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/minmax.h>
12 #include <linux/timer.h>
13 #include <linux/workqueue.h>
14 #include <linux/wait.h>
15 #include <linux/mempool.h>
16 #include <linux/pfn.h>
17 #include <linux/bio.h>
18 #include <linux/stringify.h>
19 #include <linux/gfp.h>
20 #include <linux/smp.h>
21 #include <linux/rcupdate.h>
22 #include <linux/percpu-refcount.h>
23 #include <linux/scatterlist.h>
24 #include <linux/blkzoned.h>
26 #include <linux/sbitmap.h>
30 struct elevator_queue;
35 struct blk_flush_queue;
38 struct blk_queue_stats;
39 struct blk_stat_callback;
40 struct blk_keyslot_manager;
42 #define BLKDEV_MIN_RQ 4
43 #define BLKDEV_MAX_RQ 128 /* Default maximum */
45 /* Must be consistent with blk_mq_poll_stats_bkt() */
46 #define BLK_MQ_POLL_STATS_BKTS 16
48 /* Doing classic polling */
49 #define BLK_MQ_POLL_CLASSIC -1
52 * Maximum number of blkcg policies allowed to be registered concurrently.
53 * Defined here to simplify include dependency.
55 #define BLKCG_MAX_POLS 6
57 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
61 typedef __u32 __bitwise req_flags_t;
63 /* drive already may have started this one */
64 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
65 /* may not be passed by ioscheduler */
66 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
67 /* request for flush sequence */
68 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
69 /* merge of different types, fail separately */
70 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
71 /* track inflight for MQ */
72 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
73 /* don't call prep for this one */
74 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
75 /* vaguely specified driver internal error. Ignored by the block layer */
76 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
77 /* don't warn about errors */
78 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
79 /* elevator private data attached */
80 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
81 /* account into disk and partition IO statistics */
82 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
83 /* runtime pm request */
84 #define RQF_PM ((__force req_flags_t)(1 << 15))
85 /* on IO scheduler merge hash */
86 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
87 /* track IO completion time */
88 #define RQF_STATS ((__force req_flags_t)(1 << 17))
89 /* Look at ->special_vec for the actual data payload instead of the
91 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
92 /* The per-zone write lock is held for this request */
93 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
94 /* already slept for hybrid poll */
95 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
96 /* ->timeout has been called, don't expire again */
97 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
99 /* flags that prevent us from merging requests: */
100 #define RQF_NOMERGE_FLAGS \
101 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
104 * Request state for blk-mq.
113 * Try to put the fields that are referenced together in the same cacheline.
115 * If you modify this structure, make sure to update blk_rq_init() and
116 * especially blk_mq_rq_ctx_init() to take care of the added fields.
119 struct request_queue *q;
120 struct blk_mq_ctx *mq_ctx;
121 struct blk_mq_hw_ctx *mq_hctx;
123 unsigned int cmd_flags; /* op and common flags */
124 req_flags_t rq_flags;
129 /* the following two fields are internal, NEVER access directly */
130 unsigned int __data_len; /* total data len */
131 sector_t __sector; /* sector cursor */
136 struct list_head queuelist;
139 * The hash is used inside the scheduler, and killed once the
140 * request reaches the dispatch list. The ipi_list is only used
141 * to queue the request for softirq completion, which is long
142 * after the request has been unhashed (and even removed from
143 * the dispatch list).
146 struct hlist_node hash; /* merge hash */
147 struct llist_node ipi_list;
151 * The rb_node is only used inside the io scheduler, requests
152 * are pruned when moved to the dispatch queue. So let the
153 * completion_data share space with the rb_node.
156 struct rb_node rb_node; /* sort/lookup */
157 struct bio_vec special_vec;
158 void *completion_data;
159 int error_count; /* for legacy drivers, don't use */
163 * Three pointers are available for the IO schedulers, if they need
164 * more they have to dynamically allocate it. Flush requests are
165 * never put on the IO scheduler. So let the flush fields share
166 * space with the elevator data.
176 struct list_head list;
177 rq_end_io_fn *saved_end_io;
181 struct gendisk *rq_disk;
182 struct block_device *part;
183 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
184 /* Time that the first bio started allocating this request. */
187 /* Time that this request was allocated for this IO. */
189 /* Time that I/O was submitted to the device. */
190 u64 io_start_time_ns;
192 #ifdef CONFIG_BLK_WBT
193 unsigned short wbt_flags;
196 * rq sectors used for blk stats. It has the same value
197 * with blk_rq_sectors(rq), except that it never be zeroed
200 unsigned short stats_sectors;
203 * Number of scatter-gather DMA addr+len pairs after
204 * physical address coalescing is performed.
206 unsigned short nr_phys_segments;
208 #if defined(CONFIG_BLK_DEV_INTEGRITY)
209 unsigned short nr_integrity_segments;
212 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
213 struct bio_crypt_ctx *crypt_ctx;
214 struct blk_ksm_keyslot *crypt_keyslot;
217 unsigned short write_hint;
218 unsigned short ioprio;
220 enum mq_rq_state state;
223 unsigned int timeout;
224 unsigned long deadline;
227 struct __call_single_data csd;
232 * completion callback.
