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/bsg.h>
21 #include <linux/smp.h>
22 #include <linux/rcupdate.h>
23 #include <linux/percpu-refcount.h>
24 #include <linux/scatterlist.h>
25 #include <linux/blkzoned.h>
27 #include <linux/sbitmap.h>
30 struct scsi_ioctl_command;
33 struct elevator_queue;
39 struct blk_flush_queue;
42 struct blk_queue_stats;
43 struct blk_stat_callback;
44 struct blk_keyslot_manager;
46 #define BLKDEV_MIN_RQ 4
47 #define BLKDEV_MAX_RQ 128 /* Default maximum */
49 /* Must be consistent with blk_mq_poll_stats_bkt() */
50 #define BLK_MQ_POLL_STATS_BKTS 16
52 /* Doing classic polling */
53 #define BLK_MQ_POLL_CLASSIC -1
56 * Maximum number of blkcg policies allowed to be registered concurrently.
57 * Defined here to simplify include dependency.
59 #define BLKCG_MAX_POLS 6
61 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
65 typedef __u32 __bitwise req_flags_t;
67 /* drive already may have started this one */
68 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
69 /* may not be passed by ioscheduler */
70 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
71 /* request for flush sequence */
72 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
73 /* merge of different types, fail separately */
74 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
75 /* track inflight for MQ */
76 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
77 /* don't call prep for this one */
78 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
79 /* vaguely specified driver internal error. Ignored by the block layer */
80 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
81 /* don't warn about errors */
82 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
83 /* elevator private data attached */
84 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
85 /* account into disk and partition IO statistics */
86 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
87 /* runtime pm request */
88 #define RQF_PM ((__force req_flags_t)(1 << 15))
89 /* on IO scheduler merge hash */
90 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
91 /* track IO completion time */
92 #define RQF_STATS ((__force req_flags_t)(1 << 17))
93 /* Look at ->special_vec for the actual data payload instead of the
95 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
96 /* The per-zone write lock is held for this request */
97 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
98 /* already slept for hybrid poll */
99 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
100 /* ->timeout has been called, don't expire again */
101 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
103 /* flags that prevent us from merging requests: */
104 #define RQF_NOMERGE_FLAGS \
105 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
108 * Request state for blk-mq.
117 * Try to put the fields that are referenced together in the same cacheline.
119 * If you modify this structure, make sure to update blk_rq_init() and
120 * especially blk_mq_rq_ctx_init() to take care of the added fields.
123 struct request_queue *q;
124 struct blk_mq_ctx *mq_ctx;
125 struct blk_mq_hw_ctx *mq_hctx;
127 unsigned int cmd_flags; /* op and common flags */
128 req_flags_t rq_flags;
133 /* the following two fields are internal, NEVER access directly */
134 unsigned int __data_len; /* total data len */
135 sector_t __sector; /* sector cursor */
140 struct list_head queuelist;
143 * The hash is used inside the scheduler, and killed once the
144 * request reaches the dispatch list. The ipi_list is only used
145 * to queue the request for softirq completion, which is long
146 * after the request has been unhashed (and even removed from
147 * the dispatch list).
150 struct hlist_node hash; /* merge hash */
151 struct llist_node ipi_list;
155 * The rb_node is only used inside the io scheduler, requests
156 * are pruned when moved to the dispatch queue. So let the
157 * completion_data share space with the rb_node.
160 struct rb_node rb_node; /* sort/lookup */
161 struct bio_vec special_vec;
162 void *completion_data;
163 int error_count; /* for legacy drivers, don't use */
167 * Three pointers are available for the IO schedulers, if they need
168 * more they have to dynamically allocate it. Flush requests are
169 * never put on the IO scheduler. So let the flush fields share
170 * space with the elevator data.
180 struct list_head list;
181 rq_end_io_fn *saved_end_io;
185 struct gendisk *rq_disk;
186 struct block_device *part;
187 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
188 /* Time that the first bio started allocating this request. */
191 /* Time that this request was allocated for this IO. */
193 /* Time that I/O was submitted to the device. */
194 u64 io_start_time_ns;
196 #ifdef CONFIG_BLK_WBT
197 unsigned short wbt_flags;
200 * rq sectors used for blk stats. It has the same value
201 * with blk_rq_sectors(rq), except that it never be zeroed
204 unsigned short stats_sectors;
207 * Number of scatter-gather DMA addr+len pairs after
208 * physical address coalescing is performed.
210 unsigned short nr_phys_segments;
212 #if defined(CONFIG_BLK_DEV_INTEGRITY)
213 unsigned short nr_integrity_segments;
216 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
217 struct bio_crypt_ctx *crypt_ctx;
218 struct blk_ksm_keyslot *crypt_keyslot;
221 unsigned short write_hint;
222 unsigned short ioprio;
224 enum mq_rq_state state;
227 unsigned int timeout;
228 unsigned long deadline;
231 struct __call_single_data csd;
236 * completion callback.
238 rq_end_io_fn *end_io;
242 static inline bool blk_op_is_passthrough(unsigned int op)
245 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
248 static inline bool blk_rq_is_passthrough(struct request *rq)
250 return blk_op_is_passthrough(req_op(rq));
253 static inline unsigned short req_get_ioprio(struct request *req)
258 #include <linux/elevator.h>
260 struct blk_queue_ctx;
264 enum blk_eh_timer_return {
265 BLK_EH_DONE, /* drivers has completed the command */
266 BLK_EH_RESET_TIMER, /* reset timer and try again */
269 enum blk_queue_state {
274 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
275 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
277 #define BLK_SCSI_MAX_CMDS (256)
278 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
281 * Zoned block device models (zoned limit).
