1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef _LINUX_BLKDEV_H
3 #define _LINUX_BLKDEV_H
5 #include <linux/sched.h>
6 #include <linux/sched/clock.h>
10 #include <linux/major.h>
11 #include <linux/genhd.h>
12 #include <linux/list.h>
13 #include <linux/llist.h>
14 #include <linux/timer.h>
15 #include <linux/workqueue.h>
16 #include <linux/pagemap.h>
17 #include <linux/backing-dev-defs.h>
18 #include <linux/wait.h>
19 #include <linux/mempool.h>
20 #include <linux/pfn.h>
21 #include <linux/bio.h>
22 #include <linux/stringify.h>
23 #include <linux/gfp.h>
24 #include <linux/bsg.h>
25 #include <linux/smp.h>
26 #include <linux/rcupdate.h>
27 #include <linux/percpu-refcount.h>
28 #include <linux/scatterlist.h>
29 #include <linux/blkzoned.h>
32 struct scsi_ioctl_command
;
35 struct elevator_queue
;
41 struct blk_flush_queue
;
44 struct blk_queue_stats
;
45 struct blk_stat_callback
;
47 #define BLKDEV_MIN_RQ 4
48 #define BLKDEV_MAX_RQ 128 /* Default maximum */
50 /* Must be consistent with blk_mq_poll_stats_bkt() */
51 #define BLK_MQ_POLL_STATS_BKTS 16
54 * Maximum number of blkcg policies allowed to be registered concurrently.
55 * Defined here to simplify include dependency.
57 #define BLKCG_MAX_POLS 5
59 typedef void (rq_end_io_fn
)(struct request
*, blk_status_t
);
63 typedef __u32 __bitwise req_flags_t
;
65 /* elevator knows about this request */
66 #define RQF_SORTED ((__force req_flags_t)(1 << 0))
67 /* drive already may have started this one */
68 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
69 /* may not be passed by ioscheduler */
70 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
71 /* request for flush sequence */
72 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
73 /* merge of different types, fail separately */
74 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
75 /* track inflight for MQ */
76 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
77 /* don't call prep for this one */
78 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
79 /* set for "ide_preempt" requests and also for requests for which the SCSI
80 "quiesce" state must be ignored. */
81 #define RQF_PREEMPT ((__force req_flags_t)(1 << 8))
82 /* contains copies of user pages */
83 #define RQF_COPY_USER ((__force req_flags_t)(1 << 9))
84 /* vaguely specified driver internal error. Ignored by the block layer */
85 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
86 /* don't warn about errors */
87 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
88 /* elevator private data attached */
89 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
90 /* account into disk and partition IO statistics */
91 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
92 /* request came from our alloc pool */
93 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
94 /* runtime pm request */
95 #define RQF_PM ((__force req_flags_t)(1 << 15))
96 /* on IO scheduler merge hash */
97 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
98 /* track IO completion time */
99 #define RQF_STATS ((__force req_flags_t)(1 << 17))
100 /* Look at ->special_vec for the actual data payload instead of the
102 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
103 /* The per-zone write lock is held for this request */
104 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
105 /* already slept for hybrid poll */
106 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
107 /* ->timeout has been called, don't expire again */
108 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
110 /* flags that prevent us from merging requests: */
111 #define RQF_NOMERGE_FLAGS \
112 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
115 * Request state for blk-mq.
124 * Try to put the fields that are referenced together in the same cacheline.
126 * If you modify this structure, make sure to update blk_rq_init() and
127 * especially blk_mq_rq_ctx_init() to take care of the added fields.
130 struct request_queue
*q
;
131 struct blk_mq_ctx
*mq_ctx
;
132 struct blk_mq_hw_ctx
*mq_hctx
;
134 unsigned int cmd_flags
; /* op and common flags */
135 req_flags_t rq_flags
;
139 /* the following two fields are internal, NEVER access directly */
140 unsigned int __data_len
; /* total data len */
142 sector_t __sector
; /* sector cursor */
147 struct list_head queuelist
;
150 * The hash is used inside the scheduler, and killed once the
151 * request reaches the dispatch list. The ipi_list is only used
152 * to queue the request for softirq completion, which is long
153 * after the request has been unhashed (and even removed from
154 * the dispatch list).
157 struct hlist_node hash
; /* merge hash */
158 struct list_head ipi_list
;
162 * The rb_node is only used inside the io scheduler, requests
163 * are pruned when moved to the dispatch queue. So let the
164 * completion_data share space with the rb_node.
167 struct rb_node rb_node
; /* sort/lookup */
168 struct bio_vec special_vec
;
169 void *completion_data
;
170 int error_count
; /* for legacy drivers, don't use */
174 * Three pointers are available for the IO schedulers, if they need
175 * more they have to dynamically allocate it. Flush requests are
176 * never put on the IO scheduler. So let the flush fields share
177 * space with the elevator data.
187 struct list_head list
;
188 rq_end_io_fn
*saved_end_io
;
192 struct gendisk
*rq_disk
;
193 struct hd_struct
*part
;
194 /* Time that I/O was submitted to the kernel. */
196 /* Time that I/O was submitted to the device. */
197 u64 io_start_time_ns
;
199 #ifdef CONFIG_BLK_WBT
200 unsigned short wbt_flags
;
202 #ifdef CONFIG_BLK_DEV_THROTTLING_LOW
203 unsigned short throtl_size
;
207 * Number of scatter-gather DMA addr+len pairs after
208 * physical address coalescing is performed.
210 unsigned short nr_phys_segments
;
212 #if defined(CONFIG_BLK_DEV_INTEGRITY)
213 unsigned short nr_integrity_segments
;
216 unsigned short write_hint
;
217 unsigned short ioprio
;
219 unsigned int extra_len
; /* length of alignment and padding */
221 enum mq_rq_state state
;
224 unsigned int timeout
;
225 unsigned long deadline
;
228 struct __call_single_data csd
;
233 * completion callback.