234 rq_end_io_fn *end_io;
238 static inline int blk_validate_block_size(unsigned int bsize)
240 if (bsize < 512 || bsize > PAGE_SIZE || !is_power_of_2(bsize))
246 static inline bool blk_op_is_passthrough(unsigned int op)
249 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
252 static inline bool blk_rq_is_passthrough(struct request *rq)
254 return blk_op_is_passthrough(req_op(rq));
257 static inline unsigned short req_get_ioprio(struct request *req)
262 #include <linux/elevator.h>
266 enum blk_eh_timer_return {
267 BLK_EH_DONE, /* drivers has completed the command */
268 BLK_EH_RESET_TIMER, /* reset timer and try again */
271 enum blk_queue_state {
276 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
277 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
280 * Zoned block device models (zoned limit).
282 * Note: This needs to be ordered from the least to the most severe
283 * restrictions for the inheritance in blk_stack_limits() to work.
285 enum blk_zoned_model {
286 BLK_ZONED_NONE = 0, /* Regular block device */
287 BLK_ZONED_HA, /* Host-aware zoned block device */
288 BLK_ZONED_HM, /* Host-managed zoned block device */
292 * BLK_BOUNCE_NONE: never bounce (default)
293 * BLK_BOUNCE_HIGH: bounce all highmem pages
300 struct queue_limits {
301 enum blk_bounce bounce;
302 unsigned long seg_boundary_mask;
303 unsigned long virt_boundary_mask;
305 unsigned int max_hw_sectors;
306 unsigned int max_dev_sectors;
307 unsigned int chunk_sectors;
308 unsigned int max_sectors;
309 unsigned int max_segment_size;
310 unsigned int physical_block_size;
311 unsigned int logical_block_size;
312 unsigned int alignment_offset;
315 unsigned int max_discard_sectors;
316 unsigned int max_hw_discard_sectors;
317 unsigned int max_write_same_sectors;
318 unsigned int max_write_zeroes_sectors;
319 unsigned int max_zone_append_sectors;
320 unsigned int discard_granularity;
321 unsigned int discard_alignment;
322 unsigned int zone_write_granularity;
324 unsigned short max_segments;
325 unsigned short max_integrity_segments;
326 unsigned short max_discard_segments;
328 unsigned char misaligned;
329 unsigned char discard_misaligned;
330 unsigned char raid_partial_stripes_expensive;
331 enum blk_zoned_model zoned;
334 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
337 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
339 #ifdef CONFIG_BLK_DEV_ZONED
341 #define BLK_ALL_ZONES ((unsigned int)-1)
342 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
343 unsigned int nr_zones, report_zones_cb cb, void *data);
344 unsigned int blkdev_nr_zones(struct gendisk *disk);
345 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
346 sector_t sectors, sector_t nr_sectors,
348 int blk_revalidate_disk_zones(struct gendisk *disk,
349 void (*update_driver_data)(struct gendisk *disk));
351 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
352 unsigned int cmd, unsigned long arg);
353 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
354 unsigned int cmd, unsigned long arg);
356 #else /* CONFIG_BLK_DEV_ZONED */
358 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
363 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
364 fmode_t mode, unsigned int cmd,
370 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
371 fmode_t mode, unsigned int cmd,
377 #endif /* CONFIG_BLK_DEV_ZONED */
379 struct request_queue {
380 struct request *last_merge;
381 struct elevator_queue *elevator;
383 struct percpu_ref q_usage_counter;
385 struct blk_queue_stats *stats;
386 struct rq_qos *rq_qos;
388 const struct blk_mq_ops *mq_ops;
391 struct blk_mq_ctx __percpu *queue_ctx;
393 unsigned int queue_depth;
395 /* hw dispatch queues */
396 struct blk_mq_hw_ctx **queue_hw_ctx;
397 unsigned int nr_hw_queues;
400 * The queue owner gets to use this for whatever they like.
401 * ll_rw_blk doesn't touch it.
406 * various queue flags, see QUEUE_* below
408 unsigned long queue_flags;
410 * Number of contexts that have called blk_set_pm_only(). If this
411 * counter is above zero then only RQF_PM requests are processed.
416 * ida allocated id for this queue. Used to index queues from
421 spinlock_t queue_lock;
423 struct gendisk *disk;
433 struct kobject *mq_kobj;
435 #ifdef CONFIG_BLK_DEV_INTEGRITY
436 struct blk_integrity integrity;
437 #endif /* CONFIG_BLK_DEV_INTEGRITY */
441 enum rpm_status rpm_status;
447 unsigned long nr_requests; /* Max # of requests */
449 unsigned int dma_pad_mask;
450 unsigned int dma_alignment;
452 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
453 /* Inline crypto capabilities */
454 struct blk_keyslot_manager *ksm;
457 unsigned int rq_timeout;
460 struct blk_stat_callback *poll_cb;
461 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
463 struct timer_list timeout;
464 struct work_struct timeout_work;
466 atomic_t nr_active_requests_shared_sbitmap;
468 struct sbitmap_queue sched_bitmap_tags;
469 struct sbitmap_queue sched_breserved_tags;
471 struct list_head icq_list;
472 #ifdef CONFIG_BLK_CGROUP
473 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
474 struct blkcg_gq *root_blkg;
475 struct list_head blkg_list;
478 struct queue_limits limits;
480 unsigned int required_elevator_features;
482 #ifdef CONFIG_BLK_DEV_ZONED
484 * Zoned block device information for request dispatch control.
485 * nr_zones is the total number of zones of the device. This is always
486 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
487 * bits which indicates if a zone is conventional (bit set) or
488 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
489 * bits which indicates if a zone is write locked, that is, if a write
490 * request targeting the zone was dispatched. All three fields are
491 * initialized by the low level device driver (e.g. scsi/sd.c).