283 * Note: This needs to be ordered from the least to the most severe
284 * restrictions for the inheritance in blk_stack_limits() to work.
286 enum blk_zoned_model {
287 BLK_ZONED_NONE = 0, /* Regular block device */
288 BLK_ZONED_HA, /* Host-aware zoned block device */
289 BLK_ZONED_HM, /* Host-managed zoned block device */
293 * BLK_BOUNCE_NONE: never bounce (default)
294 * BLK_BOUNCE_HIGH: bounce all highmem pages
301 struct queue_limits {
302 enum blk_bounce bounce;
303 unsigned long seg_boundary_mask;
304 unsigned long virt_boundary_mask;
306 unsigned int max_hw_sectors;
307 unsigned int max_dev_sectors;
308 unsigned int chunk_sectors;
309 unsigned int max_sectors;
310 unsigned int max_segment_size;
311 unsigned int physical_block_size;
312 unsigned int logical_block_size;
313 unsigned int alignment_offset;
316 unsigned int max_discard_sectors;
317 unsigned int max_hw_discard_sectors;
318 unsigned int max_write_same_sectors;
319 unsigned int max_write_zeroes_sectors;
320 unsigned int max_zone_append_sectors;
321 unsigned int discard_granularity;
322 unsigned int discard_alignment;
323 unsigned int zone_write_granularity;
325 unsigned short max_segments;
326 unsigned short max_integrity_segments;
327 unsigned short max_discard_segments;
329 unsigned char misaligned;
330 unsigned char discard_misaligned;
331 unsigned char raid_partial_stripes_expensive;
332 enum blk_zoned_model zoned;
335 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
338 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
340 #ifdef CONFIG_BLK_DEV_ZONED
342 #define BLK_ALL_ZONES ((unsigned int)-1)
343 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
344 unsigned int nr_zones, report_zones_cb cb, void *data);
345 unsigned int blkdev_nr_zones(struct gendisk *disk);
346 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
347 sector_t sectors, sector_t nr_sectors,
349 int blk_revalidate_disk_zones(struct gendisk *disk,
350 void (*update_driver_data)(struct gendisk *disk));
352 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
353 unsigned int cmd, unsigned long arg);
354 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
355 unsigned int cmd, unsigned long arg);
357 #else /* CONFIG_BLK_DEV_ZONED */
359 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
364 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
365 fmode_t mode, unsigned int cmd,
371 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
372 fmode_t mode, unsigned int cmd,
378 #endif /* CONFIG_BLK_DEV_ZONED */
380 struct request_queue {
381 struct request *last_merge;
382 struct elevator_queue *elevator;
384 struct percpu_ref q_usage_counter;
386 struct blk_queue_stats *stats;
387 struct rq_qos *rq_qos;
389 const struct blk_mq_ops *mq_ops;
392 struct blk_mq_ctx __percpu *queue_ctx;
394 unsigned int queue_depth;
396 /* hw dispatch queues */
397 struct blk_mq_hw_ctx **queue_hw_ctx;
398 unsigned int nr_hw_queues;
401 * The queue owner gets to use this for whatever they like.
402 * ll_rw_blk doesn't touch it.
407 * various queue flags, see QUEUE_* below
409 unsigned long queue_flags;
411 * Number of contexts that have called blk_set_pm_only(). If this
412 * counter is above zero then only RQF_PM requests are processed.
417 * ida allocated id for this queue. Used to index queues from
422 spinlock_t queue_lock;
424 struct gendisk *disk;
434 struct kobject *mq_kobj;
436 #ifdef CONFIG_BLK_DEV_INTEGRITY
437 struct blk_integrity integrity;
438 #endif /* CONFIG_BLK_DEV_INTEGRITY */
442 enum rpm_status rpm_status;
448 unsigned long nr_requests; /* Max # of requests */
450 unsigned int dma_pad_mask;
451 unsigned int dma_alignment;
453 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
454 /* Inline crypto capabilities */
455 struct blk_keyslot_manager *ksm;
458 unsigned int rq_timeout;
461 struct blk_stat_callback *poll_cb;
462 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
464 struct timer_list timeout;
465 struct work_struct timeout_work;
467 atomic_t nr_active_requests_shared_sbitmap;
469 struct sbitmap_queue sched_bitmap_tags;
470 struct sbitmap_queue sched_breserved_tags;
472 struct list_head icq_list;
473 #ifdef CONFIG_BLK_CGROUP
474 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
475 struct blkcg_gq *root_blkg;
476 struct list_head blkg_list;
479 struct queue_limits limits;
481 unsigned int required_elevator_features;
483 #ifdef CONFIG_BLK_DEV_ZONED
485 * Zoned block device information for request dispatch control.
486 * nr_zones is the total number of zones of the device. This is always
487 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
488 * bits which indicates if a zone is conventional (bit set) or
489 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
490 * bits which indicates if a zone is write locked, that is, if a write
491 * request targeting the zone was dispatched. All three fields are
492 * initialized by the low level device driver (e.g. scsi/sd.c).
