*, int);
1135 extern void blk_queue_update_dma_alignment(struct request_queue
*, int);
1136 extern void blk_queue_softirq_done(struct request_queue
*, softirq_done_fn
*);
1137 extern void blk_queue_rq_timed_out(struct request_queue
*, rq_timed_out_fn
*);
1138 extern void blk_queue_rq_timeout(struct request_queue
*, unsigned int);
1139 extern void blk_queue_flush_queueable(struct request_queue
*q
, bool queueable
);
1140 extern void blk_queue_write_cache(struct request_queue
*q
, bool enabled
, bool fua
);
1141 extern struct backing_dev_info
*blk_get_backing_dev_info(struct block_device
*bdev
);
1143 extern int blk_rq_map_sg(struct request_queue
*, struct request
*, struct scatterlist
*);
1144 extern void blk_dump_rq_flags(struct request
*, char *);
1145 extern long nr_blockdev_pages(void);
1147 bool __must_check
blk_get_queue(struct request_queue
*);
1148 struct request_queue
*blk_alloc_queue(gfp_t
);
1149 struct request_queue
*blk_alloc_queue_node(gfp_t
, int);
1150 extern void blk_put_queue(struct request_queue
*);
1151 extern void blk_set_queue_dying(struct request_queue
*);
1154 * block layer runtime pm functions
1157 extern void blk_pm_runtime_init(struct request_queue
*q
, struct device
*dev
);
1158 extern int blk_pre_runtime_suspend(struct request_queue
*q
);
1159 extern void blk_post_runtime_suspend(struct request_queue
*q
, int err
);
1160 extern void blk_pre_runtime_resume(struct request_queue
*q
);
1161 extern void blk_post_runtime_resume(struct request_queue
*q
, int err
);
1162 extern void blk_set_runtime_active(struct request_queue
*q
);
1164 static inline void blk_pm_runtime_init(struct request_queue
*q
,
1165 struct device
*dev
) {}
1166 static inline int blk_pre_runtime_suspend(struct request_queue
*q
)
1170 static inline void blk_post_runtime_suspend(struct request_queue
*q
, int err
) {}
1171 static inline void blk_pre_runtime_resume(struct request_queue
*q
) {}
1172 static inline void blk_post_runtime_resume(struct request_queue
*q
, int err
) {}
1173 extern inline void blk_set_runtime_active(struct request_queue
*q
) {}
1177 * blk_plug permits building a queue of related requests by holding the I/O
1178 * fragments for a short period. This allows merging of sequential requests
1179 * into single larger request. As the requests are moved from a per-task list to
1180 * the device's request_queue in a batch, this results in improved scalability
1181 * as the lock contention for request_queue lock is reduced.
1183 * It is ok not to disable preemption when adding the request to the plug list
1184 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1185 * the plug list when the task sleeps by itself. For details, please see
1186 * schedule() where blk_schedule_flush_plug() is called.