235 rq_end_io_fn
*end_io
;
239 static inline bool blk_op_is_scsi(unsigned int op
)
241 return op
== REQ_OP_SCSI_IN
|| op
== REQ_OP_SCSI_OUT
;
244 static inline bool blk_op_is_private(unsigned int op
)
246 return op
== REQ_OP_DRV_IN
|| op
== REQ_OP_DRV_OUT
;
249 static inline bool blk_rq_is_scsi(struct request
*rq
)
251 return blk_op_is_scsi(req_op(rq
));
254 static inline bool blk_rq_is_private(struct request
*rq
)
256 return blk_op_is_private(req_op(rq
));
259 static inline bool blk_rq_is_passthrough(struct request
*rq
)
261 return blk_rq_is_scsi(rq
) || blk_rq_is_private(rq
);
264 static inline bool bio_is_passthrough(struct bio
*bio
)
266 unsigned op
= bio_op(bio
);
268 return blk_op_is_scsi(op
) || blk_op_is_private(op
);
271 static inline unsigned short req_get_ioprio(struct request
*req
)
276 #include <linux/elevator.h>
278 struct blk_queue_ctx
;
280 typedef blk_qc_t (make_request_fn
) (struct request_queue
*q
, struct bio
*bio
);
283 typedef int (dma_drain_needed_fn
)(struct request
*);
285 enum blk_eh_timer_return
{
286 BLK_EH_DONE
, /* drivers has completed the command */
287 BLK_EH_RESET_TIMER
, /* reset timer and try again */
290 enum blk_queue_state
{
295 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
296 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
298 #define BLK_SCSI_MAX_CMDS (256)
299 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
302 * Zoned block device models (zoned limit).
304 enum blk_zoned_model
{
305 BLK_ZONED_NONE
, /* Regular block device */
306 BLK_ZONED_HA
, /* Host-aware zoned block device */
307 BLK_ZONED_HM
, /* Host-managed zoned block device */
310 struct queue_limits
{
311 unsigned long bounce_pfn
;
312 unsigned long seg_boundary_mask
;
313 unsigned long virt_boundary_mask
;
315 unsigned int max_hw_sectors
;
316 unsigned int max_dev_sectors
;
317 unsigned int chunk_sectors
;
318 unsigned int max_sectors
;
319 unsigned int max_segment_size
;
320 unsigned int physical_block_size
;
321 unsigned int alignment_offset
;
324 unsigned int max_discard_sectors
;
325 unsigned int max_hw_discard_sectors
;
326 unsigned int max_write_same_sectors
;
327 unsigned int max_write_zeroes_sectors
;
328 unsigned int discard_granularity
;
329 unsigned int discard_alignment
;
331 unsigned short logical_block_size
;
332 unsigned short max_segments
;
333 unsigned short max_integrity_segments
;
334 unsigned short max_discard_segments
;
336 unsigned char misaligned
;
337 unsigned char discard_misaligned
;
338 unsigned char raid_partial_stripes_expensive
;
339 enum blk_zoned_model zoned
;
342 #ifdef CONFIG_BLK_DEV_ZONED
344 extern unsigned int blkdev_nr_zones(struct block_device
*bdev
);
345 extern int blkdev_report_zones(struct block_device
*bdev
,
346 sector_t sector
, struct blk_zone
*zones
,
347 unsigned int *nr_zones
, gfp_t gfp_mask
);
348 extern int blkdev_reset_zones(struct block_device
*bdev
, sector_t sectors
,
349 sector_t nr_sectors
, gfp_t gfp_mask
);
350 extern int blk_revalidate_disk_zones(struct gendisk
*disk
);
352 extern int blkdev_report_zones_ioctl(struct block_device
*bdev
, fmode_t mode
,
353 unsigned int cmd
, unsigned long arg
);
354 extern int blkdev_reset_zones_ioctl(struct block_device
*bdev
, fmode_t mode
,
355 unsigned int cmd
, unsigned long arg
);
357 #else /* CONFIG_BLK_DEV_ZONED */
359 static inline unsigned int blkdev_nr_zones(struct block_device
*bdev
)
364 static inline int blk_revalidate_disk_zones(struct gendisk
*disk
)
369 static inline int blkdev_report_zones_ioctl(struct block_device
*bdev
,
370 fmode_t mode
, unsigned int cmd
,
376 static inline int blkdev_reset_zones_ioctl(struct block_device
*bdev
,
377 fmode_t mode
, unsigned int cmd
,
383 #endif /* CONFIG_BLK_DEV_ZONED */
385 struct request_queue
{
387 * Together with queue_head for cacheline sharing
389 struct list_head queue_head
;
390 struct request
*last_merge
;
391 struct elevator_queue
*elevator
;
393 struct blk_queue_stats
*stats
;
394 struct rq_qos
*rq_qos
;
396 make_request_fn
*make_request_fn
;
397 dma_drain_needed_fn
*dma_drain_needed
;
399 const struct blk_mq_ops
*mq_ops
;
402 struct blk_mq_ctx __percpu
*queue_ctx
;
403 unsigned int nr_queues
;
405 unsigned int queue_depth
;
407 /* hw dispatch queues */
408 struct blk_mq_hw_ctx
**queue_hw_ctx
;
409 unsigned int nr_hw_queues
;
411 struct backing_dev_info
*backing_dev_info
;
414 * The queue owner gets to use this for whatever they like.
415 * ll_rw_blk doesn't touch it.
420 * various queue flags, see QUEUE_* below
422 unsigned long queue_flags
;
424 * Number of contexts that have called blk_set_pm_only(). If this
425 * counter is above zero then only RQF_PM and RQF_PREEMPT requests are
431 * ida allocated id for this queue. Used to index queues from
437 * queue needs bounce pages for pages above this limit
441 spinlock_t queue_lock
;
451 struct kobject
*mq_kobj
;
453 #ifdef CONFIG_BLK_DEV_INTEGRITY
454 struct blk_integrity integrity
;
455 #endif /* CONFIG_BLK_DEV_INTEGRITY */
460 unsigned int nr_pending
;
466 unsigned long nr_requests
; /* Max # of requests */
468 unsigned int dma_drain_size
;
469 void *dma_drain_buffer
;
470 unsigned int dma_pad_mask
;
471 unsigned int dma_alignment
;
473 unsigned int rq_timeout
;
476 struct blk_stat_callback
*poll_cb
;
477 struct blk_rq_stat poll_stat
[BLK_MQ_POLL_STATS_BKTS
];
479 struct timer_list timeout
;
480 struct work_struct timeout_work
;
482 struct list_head icq_list
;
483 #ifdef CONFIG_BLK_CGROUP
484 DECLARE_BITMAP (blkcg_pols
, BLKCG_MAX_POLS
);
485 struct blkcg_gq
*root_blkg
;
486 struct list_head blkg_list
;
489 struct queue_limits limits
;
491 #ifdef CONFIG_BLK_DEV_ZONED
493 * Zoned block device information for request dispatch control.
494 * nr_zones is the total number of zones of the device. This is always
495 * 0 for regular block devices. seq_zones_bitmap is a bitmap of nr_zones
496 * bits which indicates if a zone is conventional (bit clear) or
497 * sequential (bit set). seq_zones_wlock is a bitmap of nr_zones
498 * bits which indicates if a zone is write locked, that is, if a write
499 * request targeting the zone was dispatched. All three fields are
500 * initialized by the low level device driver (e.g. scsi/sd.c).
