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/backing-dev-defs.h>
15 #include <linux/wait.h>
16 #include <linux/mempool.h>
17 #include <linux/pfn.h>
18 #include <linux/bio.h>
19 #include <linux/stringify.h>
20 #include <linux/gfp.h>
21 #include <linux/bsg.h>
22 #include <linux/smp.h>
23 #include <linux/rcupdate.h>
24 #include <linux/percpu-refcount.h>
25 #include <linux/scatterlist.h>
26 #include <linux/blkzoned.h>
28 #include <linux/sbitmap.h>
31 struct scsi_ioctl_command;
34 struct elevator_queue;
40 struct blk_flush_queue;
43 struct blk_queue_stats;
44 struct blk_stat_callback;
45 struct blk_keyslot_manager;
47 #define BLKDEV_MIN_RQ 4
48 #define BLKDEV_MAX_RQ 128 /* Default maximum */
50 /* Must be consistent with blk_mq_poll_stats_bkt() */
51 #define BLK_MQ_POLL_STATS_BKTS 16
53 /* Doing classic polling */
54 #define BLK_MQ_POLL_CLASSIC -1
57 * Maximum number of blkcg policies allowed to be registered concurrently.
58 * Defined here to simplify include dependency.
60 #define BLKCG_MAX_POLS 5
62 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
66 typedef __u32 __bitwise req_flags_t;
68 /* drive already may have started this one */
69 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
70 /* may not be passed by ioscheduler */
71 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
72 /* request for flush sequence */
73 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
74 /* merge of different types, fail separately */
75 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
76 /* track inflight for MQ */
77 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
78 /* don't call prep for this one */
79 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
80 /* vaguely specified driver internal error. Ignored by the block layer */
81 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
82 /* don't warn about errors */
83 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
84 /* elevator private data attached */
85 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
86 /* account into disk and partition IO statistics */
87 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
88 /* runtime pm request */
89 #define RQF_PM ((__force req_flags_t)(1 << 15))
90 /* on IO scheduler merge hash */
91 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
92 /* track IO completion time */
93 #define RQF_STATS ((__force req_flags_t)(1 << 17))
94 /* Look at ->special_vec for the actual data payload instead of the
96 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
97 /* The per-zone write lock is held for this request */
98 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
99 /* already slept for hybrid poll */
100 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
101 /* ->timeout has been called, don't expire again */
102 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
104 /* flags that prevent us from merging requests: */
105 #define RQF_NOMERGE_FLAGS \
106 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
109 * Request state for blk-mq.
118 * Try to put the fields that are referenced together in the same cacheline.
120 * If you modify this structure, make sure to update blk_rq_init() and
121 * especially blk_mq_rq_ctx_init() to take care of the added fields.
124 struct request_queue *q;
125 struct blk_mq_ctx *mq_ctx;
126 struct blk_mq_hw_ctx *mq_hctx;
128 unsigned int cmd_flags; /* op and common flags */
129 req_flags_t rq_flags;
134 /* the following two fields are internal, NEVER access directly */
135 unsigned int __data_len; /* total data len */
136 sector_t __sector; /* sector cursor */
141 struct list_head queuelist;
144 * The hash is used inside the scheduler, and killed once the
145 * request reaches the dispatch list. The ipi_list is only used
146 * to queue the request for softirq completion, which is long
147 * after the request has been unhashed (and even removed from
148 * the dispatch list).
151 struct hlist_node hash; /* merge hash */
152 struct llist_node ipi_list;
156 * The rb_node is only used inside the io scheduler, requests
157 * are pruned when moved to the dispatch queue. So let the
158 * completion_data share space with the rb_node.
161 struct rb_node rb_node; /* sort/lookup */
162 struct bio_vec special_vec;
163 void *completion_data;
164 int error_count; /* for legacy drivers, don't use */
168 * Three pointers are available for the IO schedulers, if they need
169 * more they have to dynamically allocate it. Flush requests are
170 * never put on the IO scheduler. So let the flush fields share
171 * space with the elevator data.
181 struct list_head list;
182 rq_end_io_fn *saved_end_io;
186 struct gendisk *rq_disk;
187 struct block_device *part;
188 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
189 /* Time that the first bio started allocating this request. */
192 /* Time that this request was allocated for this IO. */
194 /* Time that I/O was submitted to the device. */
195 u64 io_start_time_ns;
197 #ifdef CONFIG_BLK_WBT
198 unsigned short wbt_flags;
201 * rq sectors used for blk stats. It has the same value
202 * with blk_rq_sectors(rq), except that it never be zeroed
205 unsigned short stats_sectors;
208 * Number of scatter-gather DMA addr+len pairs after
209 * physical address coalescing is performed.
211 unsigned short nr_phys_segments;
213 #if defined(CONFIG_BLK_DEV_INTEGRITY)
214 unsigned short nr_integrity_segments;
217 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
218 struct bio_crypt_ctx *crypt_ctx;
219 struct blk_ksm_keyslot *crypt_keyslot;
222 unsigned short write_hint;
223 unsigned short ioprio;
225 enum mq_rq_state state;
228 unsigned int timeout;
229 unsigned long deadline;
232 struct __call_single_data csd;
237 * completion callback.
