1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef _LINUX_BLKDEV_H
3 #define _LINUX_BLKDEV_H
5 #include <linux/sched.h>
6 #include <linux/sched/clock.h>
7 #include <linux/major.h>
8 #include <linux/genhd.h>
9 #include <linux/list.h>
10 #include <linux/llist.h>
11 #include <linux/timer.h>
12 #include <linux/workqueue.h>
13 #include <linux/pagemap.h>
14 #include <linux/backing-dev-defs.h>
15 #include <linux/wait.h>
16 #include <linux/mempool.h>
17 #include <linux/pfn.h>
18 #include <linux/bio.h>
19 #include <linux/stringify.h>
20 #include <linux/gfp.h>
21 #include <linux/bsg.h>
22 #include <linux/smp.h>
23 #include <linux/rcupdate.h>
24 #include <linux/percpu-refcount.h>
25 #include <linux/scatterlist.h>
26 #include <linux/blkzoned.h>
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 5
61 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
65 typedef __u32 __bitwise req_flags_t;
67 /* elevator knows about this request */
68 #define RQF_SORTED ((__force req_flags_t)(1 << 0))
69 /* drive already may have started this one */
70 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
71 /* may not be passed by ioscheduler */
72 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
73 /* request for flush sequence */
74 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
75 /* merge of different types, fail separately */
76 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
77 /* track inflight for MQ */
78 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
79 /* don't call prep for this one */
80 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
81 /* set for "ide_preempt" requests and also for requests for which the SCSI
82 "quiesce" state must be ignored. */
83 #define RQF_PREEMPT ((__force req_flags_t)(1 << 8))
84 /* vaguely specified driver internal error. Ignored by the block layer */
85 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
86 /* don't warn about errors */
87 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
88 /* elevator private data attached */
89 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
90 /* account into disk and partition IO statistics */
91 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
92 /* request came from our alloc pool */
93 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
94 /* runtime pm request */
95 #define RQF_PM ((__force req_flags_t)(1 << 15))
96 /* on IO scheduler merge hash */
97 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
98 /* track IO completion time */
99 #define RQF_STATS ((__force req_flags_t)(1 << 17))
100 /* Look at ->special_vec for the actual data payload instead of the
102 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
103 /* The per-zone write lock is held for this request */
104 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
105 /* already slept for hybrid poll */
106 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
107 /* ->timeout has been called, don't expire again */
108 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
110 /* flags that prevent us from merging requests: */
111 #define RQF_NOMERGE_FLAGS \
112 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
115 * Request state for blk-mq.
124 * Try to put the fields that are referenced together in the same cacheline.
126 * If you modify this structure, make sure to update blk_rq_init() and
127 * especially blk_mq_rq_ctx_init() to take care of the added fields.
130 struct request_queue *q;
131 struct blk_mq_ctx *mq_ctx;
132 struct blk_mq_hw_ctx *mq_hctx;
134 unsigned int cmd_flags; /* op and common flags */
135 req_flags_t rq_flags;
140 /* the following two fields are internal, NEVER access directly */
141 unsigned int __data_len; /* total data len */
142 sector_t __sector; /* sector cursor */
147 struct list_head queuelist;
150 * The hash is used inside the scheduler, and killed once the
151 * request reaches the dispatch list. The ipi_list is only used
152 * to queue the request for softirq completion, which is long
153 * after the request has been unhashed (and even removed from
154 * the dispatch list).
157 struct hlist_node hash; /* merge hash */
158 struct list_head ipi_list;
162 * The rb_node is only used inside the io scheduler, requests
163 * are pruned when moved to the dispatch queue. So let the
164 * completion_data share space with the rb_node.
167 struct rb_node rb_node; /* sort/lookup */
168 struct bio_vec special_vec;
169 void *completion_data;
170 int error_count; /* for legacy drivers, don't use */
174 * Three pointers are available for the IO schedulers, if they need
175 * more they have to dynamically allocate it. Flush requests are
176 * never put on the IO scheduler. So let the flush fields share
177 * space with the elevator data.