493 * Stacking drivers (device mappers) may or may not initialize
496 * Reads of this information must be protected with blk_queue_enter() /
497 * blk_queue_exit(). Modifying this information is only allowed while
498 * no requests are being processed. See also blk_mq_freeze_queue() and
499 * blk_mq_unfreeze_queue().
501 unsigned int nr_zones;
502 unsigned long *conv_zones_bitmap;
503 unsigned long *seq_zones_wlock;
504 unsigned int max_open_zones;
505 unsigned int max_active_zones;
506 #endif /* CONFIG_BLK_DEV_ZONED */
511 unsigned int sg_timeout;
512 unsigned int sg_reserved_size;
514 struct mutex debugfs_mutex;
515 #ifdef CONFIG_BLK_DEV_IO_TRACE
516 struct blk_trace __rcu *blk_trace;
519 * for flush operations
521 struct blk_flush_queue *fq;
523 struct list_head requeue_list;
524 spinlock_t requeue_lock;
525 struct delayed_work requeue_work;
527 struct mutex sysfs_lock;
528 struct mutex sysfs_dir_lock;
531 * for reusing dead hctx instance in case of updating
534 struct list_head unused_hctx_list;
535 spinlock_t unused_hctx_lock;
539 #if defined(CONFIG_BLK_DEV_BSG)
540 struct bsg_class_device bsg_dev;
543 #ifdef CONFIG_BLK_DEV_THROTTLING
545 struct throtl_data *td;
547 struct rcu_head rcu_head;
548 wait_queue_head_t mq_freeze_wq;
550 * Protect concurrent access to q_usage_counter by
551 * percpu_ref_kill() and percpu_ref_reinit().
553 struct mutex mq_freeze_lock;
555 struct blk_mq_tag_set *tag_set;
556 struct list_head tag_set_list;
557 struct bio_set bio_split;
559 struct dentry *debugfs_dir;
561 #ifdef CONFIG_BLK_DEBUG_FS
562 struct dentry *sched_debugfs_dir;
563 struct dentry *rqos_debugfs_dir;
566 bool mq_sysfs_init_done;
570 #define BLK_MAX_WRITE_HINTS 5
571 u64 write_hints[BLK_MAX_WRITE_HINTS];
574 /* Keep blk_queue_flag_name[] in sync with the definitions below */
575 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
576 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
577 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
578 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
579 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
580 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
581 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
582 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
583 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
584 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
585 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
586 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
587 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
588 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
589 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
590 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
591 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
592 #define QUEUE_FLAG_WC 17 /* Write back caching */
593 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
594 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
595 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
596 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
597 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
598 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
599 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
600 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
601 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
602 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
603 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
604 #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */
606 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
607 (1 << QUEUE_FLAG_SAME_COMP) | \
608 (1 << QUEUE_FLAG_NOWAIT))
610 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
611 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
612 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
614 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
615 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
616 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
617 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
618 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
619 #define blk_queue_noxmerges(q) \
620 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
621 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
622 #define blk_queue_stable_writes(q) \
623 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
624 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
625 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
626 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
627 #define blk_queue_zone_resetall(q) \
628 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
629 #define blk_queue_secure_erase(q) \
630 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
631 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
632 #define blk_queue_scsi_passthrough(q) \
633 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
634 #define blk_queue_pci_p2pdma(q) \
635 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
636 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
637 #define blk_queue_rq_alloc_time(q) \
638 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
640 #define blk_queue_rq_alloc_time(q) false
643 #define blk_noretry_request(rq) \
644 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
645 REQ_FAILFAST_DRIVER))
646 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
647 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
648 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
649 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
650 #define blk_queue_nowait(q) test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
652 extern void blk_set_pm_only(struct request_queue *q);
653 extern void blk_clear_pm_only(struct request_queue *q);
655 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
657 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
659 #define rq_dma_dir(rq) \
660 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
662 #define dma_map_bvec(dev, bv, dir, attrs) \
663 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
666 static inline bool queue_is_mq(struct request_queue *q)
672 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
674 return q->rpm_status;
677 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
683 static inline enum blk_zoned_model
684 blk_queue_zoned_model(struct request_queue *q)
686 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
687 return q->limits.zoned;
688 return BLK_ZONED_NONE;
691 static inline bool blk_queue_is_zoned(struct request_queue *q)
693 switch (blk_queue_zoned_model(q)) {
702 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
704 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
707 #ifdef CONFIG_BLK_DEV_ZONED
708 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
710 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
713 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
716 if (!blk_queue_is_zoned(q))
718 return sector >> ilog2(q->limits.chunk_sectors);
721 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
724 if (!blk_queue_is_zoned(q))
726 if (!q->conv_zones_bitmap)
728 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
731 static inline void blk_queue_max_open_zones(struct request_queue *q,
732 unsigned int max_open_zones)
734 q->max_open_zones = max_open_zones;
737 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
739 return q->max_open_zones;