492 * Stacking drivers (device mappers) may or may not initialize
495 * Reads of this information must be protected with blk_queue_enter() /
496 * blk_queue_exit(). Modifying this information is only allowed while
497 * no requests are being processed. See also blk_mq_freeze_queue() and
498 * blk_mq_unfreeze_queue().
500 unsigned int nr_zones;
501 unsigned long *conv_zones_bitmap;
502 unsigned long *seq_zones_wlock;
503 unsigned int max_open_zones;
504 unsigned int max_active_zones;
505 #endif /* CONFIG_BLK_DEV_ZONED */
508 struct mutex debugfs_mutex;
509 #ifdef CONFIG_BLK_DEV_IO_TRACE
510 struct blk_trace __rcu *blk_trace;
513 * for flush operations
515 struct blk_flush_queue *fq;
517 struct list_head requeue_list;
518 spinlock_t requeue_lock;
519 struct delayed_work requeue_work;
521 struct mutex sysfs_lock;
522 struct mutex sysfs_dir_lock;
525 * for reusing dead hctx instance in case of updating
528 struct list_head unused_hctx_list;
529 spinlock_t unused_hctx_lock;
533 #ifdef CONFIG_BLK_DEV_THROTTLING
535 struct throtl_data *td;
537 struct rcu_head rcu_head;
538 wait_queue_head_t mq_freeze_wq;
540 * Protect concurrent access to q_usage_counter by
541 * percpu_ref_kill() and percpu_ref_reinit().
543 struct mutex mq_freeze_lock;
545 struct blk_mq_tag_set *tag_set;
546 struct list_head tag_set_list;
547 struct bio_set bio_split;
549 struct dentry *debugfs_dir;
551 #ifdef CONFIG_BLK_DEBUG_FS
552 struct dentry *sched_debugfs_dir;
553 struct dentry *rqos_debugfs_dir;
556 bool mq_sysfs_init_done;
560 #define BLK_MAX_WRITE_HINTS 5
561 u64 write_hints[BLK_MAX_WRITE_HINTS];
564 /* Keep blk_queue_flag_name[] in sync with the definitions below */
565 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
566 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
567 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
568 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
569 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
570 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
571 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
572 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
573 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
574 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
575 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
576 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
577 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
578 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
579 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
580 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
581 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
582 #define QUEUE_FLAG_WC 17 /* Write back caching */
583 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
584 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
585 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
586 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
587 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
588 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
589 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
590 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
591 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
592 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
593 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
594 #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */
596 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
597 (1 << QUEUE_FLAG_SAME_COMP) | \
598 (1 << QUEUE_FLAG_NOWAIT))
600 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
601 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
602 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
604 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
605 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
606 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
607 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
608 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
609 #define blk_queue_noxmerges(q) \
610 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
611 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
612 #define blk_queue_stable_writes(q) \
613 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
614 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
615 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
616 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
617 #define blk_queue_zone_resetall(q) \
618 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
619 #define blk_queue_secure_erase(q) \
620 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
621 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
622 #define blk_queue_scsi_passthrough(q) \
623 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
624 #define blk_queue_pci_p2pdma(q) \
625 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
626 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
627 #define blk_queue_rq_alloc_time(q) \
628 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
630 #define blk_queue_rq_alloc_time(q) false
633 #define blk_noretry_request(rq) \
634 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
635 REQ_FAILFAST_DRIVER))
636 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
637 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
638 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
639 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
640 #define blk_queue_nowait(q) test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
642 extern void blk_set_pm_only(struct request_queue *q);
643 extern void blk_clear_pm_only(struct request_queue *q);
645 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
647 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
649 #define rq_dma_dir(rq) \
650 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
652 #define dma_map_bvec(dev, bv, dir, attrs) \
653 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
656 static inline bool queue_is_mq(struct request_queue *q)
662 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
664 return q->rpm_status;
667 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
673 static inline enum blk_zoned_model
674 blk_queue_zoned_model(struct request_queue *q)
676 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
677 return q->limits.zoned;
678 return BLK_ZONED_NONE;
681 static inline bool blk_queue_is_zoned(struct request_queue *q)
683 switch (blk_queue_zoned_model(q)) {
692 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
694 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
697 #ifdef CONFIG_BLK_DEV_ZONED
698 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
700 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
703 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
706 if (!blk_queue_is_zoned(q))
708 return sector >> ilog2(q->limits.chunk_sectors);
711 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
714 if (!blk_queue_is_zoned(q))
716 if (!q->conv_zones_bitmap)
718 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
721 static inline void blk_queue_max_open_zones(struct request_queue *q,
722 unsigned int max_open_zones)
724 q->max_open_zones = max_open_zones;
727 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
729 return q->max_open_zones;