1189 struct list_head list
; /* requests */
1190 struct list_head mq_list
; /* blk-mq requests */
1191 struct list_head cb_list
; /* md requires an unplug callback */
1193 #define BLK_MAX_REQUEST_COUNT 16
1194 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1197 typedef void (*blk_plug_cb_fn
)(struct blk_plug_cb
*, bool);
1198 struct blk_plug_cb
{
1199 struct list_head list
;
1200 blk_plug_cb_fn callback
;
1203 extern struct blk_plug_cb
*blk_check_plugged(blk_plug_cb_fn unplug
,
1204 void *data
, int size
);
1205 extern void blk_start_plug(struct blk_plug
*);
1206 extern void blk_finish_plug(struct blk_plug
*);
1207 extern void blk_flush_plug_list(struct blk_plug
*, bool);
1209 static inline void blk_flush_plug(struct task_struct
*tsk
)
1211 struct blk_plug
*plug
= tsk
->plug
;
1214 blk_flush_plug_list(plug
, false);
1217 static inline void blk_schedule_flush_plug(struct task_struct
*tsk
)
1219 struct blk_plug
*plug
= tsk
->plug
;
1222 blk_flush_plug_list(plug
, true);
1225 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
1227 struct blk_plug
*plug
= tsk
->plug
;
1230 (!list_empty(&plug
->list
) ||
1231 !list_empty(&plug
->mq_list
) ||
1232 !list_empty(&plug
->cb_list
));
1238 extern int blk_queue_start_tag(struct request_queue
*, struct request
*);
1239 extern struct request
*blk_queue_find_tag(struct request_queue
*, int);
1240 extern void blk_queue_end_tag(struct request_queue
*, struct request
*);
1241 extern int blk_queue_init_tags(struct request_queue
*, int, struct blk_queue_tag
*, int);
1242 extern void blk_queue_free_tags(struct request_queue
*);
1243 extern int blk_queue_resize_tags(struct request_queue
*, int);
1244 extern void blk_queue_invalidate_tags(struct request_queue
*);
1245 extern struct blk_queue_tag
*blk_init_tags(int, int);
1246 extern void blk_free_tags(struct blk_queue_tag
*);
1248 static inline struct request
*blk_map_queue_find_tag(struct blk_queue_tag
*bqt
,
1251 if (unlikely(bqt
== NULL
|| tag
>= bqt
->real_max_depth
))
1253 return bqt
->tag_index
[tag
];
1257 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1258 #define BLKDEV_DISCARD_ZERO (1 << 1) /* must reliably zero data */
1260 extern int blkdev_issue_flush(struct block_device
*, gfp_t
, sector_t
*);
1261 extern int blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
1262 sector_t nr_sects
, gfp_t gfp_mask
, unsigned long flags
);
1263 extern int __blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
1264 sector_t nr_sects
, gfp_t gfp_mask
, int flags
,
1266 extern int blkdev_issue_write_same(struct block_device
*bdev
, sector_t sector
,
1267 sector_t nr_sects
, gfp_t gfp_mask
, struct page
*page
);
1268 extern int blkdev_issue_zeroout(struct block_device
*bdev
, sector_t sector
,
1269 sector_t nr_sects
, gfp_t gfp_mask
, bool discard
);
1270 static inline int sb_issue_discard(struct super_block
*sb
, sector_t block
,
1271 sector_t nr_blocks
, gfp_t gfp_mask
, unsigned long flags
)
1273 return blkdev_issue_discard(sb
->s_bdev
, block
<< (sb
->s_blocksize_bits
- 9),
1274 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
1277 static inline int sb_issue_zeroout(struct super_block
*sb
, sector_t block
,
1278 sector_t nr_blocks
, gfp_t gfp_mask
)
1280 return blkdev_issue_zeroout(sb
->s_bdev
,
1281 block
<< (sb
->s_blocksize_bits
- 9),
1282 nr_blocks
<< (sb
->s_blocksize_bits
- 9),
1286 extern int blk_verify_command(unsigned char *cmd
, fmode_t has_write_perm
);
1288 enum blk_default_limits
{
1289 BLK_MAX_SEGMENTS
= 128,
1290 BLK_SAFE_MAX_SECTORS
= 255,
1291 BLK_DEF_MAX_SECTORS
= 2560,
1292 BLK_MAX_SEGMENT_SIZE
= 65536,
1293 BLK_SEG_BOUNDARY_MASK
= 0xFFFFFFFFUL
,
1296 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