187 struct list_head list;
188 rq_end_io_fn *saved_end_io;
192 struct gendisk *rq_disk;
193 struct hd_struct *part;
194 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
195 /* Time that the first bio started allocating this request. */
198 /* Time that this request was allocated for this IO. */
200 /* Time that I/O was submitted to the device. */
201 u64 io_start_time_ns;
203 #ifdef CONFIG_BLK_WBT
204 unsigned short wbt_flags;
207 * rq sectors used for blk stats. It has the same value
208 * with blk_rq_sectors(rq), except that it never be zeroed
211 unsigned short stats_sectors;
214 * Number of scatter-gather DMA addr+len pairs after
215 * physical address coalescing is performed.
217 unsigned short nr_phys_segments;
219 #if defined(CONFIG_BLK_DEV_INTEGRITY)
220 unsigned short nr_integrity_segments;
223 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
224 struct bio_crypt_ctx *crypt_ctx;
225 struct blk_ksm_keyslot *crypt_keyslot;
228 unsigned short write_hint;
229 unsigned short ioprio;
231 enum mq_rq_state state;
234 unsigned int timeout;
235 unsigned long deadline;
238 struct __call_single_data csd;
243 * completion callback.
245 rq_end_io_fn *end_io;
249 static inline bool blk_op_is_scsi(unsigned int op)
251 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
254 static inline bool blk_op_is_private(unsigned int op)
256 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
259 static inline bool blk_rq_is_scsi(struct request *rq)
261 return blk_op_is_scsi(req_op(rq));
264 static inline bool blk_rq_is_private(struct request *rq)
266 return blk_op_is_private(req_op(rq));
269 static inline bool blk_rq_is_passthrough(struct request *rq)
271 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
274 static inline bool bio_is_passthrough(struct bio *bio)
276 unsigned op = bio_op(bio);
278 return blk_op_is_scsi(op) || blk_op_is_private(op);
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 */
320 struct queue_limits {
321 unsigned long bounce_pfn;
322 unsigned long seg_boundary_mask;
323 unsigned long virt_boundary_mask;
325 unsigned int max_hw_sectors;
326 unsigned int max_dev_sectors;
327 unsigned int chunk_sectors;
328 unsigned int max_sectors;
329 unsigned int max_segment_size;
330 unsigned int physical_block_size;
331 unsigned int logical_block_size;
332 unsigned int alignment_offset;
335 unsigned int max_discard_sectors;
336 unsigned int max_hw_discard_sectors;
337 unsigned int max_write_same_sectors;
338 unsigned int max_write_zeroes_sectors;
339 unsigned int max_zone_append_sectors;
340 unsigned int discard_granularity;
341 unsigned int discard_alignment;
343 unsigned short max_segments;
344 unsigned short max_integrity_segments;
345 unsigned short max_discard_segments;
347 unsigned char misaligned;
348 unsigned char discard_misaligned;
349 unsigned char raid_partial_stripes_expensive;
350 enum blk_zoned_model zoned;
353 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
356 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
358 #ifdef CONFIG_BLK_DEV_ZONED
360 #define BLK_ALL_ZONES ((unsigned int)-1)
361 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
362 unsigned int nr_zones, report_zones_cb cb, void *data);
363 unsigned int blkdev_nr_zones(struct gendisk *disk);
364 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
365 sector_t sectors, sector_t nr_sectors,
367 int blk_revalidate_disk_zones(struct gendisk *disk,
368 void (*update_driver_data)(struct gendisk *disk));
370 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
371 unsigned int cmd, unsigned long arg);
372 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
373 unsigned int cmd, unsigned long arg);
375 #else /* CONFIG_BLK_DEV_ZONED */
377 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
382 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
383 fmode_t mode, unsigned int cmd,
389 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
390 fmode_t mode, unsigned int cmd,
396 #endif /* CONFIG_BLK_DEV_ZONED */
398 struct request_queue {
399 struct request *last_merge;
400 struct elevator_queue *elevator;
402 struct percpu_ref q_usage_counter;
404 struct blk_queue_stats *stats;
405 struct rq_qos *rq_qos;
407 const struct blk_mq_ops *mq_ops;
410 struct blk_mq_ctx __percpu *queue_ctx;
412 unsigned int queue_depth;
414 /* hw dispatch queues */
415 struct blk_mq_hw_ctx **queue_hw_ctx;
416 unsigned int nr_hw_queues;
418 struct backing_dev_info *backing_dev_info;
421 * The queue owner gets to use this for whatever they like.
