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>
7 #include <linux/major.h>
8 #include <linux/genhd.h>
9 #include <linux/list.h>
10 #include <linux/llist.h>
11 #include <linux/timer.h>
12 #include <linux/workqueue.h>
13 #include <linux/pagemap.h>
14 #include <linux/backing-dev-defs.h>
15 #include <linux/wait.h>
16 #include <linux/mempool.h>
17 #include <linux/pfn.h>
18 #include <linux/bio.h>
19 #include <linux/stringify.h>
20 #include <linux/gfp.h>
21 #include <linux/bsg.h>
22 #include <linux/smp.h>
23 #include <linux/rcupdate.h>
24 #include <linux/percpu-refcount.h>
25 #include <linux/scatterlist.h>
26 #include <linux/blkzoned.h>
29 struct scsi_ioctl_command;
32 struct elevator_queue;
38 struct blk_flush_queue;
41 struct blk_queue_stats;
42 struct blk_stat_callback;
43 struct blk_keyslot_manager;
45 #define BLKDEV_MIN_RQ 4
46 #define BLKDEV_MAX_RQ 128 /* Default maximum */
48 /* Must be consistent with blk_mq_poll_stats_bkt() */
49 #define BLK_MQ_POLL_STATS_BKTS 16
51 /* Doing classic polling */
52 #define BLK_MQ_POLL_CLASSIC -1
55 * Maximum number of blkcg policies allowed to be registered concurrently.
56 * Defined here to simplify include dependency.
58 #define BLKCG_MAX_POLS 5
60 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
64 typedef __u32 __bitwise req_flags_t;
66 /* elevator knows about this request */
67 #define RQF_SORTED ((__force req_flags_t)(1 << 0))
68 /* drive already may have started this one */
69 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
70 /* may not be passed by ioscheduler */
71 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
72 /* request for flush sequence */
73 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
74 /* merge of different types, fail separately */
75 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
76 /* track inflight for MQ */
77 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
78 /* don't call prep for this one */
79 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
80 /* set for "ide_preempt" requests and also for requests for which the SCSI
81 "quiesce" state must be ignored. */
82 #define RQF_PREEMPT ((__force req_flags_t)(1 << 8))
83 /* vaguely specified driver internal error. Ignored by the block layer */
84 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
85 /* don't warn about errors */
86 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
87 /* elevator private data attached */
88 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
89 /* account into disk and partition IO statistics */
90 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
91 /* request came from our alloc pool */
92 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
93 /* runtime pm request */
94 #define RQF_PM ((__force req_flags_t)(1 << 15))
95 /* on IO scheduler merge hash */
96 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
97 /* track IO completion time */
98 #define RQF_STATS ((__force req_flags_t)(1 << 17))
99 /* Look at ->special_vec for the actual data payload instead of the
101 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
102 /* The per-zone write lock is held for this request */
103 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
104 /* already slept for hybrid poll */
105 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
106 /* ->timeout has been called, don't expire again */
107 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
109 /* flags that prevent us from merging requests: */
110 #define RQF_NOMERGE_FLAGS \
111 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
114 * Request state for blk-mq.
123 * Try to put the fields that are referenced together in the same cacheline.
125 * If you modify this structure, make sure to update blk_rq_init() and
126 * especially blk_mq_rq_ctx_init() to take care of the added fields.
129 struct request_queue *q;
130 struct blk_mq_ctx *mq_ctx;
131 struct blk_mq_hw_ctx *mq_hctx;
133 unsigned int cmd_flags; /* op and common flags */
134 req_flags_t rq_flags;
139 /* the following two fields are internal, NEVER access directly */
140 unsigned int __data_len; /* total data len */
141 sector_t __sector; /* sector cursor */
146 struct list_head queuelist;
149 * The hash is used inside the scheduler, and killed once the
150 * request reaches the dispatch list. The ipi_list is only used
151 * to queue the request for softirq completion, which is long
152 * after the request has been unhashed (and even removed from
153 * the dispatch list).
156 struct hlist_node hash; /* merge hash */
157 struct list_head ipi_list;
161 * The rb_node is only used inside the io scheduler, requests
162 * are pruned when moved to the dispatch queue. So let the
163 * completion_data share space with the rb_node.
166 struct rb_node rb_node; /* sort/lookup */
167 struct bio_vec special_vec;
168 void *completion_data;
169 int error_count; /* for legacy drivers, don't use */
173 * Three pointers are available for the IO schedulers, if they need
174 * more they have to dynamically allocate it. Flush requests are
175 * never put on the IO scheduler. So let the flush fields share
176 * space with the elevator data.
186 struct list_head list;
187 rq_end_io_fn *saved_end_io;
191 struct gendisk *rq_disk;
192 struct hd_struct *part;
193 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
194 /* Time that the first bio started allocating this request. */
197 /* Time that this request was allocated for this IO. */
199 /* Time that I/O was submitted to the device. */
200 u64 io_start_time_ns;
202 #ifdef CONFIG_BLK_WBT
203 unsigned short wbt_flags;
206 * rq sectors used for blk stats. It has the same value
207 * with blk_rq_sectors(rq), except that it never be zeroed
210 unsigned short stats_sectors;
213 * Number of scatter-gather DMA addr+len pairs after
214 * physical address coalescing is performed.