1298 static inline unsigned long queue_bounce_pfn(struct request_queue
*q
)
1300 return q
->limits
.bounce_pfn
;
1303 static inline unsigned long queue_segment_boundary(struct request_queue
*q
)
1305 return q
->limits
.seg_boundary_mask
;
1308 static inline unsigned long queue_virt_boundary(struct request_queue
*q
)
1310 return q
->limits
.virt_boundary_mask
;
1313 static inline unsigned int queue_max_sectors(struct request_queue
*q
)
1315 return q
->limits
.max_sectors
;
1318 static inline unsigned int queue_max_hw_sectors(struct request_queue
*q
)
1320 return q
->limits
.max_hw_sectors
;
1323 static inline unsigned short queue_max_segments(struct request_queue
*q
)
1325 return q
->limits
.max_segments
;
1328 static inline unsigned int queue_max_segment_size(struct request_queue
*q
)
1330 return q
->limits
.max_segment_size
;
1333 static inline unsigned short queue_logical_block_size(struct request_queue
*q
)
1337 if (q
&& q
->limits
.logical_block_size
)
1338 retval
= q
->limits
.logical_block_size
;
1343 static inline unsigned short bdev_logical_block_size(struct block_device
*bdev
)
1345 return queue_logical_block_size(bdev_get_queue(bdev
));
1348 static inline unsigned int queue_physical_block_size(struct request_queue
*q
)
1350 return q
->limits
.physical_block_size
;
1353 static inline unsigned int bdev_physical_block_size(struct block_device
*bdev
)
1355 return queue_physical_block_size(bdev_get_queue(bdev
));
1358 static inline unsigned int queue_io_min(struct request_queue
*q
)
1360 return q
->limits
.io_min
;
1363 static inline int bdev_io_min(struct block_device
*bdev
)
1365 return queue_io_min(bdev_get_queue(bdev
));
1368 static inline unsigned int queue_io_opt(struct request_queue
*q
)
1370 return q
->limits
.io_opt
;
1373 static inline int bdev_io_opt(struct block_device
*bdev
)
1375 return queue_io_opt(bdev_get_queue(bdev
));
1378 static inline int queue_alignment_offset(struct request_queue
*q
)
1380 if (q
->limits
.misaligned
)
1383 return q
->limits
.alignment_offset
;
1386 static inline int queue_limit_alignment_offset(struct queue_limits
*lim
, sector_t sector
)
1388 unsigned int granularity
= max(lim
->physical_block_size
, lim
->io_min
);
1389 unsigned int alignment
= sector_div(sector
, granularity
>> 9) << 9;
1391 return (granularity
+ lim
->alignment_offset
- alignment
) % granularity
;
1394 static inline int bdev_alignment_offset(struct block_device
*bdev
)
1396 struct request_queue
*q
= bdev_get_queue(bdev
);
1398 if (q
->limits
.misaligned
)
1401 if (bdev
!= bdev
->bd_contains
)
1402 return bdev
->bd_part
->alignment_offset
;
1404 return q
->limits
.alignment_offset
;
1407 static inline int queue_discard_alignment(struct request_queue
*q
)
1409 if (q
->limits
.discard_misaligned
)
1412 return q
->limits
.discard_alignment
;
1415 static inline int queue_limit_discard_alignment(struct queue_limits
*lim
, sector_t sector
)
1417 unsigned int alignment
, granularity
, offset
;
1419 if (!lim
->max_discard_sectors
)
1422 /* Why are these in bytes, not sectors? */
1423 alignment
= lim
->discard_alignment
>> 9;
1424 granularity
= lim
->discard_granularity
>> 9;
1428 /* Offset of the partition start in 'granularity' sectors */
1429 offset
= sector_div(sector
, granularity
);
1431 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1432 offset
= (granularity
+ alignment
- offset
) % granularity
;
1434 /* Turn it back into bytes, gaah */
1438 static inline int bdev_discard_alignment(struct block_device
*bdev
)
1440 struct request_queue
*q
= bdev_get_queue(bdev
);
1442 if (bdev
!= bdev
->bd_contains
)
1443 return bdev
->bd_part
->discard_alignment
;