239 rq_end_io_fn *end_io;
243 static inline bool blk_op_is_scsi(unsigned int op)
245 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
248 static inline bool blk_op_is_private(unsigned int op)
250 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
253 static inline bool blk_rq_is_scsi(struct request *rq)
255 return blk_op_is_scsi(req_op(rq));
258 static inline bool blk_rq_is_private(struct request *rq)
260 return blk_op_is_private(req_op(rq));
263 static inline bool blk_rq_is_passthrough(struct request *rq)
265 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
268 static inline bool bio_is_passthrough(struct bio *bio)
270 unsigned op = bio_op(bio);
272 return blk_op_is_scsi(op) || blk_op_is_private(op);
275 static inline bool blk_op_is_passthrough(unsigned int op)
277 return (blk_op_is_scsi(op & REQ_OP_MASK) ||
278 blk_op_is_private(op & REQ_OP_MASK));
281 static inline unsigned short req_get_ioprio(struct request *req)
286 #include <linux/elevator.h>
288 struct blk_queue_ctx;
292 enum blk_eh_timer_return {
293 BLK_EH_DONE, /* drivers has completed the command */
294 BLK_EH_RESET_TIMER, /* reset timer and try again */
297 enum blk_queue_state {
302 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
303 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
305 #define BLK_SCSI_MAX_CMDS (256)
306 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
309 * Zoned block device models (zoned limit).
311 * Note: This needs to be ordered from the least to the most severe
312 * restrictions for the inheritance in blk_stack_limits() to work.
314 enum blk_zoned_model {
315 BLK_ZONED_NONE = 0, /* Regular block device */
316 BLK_ZONED_HA, /* Host-aware zoned block device */
317 BLK_ZONED_HM, /* Host-managed zoned block device */
321 * BLK_BOUNCE_NONE: never bounce (default)
322 * BLK_BOUNCE_HIGH: bounce all highmem pages
329 struct queue_limits {
330 enum blk_bounce bounce;
331 unsigned long seg_boundary_mask;
332 unsigned long virt_boundary_mask;
334 unsigned int max_hw_sectors;
335 unsigned int max_dev_sectors;
336 unsigned int chunk_sectors;
337 unsigned int max_sectors;
338 unsigned int max_segment_size;
339 unsigned int physical_block_size;
340 unsigned int logical_block_size;
341 unsigned int alignment_offset;
344 unsigned int max_discard_sectors;
345 unsigned int max_hw_discard_sectors;
346 unsigned int max_write_same_sectors;
347 unsigned int max_write_zeroes_sectors;
348 unsigned int max_zone_append_sectors;
349 unsigned int discard_granularity;
350 unsigned int discard_alignment;
351 unsigned int zone_write_granularity;
353 unsigned short max_segments;
354 unsigned short max_integrity_segments;
355 unsigned short max_discard_segments;
357 unsigned char misaligned;
358 unsigned char discard_misaligned;
359 unsigned char raid_partial_stripes_expensive;
360 enum blk_zoned_model zoned;
363 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
366 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
368 #ifdef CONFIG_BLK_DEV_ZONED
370 #define BLK_ALL_ZONES ((unsigned int)-1)
371 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
372 unsigned int nr_zones, report_zones_cb cb, void *data);
373 unsigned int blkdev_nr_zones(struct gendisk *disk);
374 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
375 sector_t sectors, sector_t nr_sectors,
377 int blk_revalidate_disk_zones(struct gendisk *disk,
378 void (*update_driver_data)(struct gendisk *disk));
380 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
381 unsigned int cmd, unsigned long arg);
382 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
383 unsigned int cmd, unsigned long arg);
385 #else /* CONFIG_BLK_DEV_ZONED */
387 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
392 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
393 fmode_t mode, unsigned int cmd,
399 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
400 fmode_t mode, unsigned int cmd,
406 #endif /* CONFIG_BLK_DEV_ZONED */
408 struct request_queue {
409 struct request *last_merge;
410 struct elevator_queue *elevator;
412 struct percpu_ref q_usage_counter;
414 struct blk_queue_stats *stats;
415 struct rq_qos *rq_qos;
417 const struct blk_mq_ops *mq_ops;
420 struct blk_mq_ctx __percpu *queue_ctx;
422 unsigned int queue_depth;
424 /* hw dispatch queues */
425 struct blk_mq_hw_ctx **queue_hw_ctx;
426 unsigned int nr_hw_queues;
428 struct backing_dev_info *backing_dev_info;
431 * The queue owner gets to use this for whatever they like.
432 * ll_rw_blk doesn't touch it.
437 * various queue flags, see QUEUE_* below
439 unsigned long queue_flags;
441 * Number of contexts that have called blk_set_pm_only(). If this
442 * counter is above zero then only RQF_PM requests are processed.
447 * ida allocated id for this queue. Used to index queues from
452 spinlock_t queue_lock;
462 struct kobject *mq_kobj;
464 #ifdef CONFIG_BLK_DEV_INTEGRITY
465 struct blk_integrity integrity;
466 #endif /* CONFIG_BLK_DEV_INTEGRITY */
470 enum rpm_status rpm_status;
476 unsigned long nr_requests; /* Max # of requests */
478 unsigned int dma_pad_mask;
479 unsigned int dma_alignment;
481 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
482 /* Inline crypto capabilities */
483 struct blk_keyslot_manager *ksm;
486 unsigned int rq_timeout;
489 struct blk_stat_callback *poll_cb;
490 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
492 struct timer_list timeout;
493 struct work_struct timeout_work;
495 atomic_t nr_active_requests_shared_sbitmap;
497 struct sbitmap_queue sched_bitmap_tags;
498 struct sbitmap_queue sched_breserved_tags;
500 struct list_head icq_list;
501 #ifdef CONFIG_BLK_CGROUP
502 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
503 struct blkcg_gq *root_blkg;
504 struct list_head blkg_list;
507 struct queue_limits limits;
509 unsigned int required_elevator_features;
511 #ifdef CONFIG_BLK_DEV_ZONED
513 * Zoned block device information for request dispatch control.