732 static inline void blk_queue_max_active_zones(struct request_queue *q,
733 unsigned int max_active_zones)
735 q->max_active_zones = max_active_zones;
738 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
740 return q->max_active_zones;
742 #else /* CONFIG_BLK_DEV_ZONED */
743 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
747 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
752 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
757 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
761 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
765 #endif /* CONFIG_BLK_DEV_ZONED */
767 static inline bool rq_is_sync(struct request *rq)
769 return op_is_sync(rq->cmd_flags);
772 static inline bool rq_mergeable(struct request *rq)
774 if (blk_rq_is_passthrough(rq))
777 if (req_op(rq) == REQ_OP_FLUSH)
780 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
783 if (req_op(rq) == REQ_OP_ZONE_APPEND)
786 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
788 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
794 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
796 if (bio_page(a) == bio_page(b) &&
797 bio_offset(a) == bio_offset(b))
803 static inline unsigned int blk_queue_depth(struct request_queue *q)
806 return q->queue_depth;
808 return q->nr_requests;
812 * default timeout for SG_IO if none specified
814 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
815 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
821 unsigned long offset;
826 struct req_iterator {
827 struct bvec_iter iter;
831 /* This should not be used directly - use rq_for_each_segment */
832 #define for_each_bio(_bio) \
833 for (; _bio; _bio = _bio->bi_next)
834 #define __rq_for_each_bio(_bio, rq) \
836 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
838 #define rq_for_each_segment(bvl, _rq, _iter) \
839 __rq_for_each_bio(_iter.bio, _rq) \
840 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
842 #define rq_for_each_bvec(bvl, _rq, _iter) \
843 __rq_for_each_bio(_iter.bio, _rq) \
844 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
846 #define rq_iter_last(bvec, _iter) \
847 (_iter.bio->bi_next == NULL && \
848 bio_iter_last(bvec, _iter.iter))
850 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
851 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
853 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
854 extern void rq_flush_dcache_pages(struct request *rq);
856 static inline void rq_flush_dcache_pages(struct request *rq)
861 extern int blk_register_queue(struct gendisk *disk);
862 extern void blk_unregister_queue(struct gendisk *disk);
863 blk_qc_t submit_bio_noacct(struct bio *bio);
864 extern void blk_rq_init(struct request_queue *q, struct request *rq);
865 extern void blk_put_request(struct request *);
866 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
867 blk_mq_req_flags_t flags);
868 extern int blk_lld_busy(struct request_queue *q);
869 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
870 struct bio_set *bs, gfp_t gfp_mask,
871 int (*bio_ctr)(struct bio *, struct bio *, void *),
873 extern void blk_rq_unprep_clone(struct request *rq);
874 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
876 int blk_rq_append_bio(struct request *rq, struct bio *bio);
877 extern void blk_queue_split(struct bio **);
878 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
879 extern void blk_queue_exit(struct request_queue *q);
880 extern void blk_sync_queue(struct request_queue *q);
881 extern int blk_rq_map_user(struct request_queue *, struct request *,
882 struct rq_map_data *, void __user *, unsigned long,
884 extern int blk_rq_unmap_user(struct bio *);
885 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
886 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
887 struct rq_map_data *, const struct iov_iter *,
889 extern void blk_execute_rq_nowait(struct gendisk *,
890 struct request *, int, rq_end_io_fn *);
892 blk_status_t blk_execute_rq(struct gendisk *bd_disk, struct request *rq,
895 /* Helper to convert REQ_OP_XXX to its string format XXX */
896 extern const char *blk_op_str(unsigned int op);
898 int blk_status_to_errno(blk_status_t status);
899 blk_status_t errno_to_blk_status(int errno);
901 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
903 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
905 return bdev->bd_disk->queue; /* this is never NULL */
909 * The basic unit of block I/O is a sector. It is used in a number of contexts
910 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
911 * bytes. Variables of type sector_t represent an offset or size that is a
912 * multiple of 512 bytes. Hence these two constants.
915 #define SECTOR_SHIFT 9
918 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
921 #define PAGE_SECTORS_SHIFT (PAGE_SHIFT - SECTOR_SHIFT)
922 #define PAGE_SECTORS (1 << PAGE_SECTORS_SHIFT)
923 #define SECTOR_MASK (PAGE_SECTORS - 1)
926 * blk_rq_pos() : the current sector
927 * blk_rq_bytes() : bytes left in the entire request
928 * blk_rq_cur_bytes() : bytes left in the current segment
929 * blk_rq_err_bytes() : bytes left till the next error boundary
930 * blk_rq_sectors() : sectors left in the entire request
931 * blk_rq_cur_sectors() : sectors left in the current segment
932 * blk_rq_stats_sectors() : sectors of the entire request used for stats
934 static inline sector_t blk_rq_pos(const struct request *rq)
939 static inline unsigned int blk_rq_bytes(const struct request *rq)
941 return rq->__data_len;
944 static inline int blk_rq_cur_bytes(const struct request *rq)
946 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
949 extern unsigned int blk_rq_err_bytes(const struct request *rq);
951 static inline unsigned int blk_rq_sectors(const struct request *rq)
953 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
956 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
958 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
961 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
963 return rq->stats_sectors;
966 #ifdef CONFIG_BLK_DEV_ZONED
968 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
969 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
971 static inline unsigned int bio_zone_no(struct bio *bio)
973 return blk_queue_zone_no(bdev_get_queue(bio->bi_bdev),
974 bio->bi_iter.bi_sector);
977 static inline unsigned int bio_zone_is_seq(struct bio *bio)
979 return blk_queue_zone_is_seq(bdev_get_queue(bio->bi_bdev),
980 bio->bi_iter.bi_sector);
983 static inline unsigned int blk_rq_zone_no(struct request *rq)
985 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
988 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
990 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
992 #endif /* CONFIG_BLK_DEV_ZONED */
995 * Some commands like WRITE SAME have a payload or data transfer size which
996 * is different from the size of the request. Any driver that supports such
997 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
998 * calculate the data transfer size.
1000 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1002 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1003 return rq->special_vec.bv_len;
1004 return blk_rq_bytes(rq);
1008 * Return the first full biovec in the request. The caller needs to check that
1009 * there are any bvecs before calling this helper.