216 unsigned short nr_phys_segments;
218 #if defined(CONFIG_BLK_DEV_INTEGRITY)
219 unsigned short nr_integrity_segments;
222 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
223 struct bio_crypt_ctx *crypt_ctx;
224 struct blk_ksm_keyslot *crypt_keyslot;
227 unsigned short write_hint;
228 unsigned short ioprio;
230 enum mq_rq_state state;
233 unsigned int timeout;
234 unsigned long deadline;
237 struct __call_single_data csd;
242 * completion callback.
244 rq_end_io_fn *end_io;
248 static inline bool blk_op_is_scsi(unsigned int op)
250 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
253 static inline bool blk_op_is_private(unsigned int op)
255 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
258 static inline bool blk_rq_is_scsi(struct request *rq)
260 return blk_op_is_scsi(req_op(rq));
263 static inline bool blk_rq_is_private(struct request *rq)
265 return blk_op_is_private(req_op(rq));
268 static inline bool blk_rq_is_passthrough(struct request *rq)
270 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
273 static inline bool bio_is_passthrough(struct bio *bio)
275 unsigned op = bio_op(bio);
277 return blk_op_is_scsi(op) || blk_op_is_private(op);
280 static inline unsigned short req_get_ioprio(struct request *req)
285 #include <linux/elevator.h>
287 struct blk_queue_ctx;
291 enum blk_eh_timer_return {
292 BLK_EH_DONE, /* drivers has completed the command */
293 BLK_EH_RESET_TIMER, /* reset timer and try again */
296 enum blk_queue_state {
301 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
302 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
304 #define BLK_SCSI_MAX_CMDS (256)
305 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
308 * Zoned block device models (zoned limit).
310 * Note: This needs to be ordered from the least to the most severe
311 * restrictions for the inheritance in blk_stack_limits() to work.
313 enum blk_zoned_model {
314 BLK_ZONED_NONE = 0, /* Regular block device */
315 BLK_ZONED_HA, /* Host-aware zoned block device */
316 BLK_ZONED_HM, /* Host-managed zoned block device */
319 struct queue_limits {
320 unsigned long bounce_pfn;
321 unsigned long seg_boundary_mask;
322 unsigned long virt_boundary_mask;
324 unsigned int max_hw_sectors;
325 unsigned int max_dev_sectors;
326 unsigned int chunk_sectors;
327 unsigned int max_sectors;
328 unsigned int max_segment_size;
329 unsigned int physical_block_size;
330 unsigned int logical_block_size;
331 unsigned int alignment_offset;
334 unsigned int max_discard_sectors;
335 unsigned int max_hw_discard_sectors;
336 unsigned int max_write_same_sectors;
337 unsigned int max_write_zeroes_sectors;
338 unsigned int max_zone_append_sectors;
339 unsigned int discard_granularity;
340 unsigned int discard_alignment;
342 unsigned short max_segments;
343 unsigned short max_integrity_segments;
344 unsigned short max_discard_segments;
346 unsigned char misaligned;
347 unsigned char discard_misaligned;
348 unsigned char raid_partial_stripes_expensive;
349 enum blk_zoned_model zoned;
352 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
355 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
357 #ifdef CONFIG_BLK_DEV_ZONED
359 #define BLK_ALL_ZONES ((unsigned int)-1)
360 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
361 unsigned int nr_zones, report_zones_cb cb, void *data);
362 unsigned int blkdev_nr_zones(struct gendisk *disk);
363 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
364 sector_t sectors, sector_t nr_sectors,
366 int blk_revalidate_disk_zones(struct gendisk *disk,
367 void (*update_driver_data)(struct gendisk *disk));
369 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
370 unsigned int cmd, unsigned long arg);
371 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
372 unsigned int cmd, unsigned long arg);
374 #else /* CONFIG_BLK_DEV_ZONED */
376 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
381 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
382 fmode_t mode, unsigned int cmd,
388 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
389 fmode_t mode, unsigned int cmd,
395 #endif /* CONFIG_BLK_DEV_ZONED */
397 struct request_queue {
398 struct request *last_merge;
399 struct elevator_queue *elevator;
401 struct blk_queue_stats *stats;
402 struct rq_qos *rq_qos;
404 const struct blk_mq_ops *mq_ops;
407 struct blk_mq_ctx __percpu *queue_ctx;
409 unsigned int queue_depth;
411 /* hw dispatch queues */
412 struct blk_mq_hw_ctx **queue_hw_ctx;
413 unsigned int nr_hw_queues;
415 struct backing_dev_info *backing_dev_info;
418 * The queue owner gets to use this for whatever they like.
419 * ll_rw_blk doesn't touch it.