742 static inline void blk_queue_max_active_zones(struct request_queue *q,
743 unsigned int max_active_zones)
745 q->max_active_zones = max_active_zones;
748 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
750 return q->max_active_zones;
752 #else /* CONFIG_BLK_DEV_ZONED */
753 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
757 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
762 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
767 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
771 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
775 #endif /* CONFIG_BLK_DEV_ZONED */
777 static inline bool rq_is_sync(struct request *rq)
779 return op_is_sync(rq->cmd_flags);
782 static inline bool rq_mergeable(struct request *rq)
784 if (blk_rq_is_passthrough(rq))
787 if (req_op(rq) == REQ_OP_FLUSH)
790 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
793 if (req_op(rq) == REQ_OP_ZONE_APPEND)
796 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
798 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
804 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
806 if (bio_page(a) == bio_page(b) &&
807 bio_offset(a) == bio_offset(b))
813 static inline unsigned int blk_queue_depth(struct request_queue *q)
816 return q->queue_depth;
818 return q->nr_requests;
822 * default timeout for SG_IO if none specified
824 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
825 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
831 unsigned long offset;
836 struct req_iterator {
837 struct bvec_iter iter;
841 /* This should not be used directly - use rq_for_each_segment */
842 #define for_each_bio(_bio) \
843 for (; _bio; _bio = _bio->bi_next)
844 #define __rq_for_each_bio(_bio, rq) \
846 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
848 #define rq_for_each_segment(bvl, _rq, _iter) \
849 __rq_for_each_bio(_iter.bio, _rq) \
850 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
852 #define rq_for_each_bvec(bvl, _rq, _iter) \
853 __rq_for_each_bio(_iter.bio, _rq) \
854 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
856 #define rq_iter_last(bvec, _iter) \
857 (_iter.bio->bi_next == NULL && \
858 bio_iter_last(bvec, _iter.iter))
860 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
861 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
863 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
864 extern void rq_flush_dcache_pages(struct request *rq);
866 static inline void rq_flush_dcache_pages(struct request *rq)
871 extern int blk_register_queue(struct gendisk *disk);
872 extern void blk_unregister_queue(struct gendisk *disk);
873 blk_qc_t submit_bio_noacct(struct bio *bio);
874 extern void blk_rq_init(struct request_queue *q, struct request *rq);
875 extern void blk_put_request(struct request *);
876 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
877 blk_mq_req_flags_t flags);
878 extern int blk_lld_busy(struct request_queue *q);
879 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
880 struct bio_set *bs, gfp_t gfp_mask,
881 int (*bio_ctr)(struct bio *, struct bio *, void *),
883 extern void blk_rq_unprep_clone(struct request *rq);
884 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
886 int blk_rq_append_bio(struct request *rq, struct bio *bio);
887 extern void blk_queue_split(struct bio **);
888 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
889 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
890 unsigned int, void __user *);
891 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
892 unsigned int, void __user *);
893 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
894 struct scsi_ioctl_command __user *);
895 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
896 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
898 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
899 extern void blk_queue_exit(struct request_queue *q);
900 extern void blk_sync_queue(struct request_queue *q);
901 extern int blk_rq_map_user(struct request_queue *, struct request *,
902 struct rq_map_data *, void __user *, unsigned long,
904 extern int blk_rq_unmap_user(struct bio *);
905 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
906 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
907 struct rq_map_data *, const struct iov_iter *,
909 extern void blk_execute_rq_nowait(struct gendisk *,
910 struct request *, int, rq_end_io_fn *);
912 blk_status_t blk_execute_rq(struct gendisk *bd_disk, struct request *rq,
915 /* Helper to convert REQ_OP_XXX to its string format XXX */
916 extern const char *blk_op_str(unsigned int op);
918 int blk_status_to_errno(blk_status_t status);
919 blk_status_t errno_to_blk_status(int errno);
921 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
923 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
925 return bdev->bd_disk->queue; /* this is never NULL */
929 * The basic unit of block I/O is a sector. It is used in a number of contexts
930 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
931 * bytes. Variables of type sector_t represent an offset or size that is a
932 * multiple of 512 bytes. Hence these two constants.
935 #define SECTOR_SHIFT 9
938 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
941 #define PAGE_SECTORS_SHIFT (PAGE_SHIFT - SECTOR_SHIFT)
942 #define PAGE_SECTORS (1 << PAGE_SECTORS_SHIFT)
943 #define SECTOR_MASK (PAGE_SECTORS - 1)
946 * blk_rq_pos() : the current sector
947 * blk_rq_bytes() : bytes left in the entire request
948 * blk_rq_cur_bytes() : bytes left in the current segment
949 * blk_rq_err_bytes() : bytes left till the next error boundary
950 * blk_rq_sectors() : sectors left in the entire request
951 * blk_rq_cur_sectors() : sectors left in the current segment
952 * blk_rq_stats_sectors() : sectors of the entire request used for stats
954 static inline sector_t blk_rq_pos(const struct request *rq)
959 static inline unsigned int blk_rq_bytes(const struct request *rq)
961 return rq->__data_len;
964 static inline int blk_rq_cur_bytes(const struct request *rq)
966 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
969 extern unsigned int blk_rq_err_bytes(const struct request *rq);
971 static inline unsigned int blk_rq_sectors(const struct request *rq)
973 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
976 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
978 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
981 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
983 return rq->stats_sectors;
986 #ifdef CONFIG_BLK_DEV_ZONED
988 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
989 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
991 static inline unsigned int bio_zone_no(struct bio *bio)
993 return blk_queue_zone_no(bdev_get_queue(bio->bi_bdev),
994 bio->bi_iter.bi_sector);
997 static inline unsigned int bio_zone_is_seq(struct bio *bio)
999 return blk_queue_zone_is_seq(bdev_get_queue(bio->bi_bdev),
1000 bio->bi_iter.bi_sector);
1003 static inline unsigned int blk_rq_zone_no(struct request *rq)
1005 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1008 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1010 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1012 #endif /* CONFIG_BLK_DEV_ZONED */
1015 * Some commands like WRITE SAME have a payload or data transfer size which
1016 * is different from the size of the request. Any driver that supports such
1017 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1018 * calculate the data transfer size.