1011 static inline struct bio_vec req_bvec(struct request *rq)
1013 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1014 return rq->special_vec;
1015 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1018 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1021 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1022 return min(q->limits.max_discard_sectors,
1023 UINT_MAX >> SECTOR_SHIFT);
1025 if (unlikely(op == REQ_OP_WRITE_SAME))
1026 return q->limits.max_write_same_sectors;
1028 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1029 return q->limits.max_write_zeroes_sectors;
1031 return q->limits.max_sectors;
1035 * Return maximum size of a request at given offset. Only valid for
1036 * file system requests.
1038 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1040 unsigned int chunk_sectors)
1042 if (!chunk_sectors) {
1043 if (q->limits.chunk_sectors)
1044 chunk_sectors = q->limits.chunk_sectors;
1046 return q->limits.max_sectors;
1049 if (likely(is_power_of_2(chunk_sectors)))
1050 chunk_sectors -= offset & (chunk_sectors - 1);
1052 chunk_sectors -= sector_div(offset, chunk_sectors);
1054 return min(q->limits.max_sectors, chunk_sectors);
1057 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1060 struct request_queue *q = rq->q;
1062 if (blk_rq_is_passthrough(rq))
1063 return q->limits.max_hw_sectors;
1065 if (!q->limits.chunk_sectors ||
1066 req_op(rq) == REQ_OP_DISCARD ||
1067 req_op(rq) == REQ_OP_SECURE_ERASE)
1068 return blk_queue_get_max_sectors(q, req_op(rq));
1070 return min(blk_max_size_offset(q, offset, 0),
1071 blk_queue_get_max_sectors(q, req_op(rq)));
1074 static inline unsigned int blk_rq_count_bios(struct request *rq)
1076 unsigned int nr_bios = 0;
1079 __rq_for_each_bio(bio, rq)
1085 void blk_steal_bios(struct bio_list *list, struct request *rq);
1088 * Request completion related functions.
1090 * blk_update_request() completes given number of bytes and updates
1091 * the request without completing it.
1093 extern bool blk_update_request(struct request *rq, blk_status_t error,
1094 unsigned int nr_bytes);
1096 extern void blk_abort_request(struct request *);
1099 * Access functions for manipulating queue properties
1101 extern void blk_cleanup_queue(struct request_queue *);
1102 void blk_queue_bounce_limit(struct request_queue *q, enum blk_bounce limit);
1103 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1104 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1105 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1106 extern void blk_queue_max_discard_segments(struct request_queue *,
1108 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1109 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1110 unsigned int max_discard_sectors);
1111 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1112 unsigned int max_write_same_sectors);
1113 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1114 unsigned int max_write_same_sectors);
1115 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1116 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1117 unsigned int max_zone_append_sectors);
1118 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1119 void blk_queue_zone_write_granularity(struct request_queue *q,
1121 extern void blk_queue_alignment_offset(struct request_queue *q,
1122 unsigned int alignment);
1123 void disk_update_readahead(struct gendisk *disk);
1124 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1125 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1126 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1127 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1128 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1129 extern void blk_set_default_limits(struct queue_limits *lim);
1130 extern void blk_set_stacking_limits(struct queue_limits *lim);
1131 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1133 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1135 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1136 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1137 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1138 extern void blk_queue_dma_alignment(struct request_queue *, int);
1139 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1140 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1141 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1142 extern void blk_queue_required_elevator_features(struct request_queue *q,
1143 unsigned int features);
1144 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1145 struct device *dev);
1148 * Number of physical segments as sent to the device.
1150 * Normally this is the number of discontiguous data segments sent by the
1151 * submitter. But for data-less command like discard we might have no
1152 * actual data segments submitted, but the driver might have to add it's
1153 * own special payload. In that case we still return 1 here so that this
1154 * special payload will be mapped.
1156 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1158 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1160 return rq->nr_phys_segments;
1164 * Number of discard segments (or ranges) the driver needs to fill in.
1165 * Each discard bio merged into a request is counted as one segment.
1167 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1169 return max_t(unsigned short, rq->nr_phys_segments, 1);
1172 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1173 struct scatterlist *sglist, struct scatterlist **last_sg);
1174 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1175 struct scatterlist *sglist)
1177 struct scatterlist *last_sg = NULL;
1179 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1181 extern void blk_dump_rq_flags(struct request *, char *);
1183 bool __must_check blk_get_queue(struct request_queue *);
1184 extern void blk_put_queue(struct request_queue *);
1186 void blk_mark_disk_dead(struct gendisk *disk);
1190 * blk_plug permits building a queue of related requests by holding the I/O
1191 * fragments for a short period. This allows merging of sequential requests
1192 * into single larger request. As the requests are moved from a per-task list to
1193 * the device's request_queue in a batch, this results in improved scalability
1194 * as the lock contention for request_queue lock is reduced.
1196 * It is ok not to disable preemption when adding the request to the plug list
1197 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1198 * the plug list when the task sleeps by itself. For details, please see
1199 * schedule() where blk_schedule_flush_plug() is called.