424 * various queue flags, see QUEUE_* below
426 unsigned long queue_flags;
428 * Number of contexts that have called blk_set_pm_only(). If this
429 * counter is above zero then only RQF_PM and RQF_PREEMPT requests are
435 * ida allocated id for this queue. Used to index queues from
441 * queue needs bounce pages for pages above this limit
445 spinlock_t queue_lock;
455 struct kobject *mq_kobj;
457 #ifdef CONFIG_BLK_DEV_INTEGRITY
458 struct blk_integrity integrity;
459 #endif /* CONFIG_BLK_DEV_INTEGRITY */
464 unsigned int nr_pending;
470 unsigned long nr_requests; /* Max # of requests */
472 unsigned int dma_pad_mask;
473 unsigned int dma_alignment;
475 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
476 /* Inline crypto capabilities */
477 struct blk_keyslot_manager *ksm;
480 unsigned int rq_timeout;
483 struct blk_stat_callback *poll_cb;
484 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
486 struct timer_list timeout;
487 struct work_struct timeout_work;
489 struct list_head icq_list;
490 #ifdef CONFIG_BLK_CGROUP
491 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
492 struct blkcg_gq *root_blkg;
493 struct list_head blkg_list;
496 struct queue_limits limits;
498 unsigned int required_elevator_features;
500 #ifdef CONFIG_BLK_DEV_ZONED
502 * Zoned block device information for request dispatch control.
503 * nr_zones is the total number of zones of the device. This is always
504 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
505 * bits which indicates if a zone is conventional (bit set) or
506 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
507 * bits which indicates if a zone is write locked, that is, if a write
508 * request targeting the zone was dispatched. All three fields are
509 * initialized by the low level device driver (e.g. scsi/sd.c).
510 * Stacking drivers (device mappers) may or may not initialize
513 * Reads of this information must be protected with blk_queue_enter() /
514 * blk_queue_exit(). Modifying this information is only allowed while
515 * no requests are being processed. See also blk_mq_freeze_queue() and
516 * blk_mq_unfreeze_queue().
518 unsigned int nr_zones;
519 unsigned long *conv_zones_bitmap;
520 unsigned long *seq_zones_wlock;
521 unsigned int max_open_zones;
522 unsigned int max_active_zones;
523 #endif /* CONFIG_BLK_DEV_ZONED */
528 unsigned int sg_timeout;
529 unsigned int sg_reserved_size;
531 struct mutex debugfs_mutex;
532 #ifdef CONFIG_BLK_DEV_IO_TRACE
533 struct blk_trace __rcu *blk_trace;
536 * for flush operations
538 struct blk_flush_queue *fq;
540 struct list_head requeue_list;
541 spinlock_t requeue_lock;
542 struct delayed_work requeue_work;
544 struct mutex sysfs_lock;
545 struct mutex sysfs_dir_lock;
548 * for reusing dead hctx instance in case of updating
551 struct list_head unused_hctx_list;
552 spinlock_t unused_hctx_lock;
556 #if defined(CONFIG_BLK_DEV_BSG)
557 struct bsg_class_device bsg_dev;
560 #ifdef CONFIG_BLK_DEV_THROTTLING
562 struct throtl_data *td;
564 struct rcu_head rcu_head;
565 wait_queue_head_t mq_freeze_wq;
567 * Protect concurrent access to q_usage_counter by
568 * percpu_ref_kill() and percpu_ref_reinit().
570 struct mutex mq_freeze_lock;
571 struct percpu_ref q_usage_counter;
573 struct blk_mq_tag_set *tag_set;
574 struct list_head tag_set_list;
575 struct bio_set bio_split;
577 struct dentry *debugfs_dir;
579 #ifdef CONFIG_BLK_DEBUG_FS
580 struct dentry *sched_debugfs_dir;
581 struct dentry *rqos_debugfs_dir;
584 bool mq_sysfs_init_done;
588 #define BLK_MAX_WRITE_HINTS 5
589 u64 write_hints[BLK_MAX_WRITE_HINTS];
592 /* Keep blk_queue_flag_name[] in sync with the definitions below */
593 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
594 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
595 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
596 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
597 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
598 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
599 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
600 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
601 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
602 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
603 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
604 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
605 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
606 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
607 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
608 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
609 #define QUEUE_FLAG_WC 17 /* Write back caching */
610 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
611 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
612 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
613 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
614 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
615 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
616 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
617 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
618 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
619 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
621 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
622 (1 << QUEUE_FLAG_SAME_COMP))
624 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
625 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
626 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
628 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
629 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
630 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
631 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
632 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
633 #define blk_queue_noxmerges(q) \
634 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
635 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
636 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
637 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
638 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
639 #define blk_queue_zone_resetall(q) \
640 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
641 #define blk_queue_secure_erase(q) \
642 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
643 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
644 #define blk_queue_scsi_passthrough(q) \
645 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
646 #define blk_queue_pci_p2pdma(q) \
647 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
648 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
649 #define blk_queue_rq_alloc_time(q) \
650 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
652 #define blk_queue_rq_alloc_time(q) false
655 #define blk_noretry_request(rq) \
656 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
657 REQ_FAILFAST_DRIVER))
658 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
659 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
660 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
661 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
663 extern void blk_set_pm_only(struct request_queue *q);