1020 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1022 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1023 return rq->special_vec.bv_len;
1024 return blk_rq_bytes(rq);
1028 * Return the first full biovec in the request. The caller needs to check that
1029 * there are any bvecs before calling this helper.
1031 static inline struct bio_vec req_bvec(struct request *rq)
1033 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1034 return rq->special_vec;
1035 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1038 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1041 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1042 return min(q->limits.max_discard_sectors,
1043 UINT_MAX >> SECTOR_SHIFT);
1045 if (unlikely(op == REQ_OP_WRITE_SAME))
1046 return q->limits.max_write_same_sectors;
1048 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1049 return q->limits.max_write_zeroes_sectors;
1051 return q->limits.max_sectors;
1055 * Return maximum size of a request at given offset. Only valid for
1056 * file system requests.
1058 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1060 unsigned int chunk_sectors)
1062 if (!chunk_sectors) {
1063 if (q->limits.chunk_sectors)
1064 chunk_sectors = q->limits.chunk_sectors;
1066 return q->limits.max_sectors;
1069 if (likely(is_power_of_2(chunk_sectors)))
1070 chunk_sectors -= offset & (chunk_sectors - 1);
1072 chunk_sectors -= sector_div(offset, chunk_sectors);
1074 return min(q->limits.max_sectors, chunk_sectors);
1077 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1080 struct request_queue *q = rq->q;
1082 if (blk_rq_is_passthrough(rq))
1083 return q->limits.max_hw_sectors;
1085 if (!q->limits.chunk_sectors ||
1086 req_op(rq) == REQ_OP_DISCARD ||
1087 req_op(rq) == REQ_OP_SECURE_ERASE)
1088 return blk_queue_get_max_sectors(q, req_op(rq));
1090 return min(blk_max_size_offset(q, offset, 0),
1091 blk_queue_get_max_sectors(q, req_op(rq)));
1094 static inline unsigned int blk_rq_count_bios(struct request *rq)
1096 unsigned int nr_bios = 0;
1099 __rq_for_each_bio(bio, rq)
1105 void blk_steal_bios(struct bio_list *list, struct request *rq);
1108 * Request completion related functions.
1110 * blk_update_request() completes given number of bytes and updates
1111 * the request without completing it.
1113 extern bool blk_update_request(struct request *rq, blk_status_t error,
1114 unsigned int nr_bytes);
1116 extern void blk_abort_request(struct request *);
1119 * Access functions for manipulating queue properties
1121 extern void blk_cleanup_queue(struct request_queue *);
1122 void blk_queue_bounce_limit(struct request_queue *q, enum blk_bounce limit);
1123 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1124 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1125 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1126 extern void blk_queue_max_discard_segments(struct request_queue *,
1128 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1129 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1130 unsigned int max_discard_sectors);
1131 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1132 unsigned int max_write_same_sectors);
1133 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1134 unsigned int max_write_same_sectors);
1135 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1136 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1137 unsigned int max_zone_append_sectors);
1138 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1139 void blk_queue_zone_write_granularity(struct request_queue *q,
1141 extern void blk_queue_alignment_offset(struct request_queue *q,
1142 unsigned int alignment);
1143 void disk_update_readahead(struct gendisk *disk);
1144 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1145 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1146 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1147 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1148 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1149 extern void blk_set_default_limits(struct queue_limits *lim);
1150 extern void blk_set_stacking_limits(struct queue_limits *lim);
1151 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1153 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1155 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1156 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1157 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1158 extern void blk_queue_dma_alignment(struct request_queue *, int);
1159 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1160 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1161 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1162 extern void blk_queue_required_elevator_features(struct request_queue *q,
1163 unsigned int features);
1164 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1165 struct device *dev);
1168 * Number of physical segments as sent to the device.
1170 * Normally this is the number of discontiguous data segments sent by the
1171 * submitter. But for data-less command like discard we might have no
1172 * actual data segments submitted, but the driver might have to add it's
1173 * own special payload. In that case we still return 1 here so that this
1174 * special payload will be mapped.
1176 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1178 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1180 return rq->nr_phys_segments;
1184 * Number of discard segments (or ranges) the driver needs to fill in.
1185 * Each discard bio merged into a request is counted as one segment.
1187 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1189 return max_t(unsigned short, rq->nr_phys_segments, 1);
1192 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1193 struct scatterlist *sglist, struct scatterlist **last_sg);
1194 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1195 struct scatterlist *sglist)
1197 struct scatterlist *last_sg = NULL;
1199 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1201 extern void blk_dump_rq_flags(struct request *, char *);
1203 bool __must_check blk_get_queue(struct request_queue *);
1204 extern void blk_put_queue(struct request_queue *);
1205 extern void blk_set_queue_dying(struct request_queue *);
1209 * blk_plug permits building a queue of related requests by holding the I/O
1210 * fragments for a short period. This allows merging of sequential requests
1211 * into single larger request. As the requests are moved from a per-task list to
1212 * the device's request_queue in a batch, this results in improved scalability
1213 * as the lock contention for request_queue lock is reduced.