1202 struct list_head mq_list; /* blk-mq requests */
1203 struct list_head cb_list; /* md requires an unplug callback */
1204 unsigned short rq_count;
1205 bool multiple_queues;
1210 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1211 struct blk_plug_cb {
1212 struct list_head list;
1213 blk_plug_cb_fn callback;
1216 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1217 void *data, int size);
1218 extern void blk_start_plug(struct blk_plug *);
1219 extern void blk_finish_plug(struct blk_plug *);
1220 extern void blk_flush_plug_list(struct blk_plug *, bool);
1222 static inline void blk_flush_plug(struct task_struct *tsk)
1224 struct blk_plug *plug = tsk->plug;
1227 blk_flush_plug_list(plug, false);
1230 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1232 struct blk_plug *plug = tsk->plug;
1235 blk_flush_plug_list(plug, true);
1238 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1240 struct blk_plug *plug = tsk->plug;
1243 (!list_empty(&plug->mq_list) ||
1244 !list_empty(&plug->cb_list));
1247 int blkdev_issue_flush(struct block_device *bdev);
1248 long nr_blockdev_pages(void);
1249 #else /* CONFIG_BLOCK */
1253 static inline void blk_start_plug(struct blk_plug *plug)
1257 static inline void blk_finish_plug(struct blk_plug *plug)
1261 static inline void blk_flush_plug(struct task_struct *task)
1265 static inline void blk_schedule_flush_plug(struct task_struct *task)
1270 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1275 static inline int blkdev_issue_flush(struct block_device *bdev)
1280 static inline long nr_blockdev_pages(void)
1284 #endif /* CONFIG_BLOCK */
1286 extern void blk_io_schedule(void);
1288 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1289 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1291 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1293 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1294 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1295 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1296 sector_t nr_sects, gfp_t gfp_mask, int flags,
1299 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1300 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1302 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1303 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1305 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1306 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1308 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1309 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1311 return blkdev_issue_discard(sb->s_bdev,
1312 block << (sb->s_blocksize_bits -
1314 nr_blocks << (sb->s_blocksize_bits -
1318 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1319 sector_t nr_blocks, gfp_t gfp_mask)
1321 return blkdev_issue_zeroout(sb->s_bdev,
1322 block << (sb->s_blocksize_bits -
1324 nr_blocks << (sb->s_blocksize_bits -
1329 static inline bool bdev_is_partition(struct block_device *bdev)
1331 return bdev->bd_partno;
1334 enum blk_default_limits {
1335 BLK_MAX_SEGMENTS = 128,
1336 BLK_SAFE_MAX_SECTORS = 255,
1337 BLK_DEF_MAX_SECTORS = 2560,
1338 BLK_MAX_SEGMENT_SIZE = 65536,
1339 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1342 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1344 return q->limits.seg_boundary_mask;
1347 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1349 return q->limits.virt_boundary_mask;
1352 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1354 return q->limits.max_sectors;
1357 static inline unsigned int queue_max_bytes(struct request_queue *q)
1359 return min_t(unsigned int, queue_max_sectors(q), INT_MAX >> 9) << 9;
1362 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1364 return q->limits.max_hw_sectors;
1367 static inline unsigned short queue_max_segments(const struct request_queue *q)
1369 return q->limits.max_segments;
1372 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1374 return q->limits.max_discard_segments;
1377 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1379 return q->limits.max_segment_size;
1382 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1385 const struct queue_limits *l = &q->limits;
1387 return min(l->max_zone_append_sectors, l->max_sectors);
1390 static inline unsigned int
1391 bdev_max_zone_append_sectors(struct block_device *bdev)
1393 return queue_max_zone_append_sectors(bdev_get_queue(bdev));
1396 static inline unsigned int bdev_max_segments(struct block_device *bdev)
1398 return queue_max_segments(bdev_get_queue(bdev));
1401 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1405 if (q && q->limits.logical_block_size)
1406 retval = q->limits.logical_block_size;
1411 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1413 return queue_logical_block_size(bdev_get_queue(bdev));
1416 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1418 return q->limits.physical_block_size;
1421 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1423 return queue_physical_block_size(bdev_get_queue(bdev));
1426 static inline unsigned int queue_io_min(const struct request_queue *q)
1428 return q->limits.io_min;
1431 static inline int bdev_io_min(struct block_device *bdev)
1433 return queue_io_min(bdev_get_queue(bdev));
1436 static inline unsigned int queue_io_opt(const struct request_queue *q)
1438 return q->limits.io_opt;
1441 static inline int bdev_io_opt(struct block_device *bdev)
1443 return queue_io_opt(bdev_get_queue(bdev));
1446 static inline unsigned int
1447 queue_zone_write_granularity(const struct request_queue *q)
1449 return q->limits.zone_write_granularity;
1452 static inline unsigned int
1453 bdev_zone_write_granularity(struct block_device *bdev)
1455 return queue_zone_write_granularity(bdev_get_queue(bdev));
1458 static inline int queue_alignment_offset(const struct request_queue *q)
1460 if (q->limits.misaligned)
1463 return q->limits.alignment_offset;
1466 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1468 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1469 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1472 return (granularity + lim->alignment_offset - alignment) % granularity;
1475 static inline int bdev_alignment_offset(struct block_device *bdev)
1477 struct request_queue *q = bdev_get_queue(bdev);
1479 if (q->limits.misaligned)
1481 if (bdev_is_partition(bdev))
1482 return queue_limit_alignment_offset(&q->limits,
1483 bdev->bd_start_sect);
1484 return q->limits.alignment_offset;
1487 static inline int queue_discard_alignment(const struct request_queue *q)
1489 if (q->limits.discard_misaligned)
1492 return q->limits.discard_alignment;
1495 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1497 unsigned int alignment, granularity, offset;
1499 if (!lim->max_discard_sectors)
1502 /* Why are these in bytes, not sectors? */
1503 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1504 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1508 /* Offset of the partition start in 'granularity' sectors */
1509 offset = sector_div(sector, granularity);
1511 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1512 offset = (granularity + alignment - offset) % granularity;
1514 /* Turn it back into bytes, gaah */
1515 return offset << SECTOR_SHIFT;
1519 * Two cases of handling DISCARD merge:
1520 * If max_discard_segments > 1, the driver takes every bio
1521 * as a range and send them to controller together. The ranges
1522 * needn't to be contiguous.