664 extern void blk_clear_pm_only(struct request_queue *q);
666 static inline bool blk_account_rq(struct request *rq)
668 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
671 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
673 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
675 #define rq_dma_dir(rq) \
676 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
678 #define dma_map_bvec(dev, bv, dir, attrs) \
679 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
682 static inline bool queue_is_mq(struct request_queue *q)
687 static inline enum blk_zoned_model
688 blk_queue_zoned_model(struct request_queue *q)
690 return q->limits.zoned;
693 static inline bool blk_queue_is_zoned(struct request_queue *q)
695 switch (blk_queue_zoned_model(q)) {
704 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
706 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
709 #ifdef CONFIG_BLK_DEV_ZONED
710 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
712 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
715 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
718 if (!blk_queue_is_zoned(q))
720 return sector >> ilog2(q->limits.chunk_sectors);
723 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
726 if (!blk_queue_is_zoned(q))
728 if (!q->conv_zones_bitmap)
730 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
733 static inline void blk_queue_max_open_zones(struct request_queue *q,
734 unsigned int max_open_zones)
736 q->max_open_zones = max_open_zones;
739 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
741 return q->max_open_zones;
744 static inline void blk_queue_max_active_zones(struct request_queue *q,
745 unsigned int max_active_zones)
747 q->max_active_zones = max_active_zones;
750 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
752 return q->max_active_zones;
754 #else /* CONFIG_BLK_DEV_ZONED */
755 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
759 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
764 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
769 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
773 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
777 #endif /* CONFIG_BLK_DEV_ZONED */
779 static inline bool rq_is_sync(struct request *rq)
781 return op_is_sync(rq->cmd_flags);
784 static inline bool rq_mergeable(struct request *rq)
786 if (blk_rq_is_passthrough(rq))
789 if (req_op(rq) == REQ_OP_FLUSH)
792 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
795 if (req_op(rq) == REQ_OP_ZONE_APPEND)
798 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
800 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
806 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
808 if (bio_page(a) == bio_page(b) &&
809 bio_offset(a) == bio_offset(b))
815 static inline unsigned int blk_queue_depth(struct request_queue *q)
818 return q->queue_depth;
820 return q->nr_requests;
823 extern unsigned long blk_max_low_pfn, blk_max_pfn;
826 * standard bounce addresses:
828 * BLK_BOUNCE_HIGH : bounce all highmem pages
829 * BLK_BOUNCE_ANY : don't bounce anything
830 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
833 #if BITS_PER_LONG == 32
834 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
836 #define BLK_BOUNCE_HIGH -1ULL
838 #define BLK_BOUNCE_ANY (-1ULL)
839 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
842 * default timeout for SG_IO if none specified
844 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
845 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
851 unsigned long offset;
856 struct req_iterator {
857 struct bvec_iter iter;
861 /* This should not be used directly - use rq_for_each_segment */
862 #define for_each_bio(_bio) \
863 for (; _bio; _bio = _bio->bi_next)
864 #define __rq_for_each_bio(_bio, rq) \
866 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
868 #define rq_for_each_segment(bvl, _rq, _iter) \
869 __rq_for_each_bio(_iter.bio, _rq) \
870 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
872 #define rq_for_each_bvec(bvl, _rq, _iter) \
873 __rq_for_each_bio(_iter.bio, _rq) \
874 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
876 #define rq_iter_last(bvec, _iter) \
877 (_iter.bio->bi_next == NULL && \
878 bio_iter_last(bvec, _iter.iter))
880 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
881 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
883 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
884 extern void rq_flush_dcache_pages(struct request *rq);
886 static inline void rq_flush_dcache_pages(struct request *rq)
891 extern int blk_register_queue(struct gendisk *disk);
892 extern void blk_unregister_queue(struct gendisk *disk);
893 blk_qc_t submit_bio_noacct(struct bio *bio);
894 extern void blk_rq_init(struct request_queue *q, struct request *rq);
895 extern void blk_put_request(struct request *);
896 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
897 blk_mq_req_flags_t flags);
898 extern int blk_lld_busy(struct request_queue *q);
899 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
900 struct bio_set *bs, gfp_t gfp_mask,
901 int (*bio_ctr)(struct bio *, struct bio *, void *),
903 extern void blk_rq_unprep_clone(struct request *rq);
904 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
906 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
907 extern void blk_queue_split(struct bio **);
908 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
909 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
910 unsigned int, void __user *);
911 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
912 unsigned int, void __user *);
913 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
914 struct scsi_ioctl_command __user *);
915 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
916 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
918 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
919 extern void blk_queue_exit(struct request_queue *q);
920 extern void blk_sync_queue(struct request_queue *q);
921 extern int blk_rq_map_user(struct request_queue *, struct request *,
922 struct rq_map_data *, void __user *, unsigned long,
924 extern int blk_rq_unmap_user(struct bio *);
925 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
926 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
927 struct rq_map_data *, const struct iov_iter *,
929 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
930 struct request *, int);
931 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
932 struct request *, int, rq_end_io_fn *);
934 /* Helper to convert REQ_OP_XXX to its string format XXX */
935 extern const char *blk_op_str(unsigned int op);
937 int blk_status_to_errno(blk_status_t status);
938 blk_status_t errno_to_blk_status(int errno);
940 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
942 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
944 return bdev->bd_disk->queue; /* this is never NULL */
948 * The basic unit of block I/O is a sector. It is used in a number of contexts
949 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
950 * bytes. Variables of type sector_t represent an offset or size that is a
951 * multiple of 512 bytes. Hence these two constants.