514 * nr_zones is the total number of zones of the device. This is always
515 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
516 * bits which indicates if a zone is conventional (bit set) or
517 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
518 * bits which indicates if a zone is write locked, that is, if a write
519 * request targeting the zone was dispatched. All three fields are
520 * initialized by the low level device driver (e.g. scsi/sd.c).
521 * Stacking drivers (device mappers) may or may not initialize
524 * Reads of this information must be protected with blk_queue_enter() /
525 * blk_queue_exit(). Modifying this information is only allowed while
526 * no requests are being processed. See also blk_mq_freeze_queue() and
527 * blk_mq_unfreeze_queue().
529 unsigned int nr_zones;
530 unsigned long *conv_zones_bitmap;
531 unsigned long *seq_zones_wlock;
532 unsigned int max_open_zones;
533 unsigned int max_active_zones;
534 #endif /* CONFIG_BLK_DEV_ZONED */
539 unsigned int sg_timeout;
540 unsigned int sg_reserved_size;
542 struct mutex debugfs_mutex;
543 #ifdef CONFIG_BLK_DEV_IO_TRACE
544 struct blk_trace __rcu *blk_trace;
547 * for flush operations
549 struct blk_flush_queue *fq;
551 struct list_head requeue_list;
552 spinlock_t requeue_lock;
553 struct delayed_work requeue_work;
555 struct mutex sysfs_lock;
556 struct mutex sysfs_dir_lock;
559 * for reusing dead hctx instance in case of updating
562 struct list_head unused_hctx_list;
563 spinlock_t unused_hctx_lock;
567 #if defined(CONFIG_BLK_DEV_BSG)
568 struct bsg_class_device bsg_dev;
571 #ifdef CONFIG_BLK_DEV_THROTTLING
573 struct throtl_data *td;
575 struct rcu_head rcu_head;
576 wait_queue_head_t mq_freeze_wq;
578 * Protect concurrent access to q_usage_counter by
579 * percpu_ref_kill() and percpu_ref_reinit().
581 struct mutex mq_freeze_lock;
583 struct blk_mq_tag_set *tag_set;
584 struct list_head tag_set_list;
585 struct bio_set bio_split;
587 struct dentry *debugfs_dir;
589 #ifdef CONFIG_BLK_DEBUG_FS
590 struct dentry *sched_debugfs_dir;
591 struct dentry *rqos_debugfs_dir;
594 bool mq_sysfs_init_done;
598 #define BLK_MAX_WRITE_HINTS 5
599 u64 write_hints[BLK_MAX_WRITE_HINTS];
602 /* Keep blk_queue_flag_name[] in sync with the definitions below */
603 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
604 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
605 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
606 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
607 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
608 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
609 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
610 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
611 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
612 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
613 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
614 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
615 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
616 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
617 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
618 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
619 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
620 #define QUEUE_FLAG_WC 17 /* Write back caching */
621 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
622 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
623 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
624 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
625 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
626 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
627 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
628 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
629 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
630 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
631 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
632 #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */
634 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
635 (1 << QUEUE_FLAG_SAME_COMP) | \
636 (1 << QUEUE_FLAG_NOWAIT))
638 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
639 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
640 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
642 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
643 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
644 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
645 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
646 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
647 #define blk_queue_noxmerges(q) \
648 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
649 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
650 #define blk_queue_stable_writes(q) \
651 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
652 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
653 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
654 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
655 #define blk_queue_zone_resetall(q) \
656 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
657 #define blk_queue_secure_erase(q) \
658 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
659 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
660 #define blk_queue_scsi_passthrough(q) \
661 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
662 #define blk_queue_pci_p2pdma(q) \
663 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
664 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
665 #define blk_queue_rq_alloc_time(q) \
666 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
668 #define blk_queue_rq_alloc_time(q) false
671 #define blk_noretry_request(rq) \
672 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
673 REQ_FAILFAST_DRIVER))
674 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
675 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
676 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
677 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
678 #define blk_queue_nowait(q) test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
680 extern void blk_set_pm_only(struct request_queue *q);
681 extern void blk_clear_pm_only(struct request_queue *q);
683 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
685 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
687 #define rq_dma_dir(rq) \
688 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
690 #define dma_map_bvec(dev, bv, dir, attrs) \
691 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