1445 return q
->limits
.discard_alignment
;
1448 static inline unsigned int queue_discard_zeroes_data(struct request_queue
*q
)
1450 if (q
->limits
.max_discard_sectors
&& q
->limits
.discard_zeroes_data
== 1)
1456 static inline unsigned int bdev_discard_zeroes_data(struct block_device
*bdev
)
1458 return queue_discard_zeroes_data(bdev_get_queue(bdev
));
1461 static inline unsigned int bdev_write_same(struct block_device
*bdev
)
1463 struct request_queue
*q
= bdev_get_queue(bdev
);
1466 return q
->limits
.max_write_same_sectors
;
1471 static inline enum blk_zoned_model
bdev_zoned_model(struct block_device
*bdev
)
1473 struct request_queue
*q
= bdev_get_queue(bdev
);
1476 return blk_queue_zoned_model(q
);
1478 return BLK_ZONED_NONE
;
1481 static inline bool bdev_is_zoned(struct block_device
*bdev
)
1483 struct request_queue
*q
= bdev_get_queue(bdev
);
1486 return blk_queue_is_zoned(q
);
1491 static inline unsigned int bdev_zone_size(struct block_device
*bdev
)
1493 struct request_queue
*q
= bdev_get_queue(bdev
);
1496 return blk_queue_zone_size(q
);
1501 static inline int queue_dma_alignment(struct request_queue
*q
)
1503 return q
? q
->dma_alignment
: 511;
1506 static inline int blk_rq_aligned(struct request_queue
*q
, unsigned long addr
,
1509 unsigned int alignment
= queue_dma_alignment(q
) | q
->dma_pad_mask
;
1510 return !(addr
& alignment
) && !(len
& alignment
);
1513 /* assumes size > 256 */
1514 static inline unsigned int blksize_bits(unsigned int size
)
1516 unsigned int bits
= 8;
1520 } while (size
> 256);
1524 static inline unsigned int block_size(struct block_device
*bdev
)
1526 return bdev
->bd_block_size
;
1529 static inline bool queue_flush_queueable(struct request_queue
*q
)
1531 return !test_bit(QUEUE_FLAG_FLUSH_NQ
, &q
->queue_flags
);
1534 typedef struct {struct page
*v
;} Sector
;
1536 unsigned char *read_dev_sector(struct block_device
*, sector_t
, Sector
*);
1538 static inline void put_dev_sector(Sector p
)
1543 static inline bool __bvec_gap_to_prev(struct request_queue
*q
,
1544 struct bio_vec
*bprv
, unsigned int offset
)
1547 ((bprv
->bv_offset
+ bprv
->bv_len
) & queue_virt_boundary(q
));
1551 * Check if adding a bio_vec after bprv with offset would create a gap in
1552 * the SG list. Most drivers don't care about this, but some do.
1554 static inline bool bvec_gap_to_prev(struct request_queue
*q
,
1555 struct bio_vec
*bprv
, unsigned int offset
)
1557 if (!queue_virt_boundary(q
))
1559 return __bvec_gap_to_prev(q
, bprv
, offset
);
1562 static inline bool bio_will_gap(struct request_queue
*q
, struct bio
*prev
,
1565 if (bio_has_data(prev
) && queue_virt_boundary(q
)) {
1566 struct bio_vec pb
, nb
;
1568 bio_get_last_bvec(prev
, &pb
);
1569 bio_get_first_bvec(next
, &nb
);
1571 return __bvec_gap_to_prev(q
, &pb
, nb
.bv_offset
);
1577 static inline bool req_gap_back_merge(struct request
*req
, struct bio
*bio
)
1579 return bio_will_gap(req
->q
, req
->biotail
, bio
);
1582 static inline bool req_gap_front_merge(struct request
*req
, struct bio
*bio
)
1584 return bio_will_gap(req
->q
, bio
, req
->bio
);
1587 int kblockd_schedule_work(struct work_struct
*work
);
1588 int kblockd_schedule_work_on(int cpu
, struct work_struct
*work
);
1589 int kblockd_schedule_delayed_work(struct delayed_work
*dwork
, unsigned long delay
);
1590 int kblockd_schedule_delayed_work_on(int cpu
, struct delayed_work
*dwork
, unsigned long delay
);
1592 #ifdef CONFIG_BLK_CGROUP
1594 * This should not be using sched_clock(). A real patch is in progress
1595 * to fix this up, until that is in place we need to disable preemption
1596 * around sched_clock() in this function and set_io_start_time_ns().