954 #define SECTOR_SHIFT 9
957 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
961 * blk_rq_pos() : the current sector
962 * blk_rq_bytes() : bytes left in the entire request
963 * blk_rq_cur_bytes() : bytes left in the current segment
964 * blk_rq_err_bytes() : bytes left till the next error boundary
965 * blk_rq_sectors() : sectors left in the entire request
966 * blk_rq_cur_sectors() : sectors left in the current segment
967 * blk_rq_stats_sectors() : sectors of the entire request used for stats
969 static inline sector_t blk_rq_pos(const struct request *rq)
974 static inline unsigned int blk_rq_bytes(const struct request *rq)
976 return rq->__data_len;
979 static inline int blk_rq_cur_bytes(const struct request *rq)
981 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
984 extern unsigned int blk_rq_err_bytes(const struct request *rq);
986 static inline unsigned int blk_rq_sectors(const struct request *rq)
988 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
991 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
993 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
996 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
998 return rq->stats_sectors;
1001 #ifdef CONFIG_BLK_DEV_ZONED
1003 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
1004 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
1006 static inline unsigned int blk_rq_zone_no(struct request *rq)
1008 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1011 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1013 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1015 #endif /* CONFIG_BLK_DEV_ZONED */
1018 * Some commands like WRITE SAME have a payload or data transfer size which
1019 * is different from the size of the request. Any driver that supports such
1020 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1021 * calculate the data transfer size.
1023 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1025 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1026 return rq->special_vec.bv_len;
1027 return blk_rq_bytes(rq);
1031 * Return the first full biovec in the request. The caller needs to check that
1032 * there are any bvecs before calling this helper.
1034 static inline struct bio_vec req_bvec(struct request *rq)
1036 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1037 return rq->special_vec;
1038 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1041 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1044 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1045 return min(q->limits.max_discard_sectors,
1046 UINT_MAX >> SECTOR_SHIFT);
1048 if (unlikely(op == REQ_OP_WRITE_SAME))
1049 return q->limits.max_write_same_sectors;
1051 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1052 return q->limits.max_write_zeroes_sectors;
1054 return q->limits.max_sectors;
1058 * Return maximum size of a request at given offset. Only valid for
1059 * file system requests.
1061 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1064 if (!q->limits.chunk_sectors)
1065 return q->limits.max_sectors;
1067 return min(q->limits.max_sectors, (unsigned int)(q->limits.chunk_sectors -
1068 (offset & (q->limits.chunk_sectors - 1))));
1071 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1074 struct request_queue *q = rq->q;
1076 if (blk_rq_is_passthrough(rq))
1077 return q->limits.max_hw_sectors;
1079 if (!q->limits.chunk_sectors ||
1080 req_op(rq) == REQ_OP_DISCARD ||
1081 req_op(rq) == REQ_OP_SECURE_ERASE)
1082 return blk_queue_get_max_sectors(q, req_op(rq));
1084 return min(blk_max_size_offset(q, offset),
1085 blk_queue_get_max_sectors(q, req_op(rq)));
1088 static inline unsigned int blk_rq_count_bios(struct request *rq)
1090 unsigned int nr_bios = 0;
1093 __rq_for_each_bio(bio, rq)
1099 void blk_steal_bios(struct bio_list *list, struct request *rq);
1102 * Request completion related functions.
1104 * blk_update_request() completes given number of bytes and updates
1105 * the request without completing it.
1107 extern bool blk_update_request(struct request *rq, blk_status_t error,
1108 unsigned int nr_bytes);
1110 extern void blk_abort_request(struct request *);
1113 * Access functions for manipulating queue properties
1115 extern void blk_cleanup_queue(struct request_queue *);
1116 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1117 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1118 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1119 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1120 extern void blk_queue_max_discard_segments(struct request_queue *,
1122 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1123 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1124 unsigned int max_discard_sectors);
1125 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1126 unsigned int max_write_same_sectors);
1127 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1128 unsigned int max_write_same_sectors);
1129 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1130 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1131 unsigned int max_zone_append_sectors);
1132 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1133 extern void blk_queue_alignment_offset(struct request_queue *q,
1134 unsigned int alignment);
1135 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1136 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1137 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1138 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1139 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1140 extern void blk_set_default_limits(struct queue_limits *lim);
1141 extern void blk_set_stacking_limits(struct queue_limits *lim);
1142 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1144 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1146 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1147 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1148 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1149 extern void blk_queue_dma_alignment(struct request_queue *, int);
1150 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1151 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1152 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1153 extern void blk_queue_required_elevator_features(struct request_queue *q,
1154 unsigned int features);
1155 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1156 struct device *dev);
1159 * Number of physical segments as sent to the device.
1161 * Normally this is the number of discontiguous data segments sent by the
1162 * submitter. But for data-less command like discard we might have no
1163 * actual data segments submitted, but the driver might have to add it's
1164 * own special payload. In that case we still return 1 here so that this
1165 * special payload will be mapped.
1167 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1169 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1171 return rq->nr_phys_segments;
1175 * Number of discard segments (or ranges) the driver needs to fill in.
1176 * Each discard bio merged into a request is counted as one segment.