501 * Stacking drivers (device mappers) may or may not initialize
504 * Reads of this information must be protected with blk_queue_enter() /
505 * blk_queue_exit(). Modifying this information is only allowed while
506 * no requests are being processed. See also blk_mq_freeze_queue() and
507 * blk_mq_unfreeze_queue().
509 unsigned int nr_zones
;
510 unsigned long *seq_zones_bitmap
;
511 unsigned long *seq_zones_wlock
;
512 #endif /* CONFIG_BLK_DEV_ZONED */
517 unsigned int sg_timeout
;
518 unsigned int sg_reserved_size
;
520 #ifdef CONFIG_BLK_DEV_IO_TRACE
521 struct blk_trace
*blk_trace
;
522 struct mutex blk_trace_mutex
;
525 * for flush operations
527 struct blk_flush_queue
*fq
;
529 struct list_head requeue_list
;
530 spinlock_t requeue_lock
;
531 struct delayed_work requeue_work
;
533 struct mutex sysfs_lock
;
535 atomic_t mq_freeze_depth
;
537 #if defined(CONFIG_BLK_DEV_BSG)
538 struct bsg_class_device bsg_dev
;
541 #ifdef CONFIG_BLK_DEV_THROTTLING
543 struct throtl_data
*td
;
545 struct rcu_head rcu_head
;
546 wait_queue_head_t mq_freeze_wq
;
547 struct percpu_ref q_usage_counter
;
548 struct list_head all_q_node
;
550 struct blk_mq_tag_set
*tag_set
;
551 struct list_head tag_set_list
;
552 struct bio_set bio_split
;
554 #ifdef CONFIG_BLK_DEBUG_FS
555 struct dentry
*debugfs_dir
;
556 struct dentry
*sched_debugfs_dir
;
557 struct dentry
*rqos_debugfs_dir
;
560 bool mq_sysfs_init_done
;
564 struct work_struct release_work
;
566 #define BLK_MAX_WRITE_HINTS 5
567 u64 write_hints
[BLK_MAX_WRITE_HINTS
];
570 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
571 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
572 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
573 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
574 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
575 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
576 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
577 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
578 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
579 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
580 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
581 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
582 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
583 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
584 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
585 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
586 #define QUEUE_FLAG_WC 17 /* Write back caching */
587 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
588 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
589 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
590 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
591 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
592 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
593 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
594 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
596 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
597 (1 << QUEUE_FLAG_SAME_COMP))
599 void blk_queue_flag_set(unsigned int flag
, struct request_queue
*q
);
600 void blk_queue_flag_clear(unsigned int flag
, struct request_queue
*q
);
601 bool blk_queue_flag_test_and_set(unsigned int flag
, struct request_queue
*q
);
603 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
604 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
605 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
606 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
607 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
608 #define blk_queue_noxmerges(q) \
609 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
610 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
611 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
612 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
613 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
614 #define blk_queue_secure_erase(q) \
615 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
616 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
617 #define blk_queue_scsi_passthrough(q) \
618 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
619 #define blk_queue_pci_p2pdma(q) \
620 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
622 #define blk_noretry_request(rq) \
623 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
624 REQ_FAILFAST_DRIVER))
625 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
626 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
627 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
629 extern void blk_set_pm_only(struct request_queue
*q
);
630 extern void blk_clear_pm_only(struct request_queue
*q
);
632 static inline bool blk_account_rq(struct request
*rq
)
634 return (rq
->rq_flags
& RQF_STARTED
) && !blk_rq_is_passthrough(rq
);
637 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
639 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
641 static inline bool queue_is_mq(struct request_queue
*q
)
646 static inline enum blk_zoned_model
647 blk_queue_zoned_model(struct request_queue
*q
)
649 return q
->limits
.zoned
;
652 static inline bool blk_queue_is_zoned(struct request_queue
*q
)
654 switch (blk_queue_zoned_model(q
)) {
663 static inline unsigned int blk_queue_zone_sectors(struct request_queue
*q
)
665 return blk_queue_is_zoned(q
) ? q
->limits
.chunk_sectors
: 0;
668 #ifdef CONFIG_BLK_DEV_ZONED
669 static inline unsigned int blk_queue_nr_zones(struct request_queue
*q
)
671 return blk_queue_is_zoned(q
) ? q
->nr_zones
: 0;
674 static inline unsigned int blk_queue_zone_no(struct request_queue
*q
,
677 if (!blk_queue_is_zoned(q
))
679 return sector
>> ilog2(q
->limits
.chunk_sectors
);
682 static inline bool blk_queue_zone_is_seq(struct request_queue
*q
,
685 if (!blk_queue_is_zoned(q
) || !q
->seq_zones_bitmap
)
687 return test_bit(blk_queue_zone_no(q
, sector
), q
->seq_zones_bitmap
);
689 #else /* CONFIG_BLK_DEV_ZONED */
690 static inline unsigned int blk_queue_nr_zones(struct request_queue
*q
)
694 #endif /* CONFIG_BLK_DEV_ZONED */
696 static inline bool rq_is_sync(struct request
*rq
)
698 return op_is_sync(rq
->cmd_flags
);
701 static inline bool rq_mergeable(struct request
*rq
)
703 if (blk_rq_is_passthrough(rq
))
706 if (req_op(rq
) == REQ_OP_FLUSH
)
709 if (req_op(rq
) == REQ_OP_WRITE_ZEROES
)
712 if (rq
->cmd_flags
& REQ_NOMERGE_FLAGS
)
714 if (rq
->rq_flags
& RQF_NOMERGE_FLAGS
)
720 static inline bool blk_write_same_mergeable(struct bio
*a
, struct bio
*b
)
722 if (bio_page(a
) == bio_page(b
) &&
723 bio_offset(a
) == bio_offset(b
))
729 static inline unsigned int blk_queue_depth(struct request_queue
*q
)
732 return q
->queue_depth
;
734 return q
->nr_requests
;
737 extern unsigned long blk_max_low_pfn
, blk_max_pfn
;
740 * standard bounce addresses:
742 * BLK_BOUNCE_HIGH : bounce all highmem pages