1215 * It is ok not to disable preemption when adding the request to the plug list
1216 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1217 * the plug list when the task sleeps by itself. For details, please see
1218 * schedule() where blk_schedule_flush_plug() is called.
1221 struct list_head mq_list; /* blk-mq requests */
1222 struct list_head cb_list; /* md requires an unplug callback */
1223 unsigned short rq_count;
1224 bool multiple_queues;
1227 #define BLK_MAX_REQUEST_COUNT 16
1228 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1231 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1232 struct blk_plug_cb {
1233 struct list_head list;
1234 blk_plug_cb_fn callback;
1237 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1238 void *data, int size);
1239 extern void blk_start_plug(struct blk_plug *);
1240 extern void blk_finish_plug(struct blk_plug *);
1241 extern void blk_flush_plug_list(struct blk_plug *, bool);
1243 static inline void blk_flush_plug(struct task_struct *tsk)
1245 struct blk_plug *plug = tsk->plug;
1248 blk_flush_plug_list(plug, false);
1251 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1253 struct blk_plug *plug = tsk->plug;
1256 blk_flush_plug_list(plug, true);
1259 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1261 struct blk_plug *plug = tsk->plug;
1264 (!list_empty(&plug->mq_list) ||
1265 !list_empty(&plug->cb_list));
1268 int blkdev_issue_flush(struct block_device *bdev);
1269 long nr_blockdev_pages(void);
1270 #else /* CONFIG_BLOCK */
1274 static inline void blk_start_plug(struct blk_plug *plug)
1278 static inline void blk_finish_plug(struct blk_plug *plug)
1282 static inline void blk_flush_plug(struct task_struct *task)
1286 static inline void blk_schedule_flush_plug(struct task_struct *task)
1291 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1296 static inline int blkdev_issue_flush(struct block_device *bdev)
1301 static inline long nr_blockdev_pages(void)
1305 #endif /* CONFIG_BLOCK */
1307 extern void blk_io_schedule(void);
1309 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1310 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1312 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1314 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1315 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1316 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1317 sector_t nr_sects, gfp_t gfp_mask, int flags,
1320 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1321 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1323 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1324 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1326 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1327 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1329 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1330 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1332 return blkdev_issue_discard(sb->s_bdev,
1333 block << (sb->s_blocksize_bits -
1335 nr_blocks << (sb->s_blocksize_bits -
1339 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1340 sector_t nr_blocks, gfp_t gfp_mask)
1342 return blkdev_issue_zeroout(sb->s_bdev,
1343 block << (sb->s_blocksize_bits -
1345 nr_blocks << (sb->s_blocksize_bits -
1350 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1352 static inline bool bdev_is_partition(struct block_device *bdev)
1354 return bdev->bd_partno;
1357 enum blk_default_limits {
1358 BLK_MAX_SEGMENTS = 128,
1359 BLK_SAFE_MAX_SECTORS = 255,
1360 BLK_DEF_MAX_SECTORS = 2560,
1361 BLK_MAX_SEGMENT_SIZE = 65536,
1362 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1365 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1367 return q->limits.seg_boundary_mask;
1370 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1372 return q->limits.virt_boundary_mask;
1375 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1377 return q->limits.max_sectors;
1380 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1382 return q->limits.max_hw_sectors;
1385 static inline unsigned short queue_max_segments(const struct request_queue *q)
1387 return q->limits.max_segments;
1390 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1392 return q->limits.max_discard_segments;
1395 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1397 return q->limits.max_segment_size;
1400 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1403 const struct queue_limits *l = &q->limits;
1405 return min(l->max_zone_append_sectors, l->max_sectors);
1408 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1412 if (q && q->limits.logical_block_size)
1413 retval = q->limits.logical_block_size;
1418 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1420 return queue_logical_block_size(bdev_get_queue(bdev));
1423 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1425 return q->limits.physical_block_size;
1428 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1430 return queue_physical_block_size(bdev_get_queue(bdev));
1433 static inline unsigned int queue_io_min(const struct request_queue *q)
1435 return q->limits.io_min;
1438 static inline int bdev_io_min(struct block_device *bdev)
1440 return queue_io_min(bdev_get_queue(bdev));
1443 static inline unsigned int queue_io_opt(const struct request_queue *q)
1445 return q->limits.io_opt;
1448 static inline int bdev_io_opt(struct block_device *bdev)
1450 return queue_io_opt(bdev_get_queue(bdev));
1453 static inline unsigned int
1454 queue_zone_write_granularity(const struct request_queue *q)
1456 return q->limits.zone_write_granularity;
1459 static inline unsigned int
1460 bdev_zone_write_granularity(struct block_device *bdev)
1462 return queue_zone_write_granularity(bdev_get_queue(bdev));
1465 static inline int queue_alignment_offset(const struct request_queue *q)
1467 if (q->limits.misaligned)
1470 return q->limits.alignment_offset;
1473 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1475 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1476 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1479 return (granularity + lim->alignment_offset - alignment) % granularity;
1482 static inline int bdev_alignment_offset(struct block_device *bdev)
1484 struct request_queue *q = bdev_get_queue(bdev);
1486 if (q->limits.misaligned)
1488 if (bdev_is_partition(bdev))
1489 return queue_limit_alignment_offset(&q->limits,
1490 bdev->bd_start_sect);
1491 return q->limits.alignment_offset;
1494 static inline int queue_discard_alignment(const struct request_queue *q)
1496 if (q->limits.discard_misaligned)
1499 return q->limits.discard_alignment;
1502 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1504 unsigned int alignment, granularity, offset;
1506 if (!lim->max_discard_sectors)
1509 /* Why are these in bytes, not sectors? */
1510 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1511 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1515 /* Offset of the partition start in 'granularity' sectors */
1516 offset = sector_div(sector, granularity);
1518 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1519 offset = (granularity + alignment - offset) % granularity;
1521 /* Turn it back into bytes, gaah */
1522 return offset << SECTOR_SHIFT;
1526 * Two cases of handling DISCARD merge:
1527 * If max_discard_segments > 1, the driver takes every bio
1528 * as a range and send them to controller together. The ranges
1529 * needn't to be contiguous.