1178 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1180 return max_t(unsigned short, rq->nr_phys_segments, 1);
1183 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1184 struct scatterlist *sglist, struct scatterlist **last_sg);
1185 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1186 struct scatterlist *sglist)
1188 struct scatterlist *last_sg = NULL;
1190 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1192 extern void blk_dump_rq_flags(struct request *, char *);
1194 bool __must_check blk_get_queue(struct request_queue *);
1195 struct request_queue *blk_alloc_queue(int node_id);
1196 extern void blk_put_queue(struct request_queue *);
1197 extern void blk_set_queue_dying(struct request_queue *);
1201 * blk_plug permits building a queue of related requests by holding the I/O
1202 * fragments for a short period. This allows merging of sequential requests
1203 * into single larger request. As the requests are moved from a per-task list to
1204 * the device's request_queue in a batch, this results in improved scalability
1205 * as the lock contention for request_queue lock is reduced.
1207 * It is ok not to disable preemption when adding the request to the plug list
1208 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1209 * the plug list when the task sleeps by itself. For details, please see
1210 * schedule() where blk_schedule_flush_plug() is called.
1213 struct list_head mq_list; /* blk-mq requests */
1214 struct list_head cb_list; /* md requires an unplug callback */
1215 unsigned short rq_count;
1216 bool multiple_queues;
1219 #define BLK_MAX_REQUEST_COUNT 16
1220 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1223 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1224 struct blk_plug_cb {
1225 struct list_head list;
1226 blk_plug_cb_fn callback;
1229 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1230 void *data, int size);
1231 extern void blk_start_plug(struct blk_plug *);
1232 extern void blk_finish_plug(struct blk_plug *);
1233 extern void blk_flush_plug_list(struct blk_plug *, bool);
1235 static inline void blk_flush_plug(struct task_struct *tsk)
1237 struct blk_plug *plug = tsk->plug;
1240 blk_flush_plug_list(plug, false);
1243 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1245 struct blk_plug *plug = tsk->plug;
1248 blk_flush_plug_list(plug, true);
1251 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1253 struct blk_plug *plug = tsk->plug;
1256 (!list_empty(&plug->mq_list) ||
1257 !list_empty(&plug->cb_list));
1260 int blkdev_issue_flush(struct block_device *, gfp_t);
1261 long nr_blockdev_pages(void);
1262 #else /* CONFIG_BLOCK */
1266 static inline void blk_start_plug(struct blk_plug *plug)
1270 static inline void blk_finish_plug(struct blk_plug *plug)
1274 static inline void blk_flush_plug(struct task_struct *task)
1278 static inline void blk_schedule_flush_plug(struct task_struct *task)
1283 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1288 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask)
1293 static inline long nr_blockdev_pages(void)
1297 #endif /* CONFIG_BLOCK */
1299 extern void blk_io_schedule(void);
1301 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1302 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1304 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1306 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1307 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1308 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1309 sector_t nr_sects, gfp_t gfp_mask, int flags,
1312 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1313 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1315 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1316 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1318 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1319 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1321 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1322 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1324 return blkdev_issue_discard(sb->s_bdev,
1325 block << (sb->s_blocksize_bits -
1327 nr_blocks << (sb->s_blocksize_bits -
1331 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1332 sector_t nr_blocks, gfp_t gfp_mask)
1334 return blkdev_issue_zeroout(sb->s_bdev,
1335 block << (sb->s_blocksize_bits -
1337 nr_blocks << (sb->s_blocksize_bits -
1342 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1344 enum blk_default_limits {
1345 BLK_MAX_SEGMENTS = 128,
1346 BLK_SAFE_MAX_SECTORS = 255,
1347 BLK_DEF_MAX_SECTORS = 2560,
1348 BLK_MAX_SEGMENT_SIZE = 65536,
1349 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1352 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1354 return q->limits.seg_boundary_mask;
1357 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1359 return q->limits.virt_boundary_mask;
1362 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1364 return q->limits.max_sectors;
1367 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1369 return q->limits.max_hw_sectors;
1372 static inline unsigned short queue_max_segments(const struct request_queue *q)
1374 return q->limits.max_segments;
1377 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1379 return q->limits.max_discard_segments;
1382 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1384 return q->limits.max_segment_size;
1387 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1389 return q->limits.max_zone_append_sectors;
1392 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1396 if (q && q->limits.logical_block_size)
1397 retval = q->limits.logical_block_size;
1402 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1404 return queue_logical_block_size(bdev_get_queue(bdev));
1407 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1409 return q->limits.physical_block_size;
1412 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1414 return queue_physical_block_size(bdev_get_queue(bdev));
1417 static inline unsigned int queue_io_min(const struct request_queue *q)
1419 return q->limits.io_min;
1422 static inline int bdev_io_min(struct block_device *bdev)
1424 return queue_io_min(bdev_get_queue(bdev));
1427 static inline unsigned int queue_io_opt(const struct request_queue *q)
1429 return q->limits.io_opt;
1432 static inline int bdev_io_opt(struct block_device *bdev)
1434 return queue_io_opt(bdev_get_queue(bdev));
1437 static inline int queue_alignment_offset(const struct request_queue *q)