422 * ll_rw_blk doesn't touch it.
427 * various queue flags, see QUEUE_* below
429 unsigned long queue_flags;
431 * Number of contexts that have called blk_set_pm_only(). If this
432 * counter is above zero then only RQF_PM and RQF_PREEMPT requests are
438 * ida allocated id for this queue. Used to index queues from
444 * queue needs bounce pages for pages above this limit
448 spinlock_t queue_lock;
458 struct kobject *mq_kobj;
460 #ifdef CONFIG_BLK_DEV_INTEGRITY
461 struct blk_integrity integrity;
462 #endif /* CONFIG_BLK_DEV_INTEGRITY */
466 enum rpm_status rpm_status;
467 unsigned int nr_pending;
473 unsigned long nr_requests; /* Max # of requests */
475 unsigned int dma_pad_mask;
476 unsigned int dma_alignment;
478 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
479 /* Inline crypto capabilities */
480 struct blk_keyslot_manager *ksm;
483 unsigned int rq_timeout;
486 struct blk_stat_callback *poll_cb;
487 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
489 struct timer_list timeout;
490 struct work_struct timeout_work;
492 atomic_t nr_active_requests_shared_sbitmap;
494 struct list_head icq_list;
495 #ifdef CONFIG_BLK_CGROUP
496 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
497 struct blkcg_gq *root_blkg;
498 struct list_head blkg_list;
501 struct queue_limits limits;
503 unsigned int required_elevator_features;
505 #ifdef CONFIG_BLK_DEV_ZONED
507 * Zoned block device information for request dispatch control.
508 * nr_zones is the total number of zones of the device. This is always
509 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
510 * bits which indicates if a zone is conventional (bit set) or
511 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
512 * bits which indicates if a zone is write locked, that is, if a write
513 * request targeting the zone was dispatched. All three fields are
514 * initialized by the low level device driver (e.g. scsi/sd.c).
515 * Stacking drivers (device mappers) may or may not initialize
518 * Reads of this information must be protected with blk_queue_enter() /
519 * blk_queue_exit(). Modifying this information is only allowed while
520 * no requests are being processed. See also blk_mq_freeze_queue() and
521 * blk_mq_unfreeze_queue().
523 unsigned int nr_zones;
524 unsigned long *conv_zones_bitmap;
525 unsigned long *seq_zones_wlock;
526 unsigned int max_open_zones;
527 unsigned int max_active_zones;
528 #endif /* CONFIG_BLK_DEV_ZONED */
533 unsigned int sg_timeout;
534 unsigned int sg_reserved_size;
536 struct mutex debugfs_mutex;
537 #ifdef CONFIG_BLK_DEV_IO_TRACE
538 struct blk_trace __rcu *blk_trace;
541 * for flush operations
543 struct blk_flush_queue *fq;
545 struct list_head requeue_list;
546 spinlock_t requeue_lock;
547 struct delayed_work requeue_work;
549 struct mutex sysfs_lock;
550 struct mutex sysfs_dir_lock;
553 * for reusing dead hctx instance in case of updating
556 struct list_head unused_hctx_list;
557 spinlock_t unused_hctx_lock;
561 #if defined(CONFIG_BLK_DEV_BSG)
562 struct bsg_class_device bsg_dev;
565 #ifdef CONFIG_BLK_DEV_THROTTLING
567 struct throtl_data *td;
569 struct rcu_head rcu_head;
570 wait_queue_head_t mq_freeze_wq;
572 * Protect concurrent access to q_usage_counter by
573 * percpu_ref_kill() and percpu_ref_reinit().