1598 static inline void set_start_time_ns(struct request
*req
)
1601 req
->start_time_ns
= sched_clock();
1605 static inline void set_io_start_time_ns(struct request
*req
)
1608 req
->io_start_time_ns
= sched_clock();
1612 static inline uint64_t rq_start_time_ns(struct request
*req
)
1614 return req
->start_time_ns
;
1617 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1619 return req
->io_start_time_ns
;
1622 static inline void set_start_time_ns(struct request
*req
) {}
1623 static inline void set_io_start_time_ns(struct request
*req
) {}
1624 static inline uint64_t rq_start_time_ns(struct request
*req
)
1628 static inline uint64_t rq_io_start_time_ns(struct request
*req
)
1634 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1635 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1636 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1637 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1639 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1641 enum blk_integrity_flags
{
1642 BLK_INTEGRITY_VERIFY
= 1 << 0,
1643 BLK_INTEGRITY_GENERATE
= 1 << 1,
1644 BLK_INTEGRITY_DEVICE_CAPABLE
= 1 << 2,
1645 BLK_INTEGRITY_IP_CHECKSUM
= 1 << 3,
1648 struct blk_integrity_iter
{
1652 unsigned int data_size
;
1653 unsigned short interval
;
1654 const char *disk_name
;
1657 typedef int (integrity_processing_fn
) (struct blk_integrity_iter
*);
1659 struct blk_integrity_profile
{
1660 integrity_processing_fn
*generate_fn
;
1661 integrity_processing_fn
*verify_fn
;
1665 extern void blk_integrity_register(struct gendisk
*, struct blk_integrity
*);
1666 extern void blk_integrity_unregister(struct gendisk
*);
1667 extern int blk_integrity_compare(struct gendisk
*, struct gendisk
*);
1668 extern int blk_rq_map_integrity_sg(struct request_queue
*, struct bio
*,
1669 struct scatterlist
*);
1670 extern int blk_rq_count_integrity_sg(struct request_queue
*, struct bio
*);
1671 extern bool blk_integrity_merge_rq(struct request_queue
*, struct request
*,
1673 extern bool blk_integrity_merge_bio(struct request_queue
*, struct request
*,
1676 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1678 struct blk_integrity
*bi
= &disk
->queue
->integrity
;
1687 struct blk_integrity
*bdev_get_integrity(struct block_device
*bdev
)
1689 return blk_get_integrity(bdev
->bd_disk
);
1692 static inline bool blk_integrity_rq(struct request
*rq
)
1694 return rq
->cmd_flags
& REQ_INTEGRITY
;
1697 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1700 q
->limits
.max_integrity_segments
= segs
;
1703 static inline unsigned short
1704 queue_max_integrity_segments(struct request_queue
*q
)
1706 return q
->limits
.max_integrity_segments
;
1709 static inline bool integrity_req_gap_back_merge(struct request
*req
,
1712 struct bio_integrity_payload
*bip
= bio_integrity(req
->bio
);
1713 struct bio_integrity_payload
*bip_next
= bio_integrity(next
);
1715 return bvec_gap_to_prev(req
->q
, &bip
->bip_vec
[bip
->bip_vcnt
- 1],
1716 bip_next
->bip_vec
[0].bv_offset
);
1719 static inline bool integrity_req_gap_front_merge(struct request
*req
,
1722 struct bio_integrity_payload
*bip
= bio_integrity(bio
);
1723 struct bio_integrity_payload
*bip_next
= bio_integrity(req
->bio
);
1725 return bvec_gap_to_prev(req
->q
, &bip
->bip_vec
[bip
->bip_vcnt
- 1],
1726 bip_next
->bip_vec
[0].bv_offset
);
1729 #else /* CONFIG_BLK_DEV_INTEGRITY */
1732 struct block_device
;
1734 struct blk_integrity
;
1736 static inline int blk_integrity_rq(struct request
*rq
)
1740 static inline int blk_rq_count_integrity_sg(struct request_queue