1439 if (q->limits.misaligned)
1442 return q->limits.alignment_offset;
1445 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1447 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1448 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1451 return (granularity + lim->alignment_offset - alignment) % granularity;
1454 static inline int bdev_alignment_offset(struct block_device *bdev)
1456 struct request_queue *q = bdev_get_queue(bdev);
1458 if (q->limits.misaligned)
1461 if (bdev != bdev->bd_contains)
1462 return bdev->bd_part->alignment_offset;
1464 return q->limits.alignment_offset;
1467 static inline int queue_discard_alignment(const struct request_queue *q)
1469 if (q->limits.discard_misaligned)
1472 return q->limits.discard_alignment;
1475 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1477 unsigned int alignment, granularity, offset;
1479 if (!lim->max_discard_sectors)
1482 /* Why are these in bytes, not sectors? */
1483 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1484 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1488 /* Offset of the partition start in 'granularity' sectors */
1489 offset = sector_div(sector, granularity);
1491 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1492 offset = (granularity + alignment - offset) % granularity;
1494 /* Turn it back into bytes, gaah */
1495 return offset << SECTOR_SHIFT;
1498 static inline int bdev_discard_alignment(struct block_device *bdev)
1500 struct request_queue *q = bdev_get_queue(bdev);
1502 if (bdev != bdev->bd_contains)
1503 return bdev->bd_part->discard_alignment;
1505 return q->limits.discard_alignment;
1508 static inline unsigned int bdev_write_same(struct block_device *bdev)
1510 struct request_queue *q = bdev_get_queue(bdev);
1513 return q->limits.max_write_same_sectors;
1518 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1520 struct request_queue *q = bdev_get_queue(bdev);
1523 return q->limits.max_write_zeroes_sectors;
1528 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1530 struct request_queue *q = bdev_get_queue(bdev);
1533 return blk_queue_zoned_model(q);
1535 return BLK_ZONED_NONE;
1538 static inline bool bdev_is_zoned(struct block_device *bdev)
1540 struct request_queue *q = bdev_get_queue(bdev);
1543 return blk_queue_is_zoned(q);
1548 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1550 struct request_queue *q = bdev_get_queue(bdev);
1553 return blk_queue_zone_sectors(q);
1557 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1559 struct request_queue *q = bdev_get_queue(bdev);
1562 return queue_max_open_zones(q);
1566 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1568 struct request_queue *q = bdev_get_queue(bdev);
1571 return queue_max_active_zones(q);
1575 static inline int queue_dma_alignment(const struct request_queue *q)
1577 return q ? q->dma_alignment : 511;
1580 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1583 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1584 return !(addr & alignment) && !(len & alignment);
1587 /* assumes size > 256 */
1588 static inline unsigned int blksize_bits(unsigned int size)
1590 unsigned int bits = 8;
1594 } while (size > 256);
1598 static inline unsigned int block_size(struct block_device *bdev)
1600 return 1 << bdev->bd_inode->i_blkbits;
1603 int kblockd_schedule_work(struct work_struct *work);
1604 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1606 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1607 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1608 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1609 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1611 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1613 enum blk_integrity_flags {
1614 BLK_INTEGRITY_VERIFY = 1 << 0,
1615 BLK_INTEGRITY_GENERATE = 1 << 1,
1616 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1617 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1620 struct blk_integrity_iter {
1624 unsigned int data_size;
1625 unsigned short interval;
1626 const char *disk_name;
1629 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1630 typedef void (integrity_prepare_fn) (struct request *);
1631 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1633 struct blk_integrity_profile {
1634 integrity_processing_fn *generate_fn;
1635 integrity_processing_fn *verify_fn;
1636 integrity_prepare_fn *prepare_fn;
1637 integrity_complete_fn *complete_fn;
1641 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1642 extern void blk_integrity_unregister(struct gendisk *);
1643 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1644 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1645 struct scatterlist *);
1646 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1647 extern bool blk_integrity_merge_rq(struct request_queue *, struct request *,
1649 extern bool blk_integrity_merge_bio(struct request_queue *, struct request *,
1652 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1654 struct blk_integrity *bi = &disk->queue->integrity;
1663 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1665 return blk_get_integrity(bdev->bd_disk);
1669 blk_integrity_queue_supports_integrity(struct request_queue *q)
1671 return q->integrity.profile;
1674 static inline bool blk_integrity_rq(struct request *rq)
1676 return rq->cmd_flags & REQ_INTEGRITY;
1679 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1682 q->limits.max_integrity_segments = segs;
1685 static inline unsigned short
1686 queue_max_integrity_segments(const struct request_queue *q)
1688 return q->limits.max_integrity_segments;
1692 * bio_integrity_intervals - Return number of integrity intervals for a bio
1693 * @bi: blk_integrity profile for device
1694 * @sectors: Size of the bio in 512-byte sectors
1696 * Description: The block layer calculates everything in 512 byte
1697 * sectors but integrity metadata is done in terms of the data integrity
1698 * interval size of the storage device. Convert the block layer sectors
1699 * to the appropriate number of integrity intervals.
1701 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1702 unsigned int sectors)
1704 return sectors >> (bi->interval_exp - 9);
1707 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1708 unsigned int sectors)
1710 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1714 * Return the first bvec that contains integrity data. Only drivers that are
1715 * limited to a single integrity segment should use this helper.