694 #define queue_to_disk(q) (dev_to_disk(kobj_to_dev((q)->kobj.parent)))
696 static inline bool queue_is_mq(struct request_queue *q)
702 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
704 return q->rpm_status;
707 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
713 static inline enum blk_zoned_model
714 blk_queue_zoned_model(struct request_queue *q)
716 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
717 return q->limits.zoned;
718 return BLK_ZONED_NONE;
721 static inline bool blk_queue_is_zoned(struct request_queue *q)
723 switch (blk_queue_zoned_model(q)) {
732 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
734 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
737 #ifdef CONFIG_BLK_DEV_ZONED
738 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
740 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
743 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
746 if (!blk_queue_is_zoned(q))
748 return sector >> ilog2(q->limits.chunk_sectors);
751 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
754 if (!blk_queue_is_zoned(q))
756 if (!q->conv_zones_bitmap)
758 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
761 static inline void blk_queue_max_open_zones(struct request_queue *q,
762 unsigned int max_open_zones)
764 q->max_open_zones = max_open_zones;
767 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
769 return q->max_open_zones;
772 static inline void blk_queue_max_active_zones(struct request_queue *q,
773 unsigned int max_active_zones)
775 q->max_active_zones = max_active_zones;
778 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
780 return q->max_active_zones;
782 #else /* CONFIG_BLK_DEV_ZONED */
783 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
787 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
792 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
797 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
801 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
805 #endif /* CONFIG_BLK_DEV_ZONED */
807 static inline bool rq_is_sync(struct request *rq)
809 return op_is_sync(rq->cmd_flags);
812 static inline bool rq_mergeable(struct request *rq)
814 if (blk_rq_is_passthrough(rq))
817 if (req_op(rq) == REQ_OP_FLUSH)
820 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
823 if (req_op(rq) == REQ_OP_ZONE_APPEND)
826 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
828 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
834 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
836 if (bio_page(a) == bio_page(b) &&
837 bio_offset(a) == bio_offset(b))
843 static inline unsigned int blk_queue_depth(struct request_queue *q)
846 return q->queue_depth;
848 return q->nr_requests;
852 * default timeout for SG_IO if none specified
854 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
855 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
861 unsigned long offset;
866 struct req_iterator {
867 struct bvec_iter iter;
871 /* This should not be used directly - use rq_for_each_segment */
872 #define for_each_bio(_bio) \
873 for (; _bio; _bio = _bio->bi_next)
874 #define __rq_for_each_bio(_bio, rq) \
876 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
878 #define rq_for_each_segment(bvl, _rq, _iter) \
879 __rq_for_each_bio(_iter.bio, _rq) \
880 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
882 #define rq_for_each_bvec(bvl, _rq, _iter) \
883 __rq_for_each_bio(_iter.bio, _rq) \
884 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
886 #define rq_iter_last(bvec, _iter) \
887 (_iter.bio->bi_next == NULL && \
888 bio_iter_last(bvec, _iter.iter))
890 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
891 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
893 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
894 extern void rq_flush_dcache_pages(struct request *rq);
896 static inline void rq_flush_dcache_pages(struct request *rq)
901 extern int blk_register_queue(struct gendisk *disk);
902 extern void blk_unregister_queue(struct gendisk *disk);
903 blk_qc_t submit_bio_noacct(struct bio *bio);
904 extern void blk_rq_init(struct request_queue *q, struct request *rq);
905 extern void blk_put_request(struct request *);
906 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
907 blk_mq_req_flags_t flags);
908 extern int blk_lld_busy(struct request_queue *q);
909 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
910 struct bio_set *bs, gfp_t gfp_mask,
911 int (*bio_ctr)(struct bio *, struct bio *, void *),
913 extern void blk_rq_unprep_clone(struct request *rq);
914 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
916 int blk_rq_append_bio(struct request *rq, struct bio *bio);
917 extern void blk_queue_split(struct bio **);
918 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
919 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
920 unsigned int, void __user *);
921 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
922 unsigned int, void __user *);
923 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
924 struct scsi_ioctl_command __user *);
925 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
926 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
928 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
929 extern void blk_queue_exit(struct request_queue *q);
930 extern void blk_sync_queue(struct request_queue *q);
931 extern int blk_rq_map_user(struct request_queue *, struct request *,
932 struct rq_map_data *, void __user *, unsigned long,
934 extern int blk_rq_unmap_user(struct bio *);
935 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
936 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
937 struct rq_map_data *, const struct iov_iter *,
939 extern void blk_execute_rq(struct gendisk *, struct request *, int);
940 extern void blk_execute_rq_nowait(struct gendisk *,
941 struct request *, int, rq_end_io_fn *);
943 /* Helper to convert REQ_OP_XXX to its string format XXX */
944 extern const char *blk_op_str(unsigned int op);
946 int blk_status_to_errno(blk_status_t status);
947 blk_status_t errno_to_blk_status(int errno);
949 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
951 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
953 return bdev->bd_disk->queue; /* this is never NULL */
957 * The basic unit of block I/O is a sector. It is used in a number of contexts
958 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
959 * bytes. Variables of type sector_t represent an offset or size that is a
960 * multiple of 512 bytes. Hence these two constants.