575 struct mutex mq_freeze_lock;
577 struct blk_mq_tag_set *tag_set;
578 struct list_head tag_set_list;
579 struct bio_set bio_split;
581 struct dentry *debugfs_dir;
583 #ifdef CONFIG_BLK_DEBUG_FS
584 struct dentry *sched_debugfs_dir;
585 struct dentry *rqos_debugfs_dir;
588 bool mq_sysfs_init_done;
592 #define BLK_MAX_WRITE_HINTS 5
593 u64 write_hints[BLK_MAX_WRITE_HINTS];
596 /* Keep blk_queue_flag_name[] in sync with the definitions below */
597 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
598 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
599 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
600 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
601 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
602 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
603 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
604 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
605 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
606 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
607 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
608 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
609 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
610 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
611 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
612 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
613 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
614 #define QUEUE_FLAG_WC 17 /* Write back caching */
615 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
616 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
617 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
618 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
619 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
620 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
621 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
622 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
623 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
624 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
625 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
626 #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */
628 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
629 (1 << QUEUE_FLAG_SAME_COMP) | \
630 (1 << QUEUE_FLAG_NOWAIT))
632 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
633 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
634 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
636 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
637 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
638 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
639 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
640 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
641 #define blk_queue_noxmerges(q) \
642 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
643 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
644 #define blk_queue_stable_writes(q) \
645 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
646 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
647 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
648 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
649 #define blk_queue_zone_resetall(q) \
650 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
651 #define blk_queue_secure_erase(q) \
652 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
653 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
654 #define blk_queue_scsi_passthrough(q) \
655 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
656 #define blk_queue_pci_p2pdma(q) \
657 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
658 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
659 #define blk_queue_rq_alloc_time(q) \
660 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
662 #define blk_queue_rq_alloc_time(q) false
665 #define blk_noretry_request(rq) \
666 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
667 REQ_FAILFAST_DRIVER))
668 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
669 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
670 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
671 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
672 #define blk_queue_nowait(q) test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
674 extern void blk_set_pm_only(struct request_queue *q);
675 extern void blk_clear_pm_only(struct request_queue *q);
677 static inline bool blk_account_rq(struct request *rq)
679 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
682 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
684 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
686 #define rq_dma_dir(rq) \
687 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
689 #define dma_map_bvec(dev, bv, dir, attrs) \
690 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
693 static inline bool queue_is_mq(struct request_queue *q)
698 static inline enum blk_zoned_model
699 blk_queue_zoned_model(struct request_queue *q)
701 return q->limits.zoned;
704 static inline bool blk_queue_is_zoned(struct request_queue *q)
706 switch (blk_queue_zoned_model(q)) {
715 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
717 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
720 #ifdef CONFIG_BLK_DEV_ZONED
721 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
723 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
726 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
729 if (!blk_queue_is_zoned(q))
731 return sector >> ilog2(q->limits.chunk_sectors);
734 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
737 if (!blk_queue_is_zoned(q))
739 if (!q->conv_zones_bitmap)
741 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
744 static inline void blk_queue_max_open_zones(struct request_queue *q,
745 unsigned int max_open_zones)
747 q->max_open_zones = max_open_zones;
750 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
752 return q->max_open_zones;
755 static inline void blk_queue_max_active_zones(struct request_queue *q,
756 unsigned int max_active_zones)
758 q->max_active_zones = max_active_zones;
761 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
763 return q->max_active_zones;
765 #else /* CONFIG_BLK_DEV_ZONED */
766 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
770 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
775 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
780 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
784 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
788 #endif /* CONFIG_BLK_DEV_ZONED */
790 static inline bool rq_is_sync(struct request *rq)