1523 * Otherwise, the bios/requests will be handled as same as
1524 * others which should be contiguous.
1526 static inline bool blk_discard_mergable(struct request *req)
1528 if (req_op(req) == REQ_OP_DISCARD &&
1529 queue_max_discard_segments(req->q) > 1)
1534 static inline int bdev_discard_alignment(struct block_device *bdev)
1536 struct request_queue *q = bdev_get_queue(bdev);
1538 if (bdev_is_partition(bdev))
1539 return queue_limit_discard_alignment(&q->limits,
1540 bdev->bd_start_sect);
1541 return q->limits.discard_alignment;
1544 static inline unsigned int bdev_write_same(struct block_device *bdev)
1546 struct request_queue *q = bdev_get_queue(bdev);
1549 return q->limits.max_write_same_sectors;
1554 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1556 struct request_queue *q = bdev_get_queue(bdev);
1559 return q->limits.max_write_zeroes_sectors;
1564 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1566 struct request_queue *q = bdev_get_queue(bdev);
1569 return blk_queue_zoned_model(q);
1571 return BLK_ZONED_NONE;
1574 static inline bool bdev_is_zoned(struct block_device *bdev)
1576 struct request_queue *q = bdev_get_queue(bdev);
1579 return blk_queue_is_zoned(q);
1584 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1586 struct request_queue *q = bdev_get_queue(bdev);
1589 return blk_queue_zone_sectors(q);
1593 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1595 struct request_queue *q = bdev_get_queue(bdev);
1598 return queue_max_open_zones(q);
1602 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1604 struct request_queue *q = bdev_get_queue(bdev);
1607 return queue_max_active_zones(q);
1611 static inline int queue_dma_alignment(const struct request_queue *q)
1613 return q ? q->dma_alignment : 511;
1616 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1619 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1620 return !(addr & alignment) && !(len & alignment);
1623 /* assumes size > 256 */
1624 static inline unsigned int blksize_bits(unsigned int size)
1626 unsigned int bits = 8;
1630 } while (size > 256);
1634 static inline unsigned int block_size(struct block_device *bdev)
1636 return 1 << bdev->bd_inode->i_blkbits;
1639 int kblockd_schedule_work(struct work_struct *work);
1640 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1642 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1643 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1644 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1645 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1647 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1649 enum blk_integrity_flags {
1650 BLK_INTEGRITY_VERIFY = 1 << 0,
1651 BLK_INTEGRITY_GENERATE = 1 << 1,
1652 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1653 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1656 struct blk_integrity_iter {
1660 unsigned int data_size;
1661 unsigned short interval;
1662 const char *disk_name;
1665 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1666 typedef void (integrity_prepare_fn) (struct request *);
1667 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1669 struct blk_integrity_profile {
1670 integrity_processing_fn *generate_fn;
1671 integrity_processing_fn *verify_fn;
1672 integrity_prepare_fn *prepare_fn;
1673 integrity_complete_fn *complete_fn;
1677 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1678 extern void blk_integrity_unregister(struct gendisk *);
1679 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1680 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1681 struct scatterlist *);
1682 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1684 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1686 struct blk_integrity *bi = &disk->queue->integrity;
1695 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1697 return blk_get_integrity(bdev->bd_disk);
1701 blk_integrity_queue_supports_integrity(struct request_queue *q)
1703 return q->integrity.profile;
1706 static inline bool blk_integrity_rq(struct request *rq)
1708 return rq->cmd_flags & REQ_INTEGRITY;
1711 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1714 q->limits.max_integrity_segments = segs;
1717 static inline unsigned short
1718 queue_max_integrity_segments(const struct request_queue *q)
1720 return q->limits.max_integrity_segments;
1724 * bio_integrity_intervals - Return number of integrity intervals for a bio
1725 * @bi: blk_integrity profile for device
1726 * @sectors: Size of the bio in 512-byte sectors
1728 * Description: The block layer calculates everything in 512 byte
1729 * sectors but integrity metadata is done in terms of the data integrity
1730 * interval size of the storage device. Convert the block layer sectors
1731 * to the appropriate number of integrity intervals.
1733 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1734 unsigned int sectors)
1736 return sectors >> (bi->interval_exp - 9);
1739 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1740 unsigned int sectors)
1742 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1746 * Return the first bvec that contains integrity data. Only drivers that are
1747 * limited to a single integrity segment should use this helper.