963 #define SECTOR_SHIFT 9
966 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
970 * blk_rq_pos() : the current sector
971 * blk_rq_bytes() : bytes left in the entire request
972 * blk_rq_cur_bytes() : bytes left in the current segment
973 * blk_rq_err_bytes() : bytes left till the next error boundary
974 * blk_rq_sectors() : sectors left in the entire request
975 * blk_rq_cur_sectors() : sectors left in the current segment
976 * blk_rq_stats_sectors() : sectors of the entire request used for stats
978 static inline sector_t blk_rq_pos(const struct request *rq)
983 static inline unsigned int blk_rq_bytes(const struct request *rq)
985 return rq->__data_len;
988 static inline int blk_rq_cur_bytes(const struct request *rq)
990 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
993 extern unsigned int blk_rq_err_bytes(const struct request *rq);
995 static inline unsigned int blk_rq_sectors(const struct request *rq)
997 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
1000 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
1002 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
1005 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
1007 return rq->stats_sectors;
1010 #ifdef CONFIG_BLK_DEV_ZONED
1012 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
1013 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
1015 static inline unsigned int bio_zone_no(struct bio *bio)
1017 return blk_queue_zone_no(bdev_get_queue(bio->bi_bdev),
1018 bio->bi_iter.bi_sector);
1021 static inline unsigned int bio_zone_is_seq(struct bio *bio)
1023 return blk_queue_zone_is_seq(bdev_get_queue(bio->bi_bdev),
1024 bio->bi_iter.bi_sector);
1027 static inline unsigned int blk_rq_zone_no(struct request *rq)
1029 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1032 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1034 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1036 #endif /* CONFIG_BLK_DEV_ZONED */
1039 * Some commands like WRITE SAME have a payload or data transfer size which
1040 * is different from the size of the request. Any driver that supports such
1041 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1042 * calculate the data transfer size.
1044 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1046 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1047 return rq->special_vec.bv_len;
1048 return blk_rq_bytes(rq);
1052 * Return the first full biovec in the request. The caller needs to check that
1053 * there are any bvecs before calling this helper.
1055 static inline struct bio_vec req_bvec(struct request *rq)
1057 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1058 return rq->special_vec;
1059 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1062 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1065 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1066 return min(q->limits.max_discard_sectors,
1067 UINT_MAX >> SECTOR_SHIFT);
1069 if (unlikely(op == REQ_OP_WRITE_SAME))
1070 return q->limits.max_write_same_sectors;
1072 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1073 return q->limits.max_write_zeroes_sectors;
1075 return q->limits.max_sectors;
1079 * Return maximum size of a request at given offset. Only valid for
1080 * file system requests.
1082 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1084 unsigned int chunk_sectors)
1086 if (!chunk_sectors) {
1087 if (q->limits.chunk_sectors)
1088 chunk_sectors = q->limits.chunk_sectors;
1090 return q->limits.max_sectors;
1093 if (likely(is_power_of_2(chunk_sectors)))
1094 chunk_sectors -= offset & (chunk_sectors - 1);
1096 chunk_sectors -= sector_div(offset, chunk_sectors);
1098 return min(q->limits.max_sectors, chunk_sectors);
1101 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1104 struct request_queue *q = rq->q;
1106 if (blk_rq_is_passthrough(rq))
1107 return q->limits.max_hw_sectors;
1109 if (!q->limits.chunk_sectors ||
1110 req_op(rq) == REQ_OP_DISCARD ||
1111 req_op(rq) == REQ_OP_SECURE_ERASE)
1112 return blk_queue_get_max_sectors(q, req_op(rq));
1114 return min(blk_max_size_offset(q, offset, 0),
1115 blk_queue_get_max_sectors(q, req_op(rq)));
1118 static inline unsigned int blk_rq_count_bios(struct request *rq)
1120 unsigned int nr_bios = 0;
1123 __rq_for_each_bio(bio, rq)
1129 void blk_steal_bios(struct bio_list *list, struct request *rq);
1132 * Request completion related functions.
1134 * blk_update_request() completes given number of bytes and updates
1135 * the request without completing it.
1137 extern bool blk_update_request(struct request *rq, blk_status_t error,
1138 unsigned int nr_bytes);
1140 extern void blk_abort_request(struct request *);
1143 * Access functions for manipulating queue properties
1145 extern void blk_cleanup_queue(struct request_queue *);
1146 void blk_queue_bounce_limit(struct request_queue *q, enum blk_bounce limit);
1147 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1148 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1149 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1150 extern void blk_queue_max_discard_segments(struct request_queue *,
1152 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1153 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1154 unsigned int max_discard_sectors);
1155 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1156 unsigned int max_write_same_sectors);
1157 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1158 unsigned int max_write_same_sectors);
1159 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1160 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1161 unsigned int max_zone_append_sectors);
1162 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1163 void blk_queue_zone_write_granularity(struct request_queue *q,
1165 extern void blk_queue_alignment_offset(struct request_queue *q,
1166 unsigned int alignment);
1167 void blk_queue_update_readahead(struct request_queue *q);
1168 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1169 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1170 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1171 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1172 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1173 extern void blk_set_default_limits(struct queue_limits *lim);
1174 extern void blk_set_stacking_limits(struct queue_limits *lim);
1175 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1177 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1179 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1180 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1181 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1182 extern void blk_queue_dma_alignment(struct request_queue *, int);
1183 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1184 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1185 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1186 extern void blk_queue_required_elevator_features(struct request_queue *q,
1187 unsigned int features);
1188 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1189 struct device *dev);
1192 * Number of physical segments as sent to the device.
1194 * Normally this is the number of discontiguous data segments sent by the
1195 * submitter. But for data-less command like discard we might have no
1196 * actual data segments submitted, but the driver might have to add it's
1197 * own special payload. In that case we still return 1 here so that this
1198 * special payload will be mapped.
1200 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1202 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1204 return rq->nr_phys_segments;
1208 * Number of discard segments (or ranges) the driver needs to fill in.
1209 * Each discard bio merged into a request is counted as one segment.