*q
,
1745 static inline int blk_rq_map_integrity_sg(struct request_queue
*q
,
1747 struct scatterlist
*s
)
1751 static inline struct blk_integrity
*bdev_get_integrity(struct block_device
*b
)
1755 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1759 static inline int blk_integrity_compare(struct gendisk
*a
, struct gendisk
*b
)
1763 static inline void blk_integrity_register(struct gendisk
*d
,
1764 struct blk_integrity
*b
)
1767 static inline void blk_integrity_unregister(struct gendisk
*d
)
1770 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1774 static inline unsigned short queue_max_integrity_segments(struct request_queue
*q
)
1778 static inline bool blk_integrity_merge_rq(struct request_queue
*rq
,
1784 static inline bool blk_integrity_merge_bio(struct request_queue
*rq
,
1791 static inline bool integrity_req_gap_back_merge(struct request
*req
,
1796 static inline bool integrity_req_gap_front_merge(struct request
*req
,
1802 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1805 * struct blk_dax_ctl - control and output parameters for ->direct_access
1806 * @sector: (input) offset relative to a block_device
1807 * @addr: (output) kernel virtual address for @sector populated by driver
1808 * @pfn: (output) page frame number for @addr populated by driver
1809 * @size: (input) number of bytes requested
1811 struct blk_dax_ctl
{
1818 struct block_device_operations
{
1819 int (*open
) (struct block_device
*, fmode_t
);
1820 void (*release
) (struct gendisk
*, fmode_t
);
1821 int (*rw_page
)(struct block_device
*, sector_t
, struct page
*, bool);
1822 int (*ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1823 int (*compat_ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1824 long (*direct_access
)(struct block_device
*, sector_t
, void **, pfn_t
*,
1826 unsigned int (*check_events
) (struct gendisk
*disk
,
1827 unsigned int clearing
);
1828 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1829 int (*media_changed
) (struct gendisk
*);
1830 void (*unlock_native_capacity
) (struct gendisk
*);
1831 int (*revalidate_disk
) (struct gendisk
*);
1832 int (*getgeo
)(struct block_device
*, struct hd_geometry
*);
1833 /* this callback is with swap_lock and sometimes page table lock held */
1834 void (*swap_slot_free_notify
) (struct block_device
*, unsigned long);
1835 struct module
*owner
;
1836 const struct pr_ops
*pr_ops
;
1839 extern int __blkdev_driver_ioctl(struct block_device
*, fmode_t
, unsigned int,
1841 extern int bdev_read_page(struct block_device
*, sector_t
, struct page
*);
1842 extern int bdev_write_page(struct block_device
*, sector_t
, struct page
*,
1843 struct writeback_control
*);
1844 extern long bdev_direct_access(struct block_device
*, struct blk_dax_ctl
*);
1845 extern int bdev_dax_supported(struct super_block
*, int);
1846 extern bool bdev_dax_capable(struct block_device
*);
1847 #else /* CONFIG_BLOCK */
1849 struct block_device
;
1852 * stubs for when the block layer is configured out
1854 #define buffer_heads_over_limit 0
1856 static inline long nr_blockdev_pages(void)
1864 static inline void blk_start_plug(struct blk_plug
*plug
)
1868 static inline void blk_finish_plug(struct blk_plug
*plug
)
1872 static inline void blk_flush_plug(struct task_struct
*task
)
1876 static inline void blk_schedule_flush_plug(struct task_struct
*task
)
1881 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
1886 static inline int blkdev_issue_flush(struct block_device
*bdev
, gfp_t gfp_mask
,
1887 sector_t
*error_sector
)
1892 #endif /* CONFIG_BLOCK */