1749 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1751 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1753 return rq->bio->bi_integrity->bip_vec;
1756 #else /* CONFIG_BLK_DEV_INTEGRITY */
1759 struct block_device;
1761 struct blk_integrity;
1763 static inline int blk_integrity_rq(struct request *rq)
1767 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1772 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1774 struct scatterlist *s)
1778 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1782 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1787 blk_integrity_queue_supports_integrity(struct request_queue *q)
1791 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1795 static inline void blk_integrity_register(struct gendisk *d,
1796 struct blk_integrity *b)
1799 static inline void blk_integrity_unregister(struct gendisk *d)
1802 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1806 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1811 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1812 unsigned int sectors)
1817 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1818 unsigned int sectors)
1823 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1828 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1830 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1832 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1834 void blk_ksm_unregister(struct request_queue *q);
1836 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1838 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1839 struct request_queue *q)
1844 static inline void blk_ksm_unregister(struct request_queue *q) { }
1846 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1849 struct block_device_operations {
1850 blk_qc_t (*submit_bio) (struct bio *bio);
1851 int (*open) (struct block_device *, fmode_t);
1852 void (*release) (struct gendisk *, fmode_t);
1853 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1854 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1855 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1856 unsigned int (*check_events) (struct gendisk *disk,
1857 unsigned int clearing);
1858 void (*unlock_native_capacity) (struct gendisk *);
1859 int (*getgeo)(struct block_device *, struct hd_geometry *);
1860 int (*set_read_only)(struct block_device *bdev, bool ro);
1861 /* this callback is with swap_lock and sometimes page table lock held */
1862 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1863 int (*report_zones)(struct gendisk *, sector_t sector,
1864 unsigned int nr_zones, report_zones_cb cb, void *data);
1865 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1866 struct module *owner;
1867 const struct pr_ops *pr_ops;
1870 * Special callback for probing GPT entry at a given sector.
1871 * Needed by Android devices, used by GPT scanner and MMC blk
1874 int (*alternative_gpt_sector)(struct gendisk *disk, sector_t *sector);
1877 #ifdef CONFIG_COMPAT
1878 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1879 unsigned int, unsigned long);
1881 #define blkdev_compat_ptr_ioctl NULL
1884 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1885 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1886 struct writeback_control *);
1888 #ifdef CONFIG_BLK_DEV_ZONED
1889 bool blk_req_needs_zone_write_lock(struct request *rq);
1890 bool blk_req_zone_write_trylock(struct request *rq);
1891 void __blk_req_zone_write_lock(struct request *rq);
1892 void __blk_req_zone_write_unlock(struct request *rq);
1894 static inline void blk_req_zone_write_lock(struct request *rq)
1896 if (blk_req_needs_zone_write_lock(rq))
1897 __blk_req_zone_write_lock(rq);
1900 static inline void blk_req_zone_write_unlock(struct request *rq)
1902 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1903 __blk_req_zone_write_unlock(rq);
1906 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1908 return rq->q->seq_zones_wlock &&
1909 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1912 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1914 if (!blk_req_needs_zone_write_lock(rq))
1916 return !blk_req_zone_is_write_locked(rq);
1919 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1924 static inline void blk_req_zone_write_lock(struct request *rq)
1928 static inline void blk_req_zone_write_unlock(struct request *rq)
1931 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1936 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1940 #endif /* CONFIG_BLK_DEV_ZONED */
1942 static inline void blk_wake_io_task(struct task_struct *waiter)
1945 * If we're polling, the task itself is doing the completions. For
1946 * that case, we don't need to signal a wakeup, it's enough to just
1947 * mark us as RUNNING.
1949 if (waiter == current)
1950 __set_current_state(TASK_RUNNING);
1952 wake_up_process(waiter);
1955 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1957 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1958 unsigned long start_time);
1960 void bio_start_io_acct_time(struct bio *bio, unsigned long start_time);
1961 unsigned long bio_start_io_acct(struct bio *bio);
1962 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1963 struct block_device *orig_bdev);
1966 * bio_end_io_acct - end I/O accounting for bio based drivers
1967 * @bio: bio to end account for
1968 * @start: start time returned by bio_start_io_acct()
1970 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1972 return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1975 int bdev_read_only(struct block_device *bdev);
1976 int set_blocksize(struct block_device *bdev, int size);
1978 const char *bdevname(struct block_device *bdev, char *buffer);
1979 int lookup_bdev(const char *pathname, dev_t *dev);
1981 void blkdev_show(struct seq_file *seqf, off_t offset);
1983 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1984 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1986 #define BLKDEV_MAJOR_MAX 512
1988 #define BLKDEV_MAJOR_MAX 0
1991 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1993 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1994 int bd_prepare_to_claim(struct block_device *bdev, void *holder);
1995 void bd_abort_claiming(struct block_device *bdev, void *holder);
1996 void blkdev_put(struct block_device *bdev, fmode_t mode);
1998 /* just for blk-cgroup, don't use elsewhere */
1999 struct block_device *blkdev_get_no_open(dev_t dev);
2000 void blkdev_put_no_open(struct block_device *bdev);
2002 struct block_device *bdev_alloc(struct gendisk *disk, u8 partno);
2003 void bdev_add(struct block_device *bdev, dev_t dev);
2004 struct block_device *I_BDEV(struct inode *inode);
2005 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
2009 void invalidate_bdev(struct block_device *bdev);
2010 int sync_blockdev(struct block_device *bdev);
2011 int sync_blockdev_nowait(struct block_device *bdev);
2012 void sync_bdevs(bool wait);
2014 static inline void invalidate_bdev(struct block_device *bdev)
2017 static inline int sync_blockdev(struct block_device *bdev)
2021 static inline int sync_blockdev_nowait(struct block_device *bdev)
2025 static inline void sync_bdevs(bool wait)
2029 int fsync_bdev(struct block_device *bdev);
2031 int freeze_bdev(struct block_device *bdev);
2032 int thaw_bdev(struct block_device *bdev);
2034 #endif /* _LINUX_BLKDEV_H */