1530 * Otherwise, the bios/requests will be handled as same as
1531 * others which should be contiguous.
1533 static inline bool blk_discard_mergable(struct request *req)
1535 if (req_op(req) == REQ_OP_DISCARD &&
1536 queue_max_discard_segments(req->q) > 1)
1541 static inline int bdev_discard_alignment(struct block_device *bdev)
1543 struct request_queue *q = bdev_get_queue(bdev);
1545 if (bdev_is_partition(bdev))
1546 return queue_limit_discard_alignment(&q->limits,
1547 bdev->bd_start_sect);
1548 return q->limits.discard_alignment;
1551 static inline unsigned int bdev_write_same(struct block_device *bdev)
1553 struct request_queue *q = bdev_get_queue(bdev);
1556 return q->limits.max_write_same_sectors;
1561 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1563 struct request_queue *q = bdev_get_queue(bdev);
1566 return q->limits.max_write_zeroes_sectors;
1571 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1573 struct request_queue *q = bdev_get_queue(bdev);
1576 return blk_queue_zoned_model(q);
1578 return BLK_ZONED_NONE;
1581 static inline bool bdev_is_zoned(struct block_device *bdev)
1583 struct request_queue *q = bdev_get_queue(bdev);
1586 return blk_queue_is_zoned(q);
1591 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1593 struct request_queue *q = bdev_get_queue(bdev);
1596 return blk_queue_zone_sectors(q);
1600 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1602 struct request_queue *q = bdev_get_queue(bdev);
1605 return queue_max_open_zones(q);
1609 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1611 struct request_queue *q = bdev_get_queue(bdev);
1614 return queue_max_active_zones(q);
1618 static inline int queue_dma_alignment(const struct request_queue *q)
1620 return q ? q->dma_alignment : 511;
1623 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1626 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1627 return !(addr & alignment) && !(len & alignment);
1630 /* assumes size > 256 */
1631 static inline unsigned int blksize_bits(unsigned int size)
1633 unsigned int bits = 8;
1637 } while (size > 256);
1641 static inline unsigned int block_size(struct block_device *bdev)
1643 return 1 << bdev->bd_inode->i_blkbits;
1646 int kblockd_schedule_work(struct work_struct *work);
1647 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1649 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1650 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1651 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1652 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1654 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1656 enum blk_integrity_flags {
1657 BLK_INTEGRITY_VERIFY = 1 << 0,
1658 BLK_INTEGRITY_GENERATE = 1 << 1,
1659 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1660 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1663 struct blk_integrity_iter {
1667 unsigned int data_size;
1668 unsigned short interval;
1669 const char *disk_name;
1672 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1673 typedef void (integrity_prepare_fn) (struct request *);
1674 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1676 struct blk_integrity_profile {
1677 integrity_processing_fn *generate_fn;
1678 integrity_processing_fn *verify_fn;
1679 integrity_prepare_fn *prepare_fn;
1680 integrity_complete_fn *complete_fn;
1684 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1685 extern void blk_integrity_unregister(struct gendisk *);
1686 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1687 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1688 struct scatterlist *);
1689 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1691 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1693 struct blk_integrity *bi = &disk->queue->integrity;
1702 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1704 return blk_get_integrity(bdev->bd_disk);
1708 blk_integrity_queue_supports_integrity(struct request_queue *q)
1710 return q->integrity.profile;
1713 static inline bool blk_integrity_rq(struct request *rq)
1715 return rq->cmd_flags & REQ_INTEGRITY;
1718 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1721 q->limits.max_integrity_segments = segs;
1724 static inline unsigned short
1725 queue_max_integrity_segments(const struct request_queue *q)
1727 return q->limits.max_integrity_segments;
1731 * bio_integrity_intervals - Return number of integrity intervals for a bio
1732 * @bi: blk_integrity profile for device
1733 * @sectors: Size of the bio in 512-byte sectors
1735 * Description: The block layer calculates everything in 512 byte
1736 * sectors but integrity metadata is done in terms of the data integrity
1737 * interval size of the storage device. Convert the block layer sectors
1738 * to the appropriate number of integrity intervals.
1740 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1741 unsigned int sectors)
1743 return sectors >> (bi->interval_exp - 9);
1746 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1747 unsigned int sectors)
1749 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1753 * Return the first bvec that contains integrity data. Only drivers that are
1754 * limited to a single integrity segment should use this helper.