743 * BLK_BOUNCE_ANY : don't bounce anything
744 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
747 #if BITS_PER_LONG == 32
748 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
750 #define BLK_BOUNCE_HIGH -1ULL
752 #define BLK_BOUNCE_ANY (-1ULL)
753 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
756 * default timeout for SG_IO if none specified
758 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
759 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
765 unsigned long offset
;
770 struct req_iterator
{
771 struct bvec_iter iter
;
775 /* This should not be used directly - use rq_for_each_segment */
776 #define for_each_bio(_bio) \
777 for (; _bio; _bio = _bio->bi_next)
778 #define __rq_for_each_bio(_bio, rq) \
780 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
782 #define rq_for_each_segment(bvl, _rq, _iter) \
783 __rq_for_each_bio(_iter.bio, _rq) \
784 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
786 #define rq_for_each_bvec(bvl, _rq, _iter) \
787 __rq_for_each_bio(_iter.bio, _rq) \
788 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
790 #define rq_iter_last(bvec, _iter) \
791 (_iter.bio->bi_next == NULL && \
792 bio_iter_last(bvec, _iter.iter))
794 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
795 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
797 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
798 extern void rq_flush_dcache_pages(struct request
*rq
);
800 static inline void rq_flush_dcache_pages(struct request
*rq
)
805 extern int blk_register_queue(struct gendisk
*disk
);
806 extern void blk_unregister_queue(struct gendisk
*disk
);
807 extern blk_qc_t
generic_make_request(struct bio
*bio
);
808 extern blk_qc_t
direct_make_request(struct bio
*bio
);
809 extern void blk_rq_init(struct request_queue
*q
, struct request
*rq
);
810 extern void blk_init_request_from_bio(struct request
*req
, struct bio
*bio
);
811 extern void blk_put_request(struct request
*);
812 extern struct request
*blk_get_request(struct request_queue
*, unsigned int op
,
813 blk_mq_req_flags_t flags
);
814 extern int blk_lld_busy(struct request_queue
*q
);
815 extern int blk_rq_prep_clone(struct request
*rq
, struct request
*rq_src
,
816 struct bio_set
*bs
, gfp_t gfp_mask
,
817 int (*bio_ctr
)(struct bio
*, struct bio
*, void *),
819 extern void blk_rq_unprep_clone(struct request
*rq
);
820 extern blk_status_t
blk_insert_cloned_request(struct request_queue
*q
,
822 extern int blk_rq_append_bio(struct request
*rq
, struct bio
**bio
);
823 extern void blk_queue_split(struct request_queue
*, struct bio
**);
824 extern void blk_recount_segments(struct request_queue
*, struct bio
*);
825 extern int scsi_verify_blk_ioctl(struct block_device
*, unsigned int);
826 extern int scsi_cmd_blk_ioctl(struct block_device
*, fmode_t
,
827 unsigned int, void __user
*);
828 extern int scsi_cmd_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
829 unsigned int, void __user
*);
830 extern int sg_scsi_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
831 struct scsi_ioctl_command __user
*);
833 extern int blk_queue_enter(struct request_queue
*q
, blk_mq_req_flags_t flags
);
834 extern void blk_queue_exit(struct request_queue
*q
);
835 extern void blk_sync_queue(struct request_queue
*q
);
836 extern int blk_rq_map_user(struct request_queue
*, struct request
*,
837 struct rq_map_data
*, void __user
*, unsigned long,
839 extern int blk_rq_unmap_user(struct bio
*);
840 extern int blk_rq_map_kern(struct request_queue
*, struct request
*, void *, unsigned int, gfp_t
);
841 extern int blk_rq_map_user_iov(struct request_queue
*, struct request
*,
842 struct rq_map_data
*, const struct iov_iter
*,
844 extern void blk_execute_rq(struct request_queue
*, struct gendisk
*,
845 struct request
*, int);
846 extern void blk_execute_rq_nowait(struct request_queue
*, struct gendisk
*,
847 struct request
*, int, rq_end_io_fn
*);
849 int blk_status_to_errno(blk_status_t status
);
850 blk_status_t
errno_to_blk_status(int errno
);
852 int blk_poll(struct request_queue
*q
, blk_qc_t cookie
, bool spin
);
854 static inline struct request_queue
*bdev_get_queue(struct block_device
*bdev
)
856 return bdev
->bd_disk
->queue
; /* this is never NULL */
860 * The basic unit of block I/O is a sector. It is used in a number of contexts
861 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
862 * bytes. Variables of type sector_t represent an offset or size that is a
863 * multiple of 512 bytes. Hence these two constants.
866 #define SECTOR_SHIFT 9
869 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
873 * blk_rq_pos() : the current sector
874 * blk_rq_bytes() : bytes left in the entire request
875 * blk_rq_cur_bytes() : bytes left in the current segment
876 * blk_rq_err_bytes() : bytes left till the next error boundary
877 * blk_rq_sectors() : sectors left in the entire request
878 * blk_rq_cur_sectors() : sectors left in the current segment
880 static inline sector_t
blk_rq_pos(const struct request
*rq
)
885 static inline unsigned int blk_rq_bytes(const struct request
*rq
)
887 return rq
->__data_len
;
890 static inline int blk_rq_cur_bytes(const struct request
*rq
)
892 return rq
->bio
? bio_cur_bytes(rq
->bio
) : 0;
895 extern unsigned int blk_rq_err_bytes(const struct request
*rq
);
897 static inline unsigned int blk_rq_sectors(const struct request
*rq
)
899 return blk_rq_bytes(rq
) >> SECTOR_SHIFT
;
902 static inline unsigned int blk_rq_cur_sectors(const struct request
*rq
)
904 return blk_rq_cur_bytes(rq
) >> SECTOR_SHIFT
;
907 #ifdef CONFIG_BLK_DEV_ZONED
908 static inline unsigned int blk_rq_zone_no(struct request
*rq
)
910 return blk_queue_zone_no(rq
->q
, blk_rq_pos(rq
));
913 static inline unsigned int blk_rq_zone_is_seq(struct request
*rq
)
915 return blk_queue_zone_is_seq(rq
->q
, blk_rq_pos(rq
));
917 #endif /* CONFIG_BLK_DEV_ZONED */
920 * Some commands like WRITE SAME have a payload or data transfer size which
921 * is different from the size of the request. Any driver that supports such
922 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
923 * calculate the data transfer size.
925 static inline unsigned int blk_rq_payload_bytes(struct request
*rq
)
927 if (rq
->rq_flags
& RQF_SPECIAL_PAYLOAD
)
928 return rq
->special_vec
.bv_len
;
929 return blk_rq_bytes(rq
);
932 static inline unsigned int blk_queue_get_max_sectors(struct request_queue
*q
,
935 if (unlikely(op
== REQ_OP_DISCARD
|| op
== REQ_OP_SECURE_ERASE
))
936 return min(q
->limits
.max_discard_sectors
,
937 UINT_MAX
>> SECTOR_SHIFT
);
939 if (unlikely(op
== REQ_OP_WRITE_SAME
))
940 return q
->limits
.max_write_same_sectors
;
942 if (unlikely(op
== REQ_OP_WRITE_ZEROES
))
943 return q
->limits
.max_write_zeroes_sectors
;
945 return q
->limits
.max_sectors
;
949 * Return maximum size of a request at given offset. Only valid for
950 * file system requests.