1211 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1213 return max_t(unsigned short, rq->nr_phys_segments, 1);
1216 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1217 struct scatterlist *sglist, struct scatterlist **last_sg);
1218 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1219 struct scatterlist *sglist)
1221 struct scatterlist *last_sg = NULL;
1223 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1225 extern void blk_dump_rq_flags(struct request *, char *);
1227 bool __must_check blk_get_queue(struct request_queue *);
1228 extern void blk_put_queue(struct request_queue *);
1229 extern void blk_set_queue_dying(struct request_queue *);
1233 * blk_plug permits building a queue of related requests by holding the I/O
1234 * fragments for a short period. This allows merging of sequential requests
1235 * into single larger request. As the requests are moved from a per-task list to
1236 * the device's request_queue in a batch, this results in improved scalability
1237 * as the lock contention for request_queue lock is reduced.
1239 * It is ok not to disable preemption when adding the request to the plug list
1240 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1241 * the plug list when the task sleeps by itself. For details, please see
1242 * schedule() where blk_schedule_flush_plug() is called.
1245 struct list_head mq_list; /* blk-mq requests */
1246 struct list_head cb_list; /* md requires an unplug callback */
1247 unsigned short rq_count;
1248 bool multiple_queues;
1251 #define BLK_MAX_REQUEST_COUNT 16
1252 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1255 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1256 struct blk_plug_cb {
1257 struct list_head list;
1258 blk_plug_cb_fn callback;
1261 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1262 void *data, int size);
1263 extern void blk_start_plug(struct blk_plug *);
1264 extern void blk_finish_plug(struct blk_plug *);
1265 extern void blk_flush_plug_list(struct blk_plug *, bool);
1267 static inline void blk_flush_plug(struct task_struct *tsk)
1269 struct blk_plug *plug = tsk->plug;
1272 blk_flush_plug_list(plug, false);
1275 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1277 struct blk_plug *plug = tsk->plug;
1280 blk_flush_plug_list(plug, true);
1283 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1285 struct blk_plug *plug = tsk->plug;
1288 (!list_empty(&plug->mq_list) ||
1289 !list_empty(&plug->cb_list));
1292 int blkdev_issue_flush(struct block_device *bdev);
1293 long nr_blockdev_pages(void);
1294 #else /* CONFIG_BLOCK */
1298 static inline void blk_start_plug(struct blk_plug *plug)
1302 static inline void blk_finish_plug(struct blk_plug *plug)
1306 static inline void blk_flush_plug(struct task_struct *task)
1310 static inline void blk_schedule_flush_plug(struct task_struct *task)
1315 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1320 static inline int blkdev_issue_flush(struct block_device *bdev)
1325 static inline long nr_blockdev_pages(void)
1329 #endif /* CONFIG_BLOCK */
1331 extern void blk_io_schedule(void);
1333 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1334 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1336 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1338 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1339 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1340 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1341 sector_t nr_sects, gfp_t gfp_mask, int flags,
1344 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1345 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1347 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1348 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1350 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1351 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1353 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1354 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1356 return blkdev_issue_discard(sb->s_bdev,
1357 block << (sb->s_blocksize_bits -
1359 nr_blocks << (sb->s_blocksize_bits -
1363 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1364 sector_t nr_blocks, gfp_t gfp_mask)
1366 return blkdev_issue_zeroout(sb->s_bdev,
1367 block << (sb->s_blocksize_bits -
1369 nr_blocks << (sb->s_blocksize_bits -
1374 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1376 static inline bool bdev_is_partition(struct block_device *bdev)
1378 return bdev->bd_partno;
1381 enum blk_default_limits {
1382 BLK_MAX_SEGMENTS = 128,
1383 BLK_SAFE_MAX_SECTORS = 255,
1384 BLK_DEF_MAX_SECTORS = 2560,
1385 BLK_MAX_SEGMENT_SIZE = 65536,
1386 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1389 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1391 return q->limits.seg_boundary_mask;
1394 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1396 return q->limits.virt_boundary_mask;
1399 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1401 return q->limits.max_sectors;
1404 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1406 return q->limits.max_hw_sectors;
1409 static inline unsigned short queue_max_segments(const struct request_queue *q)
1411 return q->limits.max_segments;
1414 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1416 return q->limits.max_discard_segments;
1419 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1421 return q->limits.max_segment_size;
1424 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1427 const struct queue_limits *l = &q->limits;
1429 return min(l->max_zone_append_sectors, l->max_sectors);
1432 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1436 if (q && q->limits.logical_block_size)
1437 retval = q->limits.logical_block_size;
1442 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1444 return queue_logical_block_size(bdev_get_queue(bdev));
1447 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1449 return q->limits.physical_block_size;
1452 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1454 return queue_physical_block_size(bdev_get_queue(bdev));
1457 static inline unsigned int queue_io_min(const struct request_queue *q)
1459 return q->limits.io_min;
1462 static inline int bdev_io_min(struct block_device *bdev)
1464 return queue_io_min(bdev_get_queue(bdev));
1467 static inline unsigned int queue_io_opt(const struct request_queue *q)
1469 return q->limits.io_opt;
1472 static inline int bdev_io_opt(struct block_device *bdev)
1474 return queue_io_opt(bdev_get_queue(bdev));
1477 static inline unsigned int
1478 queue_zone_write_granularity(const struct request_queue *q)
1480 return q->limits.zone_write_granularity;