1756 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1758 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1760 return rq->bio->bi_integrity->bip_vec;
1763 #else /* CONFIG_BLK_DEV_INTEGRITY */
1766 struct block_device;
1768 struct blk_integrity;
1770 static inline int blk_integrity_rq(struct request *rq)
1774 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1779 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1781 struct scatterlist *s)
1785 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1789 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1794 blk_integrity_queue_supports_integrity(struct request_queue *q)
1798 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1802 static inline void blk_integrity_register(struct gendisk *d,
1803 struct blk_integrity *b)
1806 static inline void blk_integrity_unregister(struct gendisk *d)
1809 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1813 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1818 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1819 unsigned int sectors)
1824 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1825 unsigned int sectors)
1830 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1835 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1837 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1839 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1841 void blk_ksm_unregister(struct request_queue *q);
1843 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1845 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1846 struct request_queue *q)
1851 static inline void blk_ksm_unregister(struct request_queue *q) { }
1853 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1856 struct block_device_operations {
1857 blk_qc_t (*submit_bio) (struct bio *bio);
1858 int (*open) (struct block_device *, fmode_t);
1859 void (*release) (struct gendisk *, fmode_t);
1860 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1861 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1862 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1863 unsigned int (*check_events) (struct gendisk *disk,
1864 unsigned int clearing);
1865 void (*unlock_native_capacity) (struct gendisk *);
1866 int (*getgeo)(struct block_device *, struct hd_geometry *);
1867 int (*set_read_only)(struct block_device *bdev, bool ro);
1868 /* this callback is with swap_lock and sometimes page table lock held */
1869 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1870 int (*report_zones)(struct gendisk *, sector_t sector,
1871 unsigned int nr_zones, report_zones_cb cb, void *data);
1872 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1873 struct module *owner;
1874 const struct pr_ops *pr_ops;
1877 * Special callback for probing GPT entry at a given sector.
1878 * Needed by Android devices, used by GPT scanner and MMC blk
1881 int (*alternative_gpt_sector)(struct gendisk *disk, sector_t *sector);
1884 #ifdef CONFIG_COMPAT
1885 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1886 unsigned int, unsigned long);
1888 #define blkdev_compat_ptr_ioctl NULL
1891 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1892 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1893 struct writeback_control *);
1895 #ifdef CONFIG_BLK_DEV_ZONED
1896 bool blk_req_needs_zone_write_lock(struct request *rq);
1897 bool blk_req_zone_write_trylock(struct request *rq);
1898 void __blk_req_zone_write_lock(struct request *rq);
1899 void __blk_req_zone_write_unlock(struct request *rq);
1901 static inline void blk_req_zone_write_lock(struct request *rq)
1903 if (blk_req_needs_zone_write_lock(rq))
1904 __blk_req_zone_write_lock(rq);
1907 static inline void blk_req_zone_write_unlock(struct request *rq)
1909 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1910 __blk_req_zone_write_unlock(rq);
1913 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1915 return rq->q->seq_zones_wlock &&
1916 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1919 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1921 if (!blk_req_needs_zone_write_lock(rq))
1923 return !blk_req_zone_is_write_locked(rq);
1926 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1931 static inline void blk_req_zone_write_lock(struct request *rq)
1935 static inline void blk_req_zone_write_unlock(struct request *rq)
1938 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1943 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1947 #endif /* CONFIG_BLK_DEV_ZONED */
1949 static inline void blk_wake_io_task(struct task_struct *waiter)
1952 * If we're polling, the task itself is doing the completions. For
1953 * that case, we don't need to signal a wakeup, it's enough to just
1954 * mark us as RUNNING.
1956 if (waiter == current)
1957 __set_current_state(TASK_RUNNING);
1959 wake_up_process(waiter);
1962 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1964 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1965 unsigned long start_time);
1967 unsigned long bio_start_io_acct(struct bio *bio);
1968 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1969 struct block_device *orig_bdev);
1972 * bio_end_io_acct - end I/O accounting for bio based drivers
1973 * @bio: bio to end account for
1974 * @start: start time returned by bio_start_io_acct()
1976 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1978 return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1981 int bdev_read_only(struct block_device *bdev);
1982 int set_blocksize(struct block_device *bdev, int size);
1984 const char *bdevname(struct block_device *bdev, char *buffer);
1985 int lookup_bdev(const char *pathname, dev_t *dev);
1987 void blkdev_show(struct seq_file *seqf, off_t offset);
1989 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1990 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1992 #define BLKDEV_MAJOR_MAX 512
1994 #define BLKDEV_MAJOR_MAX 0
1997 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1999 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
2000 int bd_prepare_to_claim(struct block_device *bdev, void *holder);
2001 void bd_abort_claiming(struct block_device *bdev, void *holder);
2002 void blkdev_put(struct block_device *bdev, fmode_t mode);
2004 /* just for blk-cgroup, don't use elsewhere */
2005 struct block_device *blkdev_get_no_open(dev_t dev);
2006 void blkdev_put_no_open(struct block_device *bdev);
2008 struct block_device *bdev_alloc(struct gendisk *disk, u8 partno);
2009 void bdev_add(struct block_device *bdev, dev_t dev);
2010 struct block_device *I_BDEV(struct inode *inode);
2011 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
2015 void invalidate_bdev(struct block_device *bdev);
2016 int sync_blockdev(struct block_device *bdev);
2018 static inline void invalidate_bdev(struct block_device *bdev)
2021 static inline int sync_blockdev(struct block_device *bdev)
2026 int fsync_bdev(struct block_device *bdev);
2028 int freeze_bdev(struct block_device *bdev);
2029 int thaw_bdev(struct block_device *bdev);
2031 #endif /* _LINUX_BLKDEV_H */