952 static inline unsigned int blk_max_size_offset(struct request_queue
*q
,
955 if (!q
->limits
.chunk_sectors
)
956 return q
->limits
.max_sectors
;
958 return min(q
->limits
.max_sectors
, (unsigned int)(q
->limits
.chunk_sectors
-
959 (offset
& (q
->limits
.chunk_sectors
- 1))));
962 static inline unsigned int blk_rq_get_max_sectors(struct request
*rq
,
965 struct request_queue
*q
= rq
->q
;
967 if (blk_rq_is_passthrough(rq
))
968 return q
->limits
.max_hw_sectors
;
970 if (!q
->limits
.chunk_sectors
||
971 req_op(rq
) == REQ_OP_DISCARD
||
972 req_op(rq
) == REQ_OP_SECURE_ERASE
)
973 return blk_queue_get_max_sectors(q
, req_op(rq
));
975 return min(blk_max_size_offset(q
, offset
),
976 blk_queue_get_max_sectors(q
, req_op(rq
)));
979 static inline unsigned int blk_rq_count_bios(struct request
*rq
)
981 unsigned int nr_bios
= 0;
984 __rq_for_each_bio(bio
, rq
)
990 void blk_steal_bios(struct bio_list
*list
, struct request
*rq
);
993 * Request completion related functions.
995 * blk_update_request() completes given number of bytes and updates
996 * the request without completing it.
998 * blk_end_request() and friends. __blk_end_request() must be called
999 * with the request queue spinlock acquired.
1001 * Several drivers define their own end_request and call
1002 * blk_end_request() for parts of the original function.
1003 * This prevents code duplication in drivers.
1005 extern bool blk_update_request(struct request
*rq
, blk_status_t error
,
1006 unsigned int nr_bytes
);
1007 extern void blk_end_request_all(struct request
*rq
, blk_status_t error
);
1008 extern bool __blk_end_request(struct request
*rq
, blk_status_t error
,
1009 unsigned int nr_bytes
);
1010 extern void __blk_end_request_all(struct request
*rq
, blk_status_t error
);
1011 extern bool __blk_end_request_cur(struct request
*rq
, blk_status_t error
);
1013 extern void __blk_complete_request(struct request
*);
1014 extern void blk_abort_request(struct request
*);
1017 * Access functions for manipulating queue properties
1019 extern void blk_cleanup_queue(struct request_queue
*);
1020 extern void blk_queue_make_request(struct request_queue
*, make_request_fn
*);
1021 extern void blk_queue_bounce_limit(struct request_queue
*, u64
);
1022 extern void blk_queue_max_hw_sectors(struct request_queue
*, unsigned int);
1023 extern void blk_queue_chunk_sectors(struct request_queue
*, unsigned int);
1024 extern void blk_queue_max_segments(struct request_queue
*, unsigned short);
1025 extern void blk_queue_max_discard_segments(struct request_queue
*,
1027 extern void blk_queue_max_segment_size(struct request_queue
*, unsigned int);
1028 extern void blk_queue_max_discard_sectors(struct request_queue
*q
,
1029 unsigned int max_discard_sectors
);
1030 extern void blk_queue_max_write_same_sectors(struct request_queue
*q
,
1031 unsigned int max_write_same_sectors
);
1032 extern void blk_queue_max_write_zeroes_sectors(struct request_queue
*q
,
1033 unsigned int max_write_same_sectors
);
1034 extern void blk_queue_logical_block_size(struct request_queue
*, unsigned short);
1035 extern void blk_queue_physical_block_size(struct request_queue
*, unsigned int);
1036 extern void blk_queue_alignment_offset(struct request_queue
*q
,
1037 unsigned int alignment
);
1038 extern void blk_limits_io_min(struct queue_limits
*limits
, unsigned int min
);
1039 extern void blk_queue_io_min(struct request_queue
*q
, unsigned int min
);
1040 extern void blk_limits_io_opt(struct queue_limits
*limits
, unsigned int opt
);
1041 extern void blk_queue_io_opt(struct request_queue
*q
, unsigned int opt
);
1042 extern void blk_set_queue_depth(struct request_queue
*q
, unsigned int depth
);
1043 extern void blk_set_default_limits(struct queue_limits
*lim
);
1044 extern void blk_set_stacking_limits(struct queue_limits
*lim
);
1045 extern int blk_stack_limits(struct queue_limits
*t
, struct queue_limits
*b
,
1047 extern int bdev_stack_limits(struct queue_limits
*t
, struct block_device
*bdev
,
1049 extern void disk_stack_limits(struct gendisk
*disk
, struct block_device
*bdev
,
1051 extern void blk_queue_stack_limits(struct request_queue
*t
, struct request_queue
*b
);
1052 extern void blk_queue_dma_pad(struct request_queue
*, unsigned int);
1053 extern void blk_queue_update_dma_pad(struct request_queue
*, unsigned int);
1054 extern int blk_queue_dma_drain(struct request_queue
*q
,
1055 dma_drain_needed_fn
*dma_drain_needed
,
1056 void *buf
, unsigned int size
);
1057 extern void blk_queue_segment_boundary(struct request_queue
*, unsigned long);
1058 extern void blk_queue_virt_boundary(struct request_queue
*, unsigned long);
1059 extern void blk_queue_dma_alignment(struct request_queue
*, int);
1060 extern void blk_queue_update_dma_alignment(struct request_queue
*, int);
1061 extern void blk_queue_rq_timeout(struct request_queue
*, unsigned int);
1062 extern void blk_queue_write_cache(struct request_queue
*q
, bool enabled
, bool fua
);
1065 * Number of physical segments as sent to the device.
1067 * Normally this is the number of discontiguous data segments sent by the
1068 * submitter. But for data-less command like discard we might have no
1069 * actual data segments submitted, but the driver might have to add it's
1070 * own special payload. In that case we still return 1 here so that this
1071 * special payload will be mapped.
1073 static inline unsigned short blk_rq_nr_phys_segments(struct request
*rq
)
1075 if (rq
->rq_flags
& RQF_SPECIAL_PAYLOAD
)
1077 return rq
->nr_phys_segments
;
1081 * Number of discard segments (or ranges) the driver needs to fill in.
1082 * Each discard bio merged into a request is counted as one segment.