1483 static inline unsigned int
1484 bdev_zone_write_granularity(struct block_device *bdev)
1486 return queue_zone_write_granularity(bdev_get_queue(bdev));
1489 static inline int queue_alignment_offset(const struct request_queue *q)
1491 if (q->limits.misaligned)
1494 return q->limits.alignment_offset;
1497 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1499 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1500 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1503 return (granularity + lim->alignment_offset - alignment) % granularity;
1506 static inline int bdev_alignment_offset(struct block_device *bdev)
1508 struct request_queue *q = bdev_get_queue(bdev);
1510 if (q->limits.misaligned)
1512 if (bdev_is_partition(bdev))
1513 return queue_limit_alignment_offset(&q->limits,
1514 bdev->bd_start_sect);
1515 return q->limits.alignment_offset;
1518 static inline int queue_discard_alignment(const struct request_queue *q)
1520 if (q->limits.discard_misaligned)
1523 return q->limits.discard_alignment;
1526 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1528 unsigned int alignment, granularity, offset;
1530 if (!lim->max_discard_sectors)
1533 /* Why are these in bytes, not sectors? */
1534 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1535 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1539 /* Offset of the partition start in 'granularity' sectors */
1540 offset = sector_div(sector, granularity);
1542 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1543 offset = (granularity + alignment - offset) % granularity;
1545 /* Turn it back into bytes, gaah */
1546 return offset << SECTOR_SHIFT;
1549 static inline int bdev_discard_alignment(struct block_device *bdev)
1551 struct request_queue *q = bdev_get_queue(bdev);
1553 if (bdev_is_partition(bdev))
1554 return queue_limit_discard_alignment(&q->limits,
1555 bdev->bd_start_sect);
1556 return q->limits.discard_alignment;
1559 static inline unsigned int bdev_write_same(struct block_device *bdev)
1561 struct request_queue *q = bdev_get_queue(bdev);
1564 return q->limits.max_write_same_sectors;
1569 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1571 struct request_queue *q = bdev_get_queue(bdev);
1574 return q->limits.max_write_zeroes_sectors;
1579 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1581 struct request_queue *q = bdev_get_queue(bdev);
1584 return blk_queue_zoned_model(q);
1586 return BLK_ZONED_NONE;
1589 static inline bool bdev_is_zoned(struct block_device *bdev)
1591 struct request_queue *q = bdev_get_queue(bdev);
1594 return blk_queue_is_zoned(q);
1599 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1601 struct request_queue *q = bdev_get_queue(bdev);
1604 return blk_queue_zone_sectors(q);
1608 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1610 struct request_queue *q = bdev_get_queue(bdev);
1613 return queue_max_open_zones(q);
1617 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1619 struct request_queue *q = bdev_get_queue(bdev);
1622 return queue_max_active_zones(q);
1626 static inline int queue_dma_alignment(const struct request_queue *q)
1628 return q ? q->dma_alignment : 511;
1631 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1634 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1635 return !(addr & alignment) && !(len & alignment);
1638 /* assumes size > 256 */
1639 static inline unsigned int blksize_bits(unsigned int size)
1641 unsigned int bits = 8;
1645 } while (size > 256);
1649 static inline unsigned int block_size(struct block_device *bdev)
1651 return 1 << bdev->bd_inode->i_blkbits;
1654 int kblockd_schedule_work(struct work_struct *work);
1655 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1657 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1658 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1659 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1660 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1662 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1664 enum blk_integrity_flags {
1665 BLK_INTEGRITY_VERIFY = 1 << 0,
1666 BLK_INTEGRITY_GENERATE = 1 << 1,
1667 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1668 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1671 struct blk_integrity_iter {
1675 unsigned int data_size;
1676 unsigned short interval;
1677 const char *disk_name;
1680 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1681 typedef void (integrity_prepare_fn) (struct request *);
1682 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1684 struct blk_integrity_profile {
1685 integrity_processing_fn *generate_fn;
1686 integrity_processing_fn *verify_fn;
1687 integrity_prepare_fn *prepare_fn;
1688 integrity_complete_fn *complete_fn;
1692 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1693 extern void blk_integrity_unregister(struct gendisk *);
1694 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1695 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1696 struct scatterlist *);
1697 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1699 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1701 struct blk_integrity *bi = &disk->queue->integrity;
1710 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1712 return blk_get_integrity(bdev->bd_disk);
1716 blk_integrity_queue_supports_integrity(struct request_queue *q)
1718 return q->integrity.profile;
1721 static inline bool blk_integrity_rq(struct request *rq)
1723 return rq->cmd_flags & REQ_INTEGRITY;
1726 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1729 q->limits.max_integrity_segments = segs;
1732 static inline unsigned short
1733 queue_max_integrity_segments(const struct request_queue *q)
1735 return q->limits.max_integrity_segments;
1739 * bio_integrity_intervals - Return number of integrity intervals for a bio
1740 * @bi: blk_integrity profile for device
1741 * @sectors: Size of the bio in 512-byte sectors
1743 * Description: The block layer calculates everything in 512 byte
1744 * sectors but integrity metadata is done in terms of the data integrity
1745 * interval size of the storage device. Convert the block layer sectors
1746 * to the appropriate number of integrity intervals.
1748 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1749 unsigned int sectors)
1751 return sectors >> (bi->interval_exp - 9);
1754 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1755 unsigned int sectors)
1757 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1761 * Return the first bvec that contains integrity data. Only drivers that are
1762 * limited to a single integrity segment should use this helper.