1717 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1719 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1721 return rq->bio->bi_integrity->bip_vec;
1724 #else /* CONFIG_BLK_DEV_INTEGRITY */
1727 struct block_device;
1729 struct blk_integrity;
1731 static inline int blk_integrity_rq(struct request *rq)
1735 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1740 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1742 struct scatterlist *s)
1746 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1750 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1755 blk_integrity_queue_supports_integrity(struct request_queue *q)
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(const 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 unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1792 unsigned int sectors)
1797 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1798 unsigned int sectors)
1803 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1808 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1810 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1812 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1814 void blk_ksm_unregister(struct request_queue *q);
1816 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1818 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1819 struct request_queue *q)
1824 static inline void blk_ksm_unregister(struct request_queue *q) { }
1826 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1829 struct block_device_operations {
1830 blk_qc_t (*submit_bio) (struct bio *bio);
1831 int (*open) (struct block_device *, fmode_t);
1832 void (*release) (struct gendisk *, fmode_t);
1833 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1834 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1835 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1836 unsigned int (*check_events) (struct gendisk *disk,
1837 unsigned int clearing);
1838 void (*unlock_native_capacity) (struct gendisk *);
1839 int (*revalidate_disk) (struct gendisk *);
1840 int (*getgeo)(struct block_device *, struct hd_geometry *);
1841 /* this callback is with swap_lock and sometimes page table lock held */
1842 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1843 int (*report_zones)(struct gendisk *, sector_t sector,
1844 unsigned int nr_zones, report_zones_cb cb, void *data);
1845 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1846 struct module *owner;
1847 const struct pr_ops *pr_ops;
1850 #ifdef CONFIG_COMPAT
1851 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1852 unsigned int, unsigned long);
1854 #define blkdev_compat_ptr_ioctl NULL
1857 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1859 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1860 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1861 struct writeback_control *);
1863 #ifdef CONFIG_BLK_DEV_ZONED
1864 bool blk_req_needs_zone_write_lock(struct request *rq);
1865 bool blk_req_zone_write_trylock(struct request *rq);
1866 void __blk_req_zone_write_lock(struct request *rq);
1867 void __blk_req_zone_write_unlock(struct request *rq);
1869 static inline void blk_req_zone_write_lock(struct request *rq)
1871 if (blk_req_needs_zone_write_lock(rq))
1872 __blk_req_zone_write_lock(rq);
1875 static inline void blk_req_zone_write_unlock(struct request *rq)
1877 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1878 __blk_req_zone_write_unlock(rq);
1881 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1883 return rq->q->seq_zones_wlock &&
1884 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1887 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1889 if (!blk_req_needs_zone_write_lock(rq))
1891 return !blk_req_zone_is_write_locked(rq);
1894 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1899 static inline void blk_req_zone_write_lock(struct request *rq)
1903 static inline void blk_req_zone_write_unlock(struct request *rq)
1906 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1911 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1915 #endif /* CONFIG_BLK_DEV_ZONED */
1917 static inline void blk_wake_io_task(struct task_struct *waiter)
1920 * If we're polling, the task itself is doing the completions. For
1921 * that case, we don't need to signal a wakeup, it's enough to just
1922 * mark us as RUNNING.
1924 if (waiter == current)
1925 __set_current_state(TASK_RUNNING);
1927 wake_up_process(waiter);
1930 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1932 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1933 unsigned long start_time);
1936 * bio_start_io_acct - start I/O accounting for bio based drivers
1937 * @bio: bio to start account for
1939 * Returns the start time that should be passed back to bio_end_io_acct().
1941 static inline unsigned long bio_start_io_acct(struct bio *bio)
1943 return disk_start_io_acct(bio->bi_disk, bio_sectors(bio), bio_op(bio));
1947 * bio_end_io_acct - end I/O accounting for bio based drivers
1948 * @bio: bio to end account for
1949 * @start: start time returned by bio_start_io_acct()
1951 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1953 return disk_end_io_acct(bio->bi_disk, bio_op(bio), start_time);
1956 int bdev_read_only(struct block_device *bdev);
1957 int set_blocksize(struct block_device *bdev, int size);
1959 const char *bdevname(struct block_device *bdev, char *buffer);
1960 struct block_device *lookup_bdev(const char *);
1962 void blkdev_show(struct seq_file *seqf, off_t offset);
1964 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1965 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1967 #define BLKDEV_MAJOR_MAX 512
1969 #define BLKDEV_MAJOR_MAX 0
1972 int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
1973 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1975 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1976 int bd_prepare_to_claim(struct block_device *bdev, struct block_device *whole,
1978 void bd_abort_claiming(struct block_device *bdev, struct block_device *whole,
1980 void blkdev_put(struct block_device *bdev, fmode_t mode);
1982 struct block_device *I_BDEV(struct inode *inode);
1983 struct block_device *bdget(dev_t);
1984 struct block_device *bdgrab(struct block_device *bdev);
1985 void bdput(struct block_device *);
1988 void invalidate_bdev(struct block_device *bdev);
1989 int sync_blockdev(struct block_device *bdev);
1991 static inline void invalidate_bdev(struct block_device *bdev)
1994 static inline int sync_blockdev(struct block_device *bdev)
1999 int fsync_bdev(struct block_device *bdev);
2001 struct super_block *freeze_bdev(struct block_device *bdev);
2002 int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2004 #endif /* _LINUX_BLKDEV_H */