1084 static inline unsigned short blk_rq_nr_discard_segments(struct request
*rq
)
1086 return max_t(unsigned short, rq
->nr_phys_segments
, 1);
1089 extern int blk_rq_map_sg(struct request_queue
*, struct request
*, struct scatterlist
*);
1090 extern void blk_dump_rq_flags(struct request
*, char *);
1091 extern long nr_blockdev_pages(void);
1093 bool __must_check
blk_get_queue(struct request_queue
*);
1094 struct request_queue
*blk_alloc_queue(gfp_t
);
1095 struct request_queue
*blk_alloc_queue_node(gfp_t gfp_mask
, int node_id
);
1096 extern void blk_put_queue(struct request_queue
*);
1097 extern void blk_set_queue_dying(struct request_queue
*);
1100 * blk_plug permits building a queue of related requests by holding the I/O
1101 * fragments for a short period. This allows merging of sequential requests
1102 * into single larger request. As the requests are moved from a per-task list to
1103 * the device's request_queue in a batch, this results in improved scalability
1104 * as the lock contention for request_queue lock is reduced.
1106 * It is ok not to disable preemption when adding the request to the plug list
1107 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1108 * the plug list when the task sleeps by itself. For details, please see
1109 * schedule() where blk_schedule_flush_plug() is called.
1112 struct list_head mq_list
; /* blk-mq requests */
1113 struct list_head cb_list
; /* md requires an unplug callback */
1114 unsigned short rq_count
;
1115 bool multiple_queues
;
1117 #define BLK_MAX_REQUEST_COUNT 16
1118 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1121 typedef void (*blk_plug_cb_fn
)(struct blk_plug_cb
*, bool);
1122 struct blk_plug_cb
{
1123 struct list_head list
;
1124 blk_plug_cb_fn callback
;
1127 extern struct blk_plug_cb
*blk_check_plugged(blk_plug_cb_fn unplug
,
1128 void *data
, int size
);
1129 extern void blk_start_plug(struct blk_plug
*);
1130 extern void blk_finish_plug(struct blk_plug
*);
1131 extern void blk_flush_plug_list(struct blk_plug
*, bool);
1133 static inline void blk_flush_plug(struct task_struct
*tsk
)
1135 struct blk_plug
*plug
= tsk
->plug
;
1138 blk_flush_plug_list(plug
, false);
1141 static inline void blk_schedule_flush_plug(struct task_struct
*tsk
)
1143 struct blk_plug
*plug
= tsk
->plug
;
1146 blk_flush_plug_list(plug
, true);
1149 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
1151 struct blk_plug
*plug
= tsk
->plug
;
1154 (!list_empty(&plug
->mq_list
) ||
1155 !list_empty(&plug
->cb_list
));
1158 extern int blkdev_issue_flush(struct block_device
*, gfp_t
, sector_t
*);
1159 extern int blkdev_issue_write_same(struct block_device
*bdev
, sector_t sector
,
1160 sector_t nr_sects
, gfp_t gfp_mask
, struct page
*page
);
1162 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1164 extern int blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
1165 sector_t nr_sects
, gfp_t gfp_mask
, unsigned long flags
);
1166 extern int __blkdev_issue_discard(struct block_device
*bdev
, sector_t sector
,
1167 sector_t nr_sects
, gfp_t gfp_mask
, int flags
,
1170 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1171 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1173 extern int __blkdev_issue_zeroout(struct block_device
*bdev
, sector_t sector
,
1174 sector_t nr_sects
, gfp_t gfp_mask
, struct bio
**biop
,
1176 extern int blkdev_issue_zeroout(struct block_device
*bdev
, sector_t sector
,
1177 sector_t nr_sects
, gfp_t gfp_mask
, unsigned flags
);
1179 static inline int sb_issue_discard(struct super_block
*sb
, sector_t block
,
1180 sector_t nr_blocks
, gfp_t gfp_mask
, unsigned long flags
)
1182 return blkdev_issue_discard(sb
->s_bdev
,
1183 block
<< (sb
->s_blocksize_bits
-
1185 nr_blocks
<< (sb
->s_blocksize_bits
-
1189 static inline int sb_issue_zeroout(struct super_block
*sb
, sector_t block
,
1190 sector_t nr_blocks
, gfp_t gfp_mask
)
1192 return blkdev_issue_zeroout(sb
->s_bdev
,
1193 block
<< (sb
->s_blocksize_bits
-
1195 nr_blocks
<< (sb
->s_blocksize_bits
-
1200 extern int blk_verify_command(unsigned char *cmd
, fmode_t mode
);
1202 enum blk_default_limits
{
1203 BLK_MAX_SEGMENTS
= 128,
1204 BLK_SAFE_MAX_SECTORS
= 255,
1205 BLK_DEF_MAX_SECTORS
= 2560,
1206 BLK_MAX_SEGMENT_SIZE
= 65536,
1207 BLK_SEG_BOUNDARY_MASK
= 0xFFFFFFFFUL
,
1210 static inline unsigned long queue_segment_boundary(struct request_queue
*q
)
1212 return q
->limits
.seg_boundary_mask
;
1215 static inline unsigned long queue_virt_boundary(struct request_queue
*q
)
1217 return q
->limits
.virt_boundary_mask
;
1220 static inline unsigned int queue_max_sectors(struct request_queue
*q
)
1222 return q
->limits
.max_sectors
;
1225 static inline unsigned int queue_max_hw_sectors(struct request_queue
*q
)
1227 return q
->limits
.max_hw_sectors
;
1230 static inline unsigned short queue_max_segments(struct request_queue
*q
)
1232 return q
->limits
.max_segments
;
1235 static inline unsigned short queue_max_discard_segments(struct request_queue
*q
)
1237 return q
->limits
.max_discard_segments
;
1240 static inline unsigned int queue_max_segment_size(struct request_queue
*q
)
1242 return q
->limits
.max_segment_size
;
1245 static inline unsigned short queue_logical_block_size(struct request_queue
*q
)
1249 if (q
&& q
->limits
.logical_block_size
)
1250 retval
= q
->limits
.logical_block_size
;
1255 static inline unsigned short bdev_logical_block_size(struct block_device
*bdev
)
1257 return queue_logical_block_size(bdev_get_queue(bdev
));
1260 static inline unsigned int queue_physical_block_size(struct request_queue
*q
)
1262 return q
->limits
.physical_block_size
;
1265 static inline unsigned int bdev_physical_block_size(struct block_device
*bdev
)
1267 return queue_physical_block_size(bdev_get_queue(bdev
));
1270 static inline unsigned int queue_io_min(struct request_queue
*q
)
1272 return q
->limits
.io_min
;
1275 static inline int bdev_io_min(struct block_device
*bdev
)
1277 return queue_io_min(bdev_get_queue(bdev
));
1280 static inline unsigned int queue_io_opt(struct request_queue
*q
)
1282 return q
->limits
.io_opt
;
1285 static inline int bdev_io_opt(struct block_device
*bdev
)
1287 return queue_io_opt(bdev_get_queue(bdev
));