1764 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1766 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1768 return rq->bio->bi_integrity->bip_vec;
1771 #else /* CONFIG_BLK_DEV_INTEGRITY */
1774 struct block_device;
1776 struct blk_integrity;
1778 static inline int blk_integrity_rq(struct request *rq)
1782 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1787 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1789 struct scatterlist *s)
1793 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1797 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1802 blk_integrity_queue_supports_integrity(struct request_queue *q)
1806 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1810 static inline void blk_integrity_register(struct gendisk *d,
1811 struct blk_integrity *b)
1814 static inline void blk_integrity_unregister(struct gendisk *d)
1817 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1821 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1826 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1827 unsigned int sectors)
1832 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1833 unsigned int sectors)
1838 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1843 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1845 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1847 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1849 void blk_ksm_unregister(struct request_queue *q);
1851 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1853 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1854 struct request_queue *q)
1859 static inline void blk_ksm_unregister(struct request_queue *q) { }
1861 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1864 struct block_device_operations {
1865 blk_qc_t (*submit_bio) (struct bio *bio);
1866 int (*open) (struct block_device *, fmode_t);
1867 void (*release) (struct gendisk *, fmode_t);
1868 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1869 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1870 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1871 unsigned int (*check_events) (struct gendisk *disk,
1872 unsigned int clearing);
1873 void (*unlock_native_capacity) (struct gendisk *);
1874 int (*getgeo)(struct block_device *, struct hd_geometry *);
1875 int (*set_read_only)(struct block_device *bdev, bool ro);
1876 /* this callback is with swap_lock and sometimes page table lock held */
1877 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1878 int (*report_zones)(struct gendisk *, sector_t sector,
1879 unsigned int nr_zones, report_zones_cb cb, void *data);
1880 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1881 struct module *owner;
1882 const struct pr_ops *pr_ops;
1885 #ifdef CONFIG_COMPAT
1886 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1887 unsigned int, unsigned long);
1889 #define blkdev_compat_ptr_ioctl NULL
1892 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1893 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1894 struct writeback_control *);
1896 #ifdef CONFIG_BLK_DEV_ZONED
1897 bool blk_req_needs_zone_write_lock(struct request *rq);
1898 bool blk_req_zone_write_trylock(struct request *rq);
1899 void __blk_req_zone_write_lock(struct request *rq);
1900 void __blk_req_zone_write_unlock(struct request *rq);
1902 static inline void blk_req_zone_write_lock(struct request *rq)
1904 if (blk_req_needs_zone_write_lock(rq))
1905 __blk_req_zone_write_lock(rq);
1908 static inline void blk_req_zone_write_unlock(struct request *rq)
1910 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1911 __blk_req_zone_write_unlock(rq);
1914 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1916 return rq->q->seq_zones_wlock &&
1917 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1920 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1922 if (!blk_req_needs_zone_write_lock(rq))
1924 return !blk_req_zone_is_write_locked(rq);
1927 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1932 static inline void blk_req_zone_write_lock(struct request *rq)
1936 static inline void blk_req_zone_write_unlock(struct request *rq)
1939 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1944 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1948 #endif /* CONFIG_BLK_DEV_ZONED */
1950 static inline void blk_wake_io_task(struct task_struct *waiter)
1953 * If we're polling, the task itself is doing the completions. For
1954 * that case, we don't need to signal a wakeup, it's enough to just
1955 * mark us as RUNNING.
1957 if (waiter == current)
1958 __set_current_state(TASK_RUNNING);
1960 wake_up_process(waiter);
1963 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1965 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1966 unsigned long start_time);
1968 unsigned long bio_start_io_acct(struct bio *bio);
1969 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1970 struct block_device *orig_bdev);
1973 * bio_end_io_acct - end I/O accounting for bio based drivers
1974 * @bio: bio to end account for
1975 * @start: start time returned by bio_start_io_acct()
1977 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1979 return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1982 int bdev_read_only(struct block_device *bdev);
1983 int set_blocksize(struct block_device *bdev, int size);
1985 const char *bdevname(struct block_device *bdev, char *buffer);
1986 int lookup_bdev(const char *pathname, dev_t *dev);
1988 void blkdev_show(struct seq_file *seqf, off_t offset);
1990 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1991 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1993 #define BLKDEV_MAJOR_MAX 512
1995 #define BLKDEV_MAJOR_MAX 0
1998 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2000 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
2001 int bd_prepare_to_claim(struct block_device *bdev, void *holder);
2002 void bd_abort_claiming(struct block_device *bdev, void *holder);
2003 void blkdev_put(struct block_device *bdev, fmode_t mode);
2005 /* just for blk-cgroup, don't use elsewhere */
2006 struct block_device *blkdev_get_no_open(dev_t dev);
2007 void blkdev_put_no_open(struct block_device *bdev);
2009 struct block_device *bdev_alloc(struct gendisk *disk, u8 partno);
2010 void bdev_add(struct block_device *bdev, dev_t dev);
2011 struct block_device *I_BDEV(struct inode *inode);
2012 struct block_device *bdgrab(struct block_device *bdev);
2013 void bdput(struct block_device *);
2014 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
2018 void invalidate_bdev(struct block_device *bdev);
2019 int sync_blockdev(struct block_device *bdev);
2021 static inline void invalidate_bdev(struct block_device *bdev)
2024 static inline int sync_blockdev(struct block_device *bdev)
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 */