1290 static inline int queue_alignment_offset(struct request_queue
*q
)
1292 if (q
->limits
.misaligned
)
1295 return q
->limits
.alignment_offset
;
1298 static inline int queue_limit_alignment_offset(struct queue_limits
*lim
, sector_t sector
)
1300 unsigned int granularity
= max(lim
->physical_block_size
, lim
->io_min
);
1301 unsigned int alignment
= sector_div(sector
, granularity
>> SECTOR_SHIFT
)
1304 return (granularity
+ lim
->alignment_offset
- alignment
) % granularity
;
1307 static inline int bdev_alignment_offset(struct block_device
*bdev
)
1309 struct request_queue
*q
= bdev_get_queue(bdev
);
1311 if (q
->limits
.misaligned
)
1314 if (bdev
!= bdev
->bd_contains
)
1315 return bdev
->bd_part
->alignment_offset
;
1317 return q
->limits
.alignment_offset
;
1320 static inline int queue_discard_alignment(struct request_queue
*q
)
1322 if (q
->limits
.discard_misaligned
)
1325 return q
->limits
.discard_alignment
;
1328 static inline int queue_limit_discard_alignment(struct queue_limits
*lim
, sector_t sector
)
1330 unsigned int alignment
, granularity
, offset
;
1332 if (!lim
->max_discard_sectors
)
1335 /* Why are these in bytes, not sectors? */
1336 alignment
= lim
->discard_alignment
>> SECTOR_SHIFT
;
1337 granularity
= lim
->discard_granularity
>> SECTOR_SHIFT
;
1341 /* Offset of the partition start in 'granularity' sectors */
1342 offset
= sector_div(sector
, granularity
);
1344 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1345 offset
= (granularity
+ alignment
- offset
) % granularity
;
1347 /* Turn it back into bytes, gaah */
1348 return offset
<< SECTOR_SHIFT
;
1351 static inline int bdev_discard_alignment(struct block_device
*bdev
)
1353 struct request_queue
*q
= bdev_get_queue(bdev
);
1355 if (bdev
!= bdev
->bd_contains
)
1356 return bdev
->bd_part
->discard_alignment
;
1358 return q
->limits
.discard_alignment
;
1361 static inline unsigned int bdev_write_same(struct block_device
*bdev
)
1363 struct request_queue
*q
= bdev_get_queue(bdev
);
1366 return q
->limits
.max_write_same_sectors
;
1371 static inline unsigned int bdev_write_zeroes_sectors(struct block_device
*bdev
)
1373 struct request_queue
*q
= bdev_get_queue(bdev
);
1376 return q
->limits
.max_write_zeroes_sectors
;
1381 static inline enum blk_zoned_model
bdev_zoned_model(struct block_device
*bdev
)
1383 struct request_queue
*q
= bdev_get_queue(bdev
);
1386 return blk_queue_zoned_model(q
);
1388 return BLK_ZONED_NONE
;
1391 static inline bool bdev_is_zoned(struct block_device
*bdev
)
1393 struct request_queue
*q
= bdev_get_queue(bdev
);
1396 return blk_queue_is_zoned(q
);
1401 static inline unsigned int bdev_zone_sectors(struct block_device
*bdev
)
1403 struct request_queue
*q
= bdev_get_queue(bdev
);
1406 return blk_queue_zone_sectors(q
);
1410 static inline int queue_dma_alignment(struct request_queue
*q
)
1412 return q
? q
->dma_alignment
: 511;
1415 static inline int blk_rq_aligned(struct request_queue
*q
, unsigned long addr
,
1418 unsigned int alignment
= queue_dma_alignment(q
) | q
->dma_pad_mask
;
1419 return !(addr
& alignment
) && !(len
& alignment
);
1422 /* assumes size > 256 */
1423 static inline unsigned int blksize_bits(unsigned int size
)
1425 unsigned int bits
= 8;
1429 } while (size
> 256);
1433 static inline unsigned int block_size(struct block_device
*bdev
)
1435 return bdev
->bd_block_size
;
1438 typedef struct {struct page
*v
;} Sector
;
1440 unsigned char *read_dev_sector(struct block_device
*, sector_t
, Sector
*);
1442 static inline void put_dev_sector(Sector p
)
1447 int kblockd_schedule_work(struct work_struct
*work
);
1448 int kblockd_schedule_work_on(int cpu
, struct work_struct
*work
);
1449 int kblockd_mod_delayed_work_on(int cpu
, struct delayed_work
*dwork
, unsigned long delay
);
1451 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1452 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1453 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1454 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1456 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1458 enum blk_integrity_flags
{
1459 BLK_INTEGRITY_VERIFY
= 1 << 0,
1460 BLK_INTEGRITY_GENERATE
= 1 << 1,
1461 BLK_INTEGRITY_DEVICE_CAPABLE
= 1 << 2,
1462 BLK_INTEGRITY_IP_CHECKSUM
= 1 << 3,
1465 struct blk_integrity_iter
{
1469 unsigned int data_size
;
1470 unsigned short interval
;
1471 const char *disk_name
;
1474 typedef blk_status_t (integrity_processing_fn
) (struct blk_integrity_iter
*);
1476 struct blk_integrity_profile
{
1477 integrity_processing_fn
*generate_fn
;
1478 integrity_processing_fn
*verify_fn
;
1482 extern void blk_integrity_register(struct gendisk
*, struct blk_integrity
*);
1483 extern void blk_integrity_unregister(struct gendisk
*);
1484 extern int blk_integrity_compare(struct gendisk
*, struct gendisk
*);
1485 extern int blk_rq_map_integrity_sg(struct request_queue
*, struct bio
*,
1486 struct scatterlist
*);
1487 extern int blk_rq_count_integrity_sg(struct request_queue
*, struct bio
*);
1488 extern bool blk_integrity_merge_rq(struct request_queue
*, struct request
*,
1490 extern bool blk_integrity_merge_bio(struct request_queue
*, struct request
*,
1493 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1495 struct blk_integrity
*bi
= &disk
->queue
->integrity
;
1504 struct blk_integrity
*bdev_get_integrity(struct block_device
*bdev
)
1506 return blk_get_integrity(bdev
->bd_disk
);
1509 static inline bool blk_integrity_rq(struct request
*rq
)
1511 return rq
->cmd_flags
& REQ_INTEGRITY
;
1514 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1517 q
->limits
.max_integrity_segments
= segs
;
1520 static inline unsigned short
1521 queue_max_integrity_segments(struct request_queue
*q
)
1523 return q
->limits
.max_integrity_segments
;
1527 * bio_integrity_intervals - Return number of integrity intervals for a bio
1528 * @bi: blk_integrity profile for device
1529 * @sectors: Size of the bio in 512-byte sectors
1531 * Description: The block layer calculates everything in 512 byte
1532 * sectors but integrity metadata is done in terms of the data integrity
1533 * interval size of the storage device. Convert the block layer sectors
1534 * to the appropriate number of integrity intervals.
1536 static inline unsigned int bio_integrity_intervals(struct blk_integrity
*bi
,
1537 unsigned int sectors
)
1539 return sectors
>> (bi
->interval_exp
- 9);
1542 static inline unsigned int bio_integrity_bytes(struct blk_integrity
*bi
,
1543 unsigned int sectors
)
1545 return bio_integrity_intervals(bi
, sectors
) * bi
->tuple_size
;
1548 #else /* CONFIG_BLK_DEV_INTEGRITY */
1551 struct block_device
;
1553 struct blk_integrity
;
1555 static inline int blk_integrity_rq(struct request
*rq
)
1559 static inline int blk_rq_count_integrity_sg(struct request_queue
*q
,
1564 static inline int blk_rq_map_integrity_sg(struct request_queue
*q
,
1566 struct scatterlist
*s
)
1570 static inline struct blk_integrity
*bdev_get_integrity(struct block_device
*b
)
1574 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1578 static inline int blk_integrity_compare(struct gendisk
*a
, struct gendisk
*b
)
1582 static inline void blk_integrity_register(struct gendisk
*d
,
1583 struct blk_integrity
*b
)
1586 static inline void blk_integrity_unregister(struct gendisk
*d
)
1589 static inline void blk_queue_max_integrity_segments(struct request_queue
*q
,
1593 static inline unsigned short queue_max_integrity_segments(struct request_queue
*q
)
1597 static inline bool blk_integrity_merge_rq(struct request_queue
*rq
,
1603 static inline bool blk_integrity_merge_bio(struct request_queue
*rq
,
1610 static inline unsigned int bio_integrity_intervals(struct blk_integrity
*bi
,
1611 unsigned int sectors
)
1616 static inline unsigned int bio_integrity_bytes(struct blk_integrity
*bi
,
1617 unsigned int sectors
)
1622 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1624 struct block_device_operations
{
1625 int (*open
) (struct block_device
*, fmode_t
);
1626 void (*release
) (struct gendisk
*, fmode_t
);
1627 int (*rw_page
)(struct block_device
*, sector_t
, struct page
*, unsigned int);
1628 int (*ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1629 int (*compat_ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1630 unsigned int (*check_events
) (struct gendisk
*disk
,
1631 unsigned int clearing
);
1632 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1633 int (*media_changed
) (struct gendisk
*);
1634 void (*unlock_native_capacity
) (struct gendisk
*);
1635 int (*revalidate_disk
) (struct gendisk
*);
1636 int (*getgeo
)(struct block_device
*, struct hd_geometry
*);
1637 /* this callback is with swap_lock and sometimes page table lock held */
1638 void (*swap_slot_free_notify
) (struct block_device
*, unsigned long);
1639 int (*report_zones
)(struct gendisk
*, sector_t sector
,
1640 struct blk_zone
*zones
, unsigned int *nr_zones
,
1642 struct module
*owner
;
1643 const struct pr_ops
*pr_ops
;
1646 extern int __blkdev_driver_ioctl(struct block_device
*, fmode_t
, unsigned int,
1648 extern int bdev_read_page(struct block_device
*, sector_t
, struct page
*);
1649 extern int bdev_write_page(struct block_device
*, sector_t
, struct page
*,
1650 struct writeback_control
*);
1652 #ifdef CONFIG_BLK_DEV_ZONED
1653 bool blk_req_needs_zone_write_lock(struct request
*rq
);
1654 void __blk_req_zone_write_lock(struct request
*rq
);
1655 void __blk_req_zone_write_unlock(struct request
*rq
);
1657 static inline void blk_req_zone_write_lock(struct request
*rq
)
1659 if (blk_req_needs_zone_write_lock(rq
))
1660 __blk_req_zone_write_lock(rq
);
1663 static inline void blk_req_zone_write_unlock(struct request
*rq
)
1665 if (rq
->rq_flags
& RQF_ZONE_WRITE_LOCKED
)
1666 __blk_req_zone_write_unlock(rq
);
1669 static inline bool blk_req_zone_is_write_locked(struct request
*rq
)
1671 return rq
->q
->seq_zones_wlock
&&
1672 test_bit(blk_rq_zone_no(rq
), rq
->q
->seq_zones_wlock
);
1675 static inline bool blk_req_can_dispatch_to_zone(struct request
*rq
)
1677 if (!blk_req_needs_zone_write_lock(rq
))
1679 return !blk_req_zone_is_write_locked(rq
);
1682 static inline bool blk_req_needs_zone_write_lock(struct request
*rq
)
1687 static inline void blk_req_zone_write_lock(struct request
*rq
)
1691 static inline void blk_req_zone_write_unlock(struct request
*rq
)
1694 static inline bool blk_req_zone_is_write_locked(struct request
*rq
)
1699 static inline bool blk_req_can_dispatch_to_zone(struct request
*rq
)
1703 #endif /* CONFIG_BLK_DEV_ZONED */
1705 #else /* CONFIG_BLOCK */
1707 struct block_device
;
1710 * stubs for when the block layer is configured out
1712 #define buffer_heads_over_limit 0
1714 static inline long nr_blockdev_pages(void)
1722 static inline void blk_start_plug(struct blk_plug
*plug
)
1726 static inline void blk_finish_plug(struct blk_plug
*plug
)
1730 static inline void blk_flush_plug(struct task_struct
*task
)
1734 static inline void blk_schedule_flush_plug(struct task_struct
*task
)
1739 static inline bool blk_needs_flush_plug(struct task_struct
*tsk
)
1744 static inline int blkdev_issue_flush(struct block_device
*bdev
, gfp_t gfp_mask
,
1745 sector_t
*error_sector
)
1750 #endif /* CONFIG_BLOCK */
1752 static inline void blk_wake_io_task(struct task_struct
*waiter
)
1755 * If we're polling, the task itself is doing the completions. For
1756 * that case, we don't need to signal a wakeup, it's enough to just
1757 * mark us as RUNNING.
1759 if (waiter
== current
)
1760 __set_current_state(TASK_RUNNING
);
1762 wake_up_process(waiter
);