792 return op_is_sync(rq->cmd_flags);
795 static inline bool rq_mergeable(struct request *rq)
797 if (blk_rq_is_passthrough(rq))
800 if (req_op(rq) == REQ_OP_FLUSH)
803 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
806 if (req_op(rq) == REQ_OP_ZONE_APPEND)
809 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
811 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
817 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
819 if (bio_page(a) == bio_page(b) &&
820 bio_offset(a) == bio_offset(b))
826 static inline unsigned int blk_queue_depth(struct request_queue *q)
829 return q->queue_depth;
831 return q->nr_requests;
834 extern unsigned long blk_max_low_pfn, blk_max_pfn;
837 * standard bounce addresses:
839 * BLK_BOUNCE_HIGH : bounce all highmem pages
840 * BLK_BOUNCE_ANY : don't bounce anything
841 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
844 #if BITS_PER_LONG == 32
845 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
847 #define BLK_BOUNCE_HIGH -1ULL
849 #define BLK_BOUNCE_ANY (-1ULL)
850 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
853 * default timeout for SG_IO if none specified
855 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
856 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
862 unsigned long offset;
867 struct req_iterator {
868 struct bvec_iter iter;
872 /* This should not be used directly - use rq_for_each_segment */
873 #define for_each_bio(_bio) \
874 for (; _bio; _bio = _bio->bi_next)
875 #define __rq_for_each_bio(_bio, rq) \
877 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
879 #define rq_for_each_segment(bvl, _rq, _iter) \
880 __rq_for_each_bio(_iter.bio, _rq) \
881 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
883 #define rq_for_each_bvec(bvl, _rq, _iter) \
884 __rq_for_each_bio(_iter.bio, _rq) \
885 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
887 #define rq_iter_last(bvec, _iter) \
888 (_iter.bio->bi_next == NULL && \
889 bio_iter_last(bvec, _iter.iter))
891 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
892 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
894 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
895 extern void rq_flush_dcache_pages(struct request *rq);
897 static inline void rq_flush_dcache_pages(struct request *rq)
902 extern int blk_register_queue(struct gendisk *disk);
903 extern void blk_unregister_queue(struct gendisk *disk);
904 blk_qc_t submit_bio_noacct(struct bio *bio);
905 extern void blk_rq_init(struct request_queue *q, struct request *rq);
906 extern void blk_put_request(struct request *);
907 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
908 blk_mq_req_flags_t flags);
909 extern int blk_lld_busy(struct request_queue *q);
910 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
911 struct bio_set *bs, gfp_t gfp_mask,
912 int (*bio_ctr)(struct bio *, struct bio *, void *),
914 extern void blk_rq_unprep_clone(struct request *rq);
915 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
917 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
918 extern void blk_queue_split(struct bio **);
919 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
920 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
921 unsigned int, void __user *);
922 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
923 unsigned int, void __user *);
924 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
925 struct scsi_ioctl_command __user *);
926 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
927 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
929 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
930 extern void blk_queue_exit(struct request_queue *q);
931 extern void blk_sync_queue(struct request_queue *q);
932 extern int blk_rq_map_user(struct request_queue *, struct request *,
933 struct rq_map_data *, void __user *, unsigned long,
935 extern int blk_rq_unmap_user(struct bio *);
936 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
937 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
938 struct rq_map_data *, const struct iov_iter *,
940 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
941 struct request *, int);
942 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
943 struct request *, int, rq_end_io_fn *);
945 /* Helper to convert REQ_OP_XXX to its string format XXX */
946 extern const char *blk_op_str(unsigned int op);
948 int blk_status_to_errno(blk_status_t status);
949 blk_status_t errno_to_blk_status(int errno);
951 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
953 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
955 return bdev->bd_disk->queue; /* this is never NULL */
959 * The basic unit of block I/O is a sector. It is used in a number of contexts
960 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
961 * bytes. Variables of type sector_t represent an offset or size that is a
962 * multiple of 512 bytes. Hence these two constants.
965 #define SECTOR_SHIFT 9
968 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
972 * blk_rq_pos() : the current sector
973 * blk_rq_bytes() : bytes left in the entire request
974 * blk_rq_cur_bytes() : bytes left in the current segment
975 * blk_rq_err_bytes() : bytes left till the next error boundary
976 * blk_rq_sectors() : sectors left in the entire request
977 * blk_rq_cur_sectors() : sectors left in the current segment
978 * blk_rq_stats_sectors() : sectors of the entire request used for stats
980 static inline sector_t blk_rq_pos(const struct request *rq)
985 static inline unsigned int blk_rq_bytes(const struct request *rq)
987 return rq->__data_len;
990 static inline int blk_rq_cur_bytes(const struct request *rq)
992 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
995 extern unsigned int blk_rq_err_bytes(const struct request *rq);
997 static inline unsigned int blk_rq_sectors(const struct request *rq)
999 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
1002 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
1004 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
1007 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
1009 return rq->stats_sectors;
1012 #ifdef CONFIG_BLK_DEV_ZONED
1014 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
1015 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
1017 static inline unsigned int blk_rq_zone_no(struct request *rq)
1019 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1022 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1024 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1026 #endif /* CONFIG_BLK_DEV_ZONED */
1029 * Some commands like WRITE SAME have a payload or data transfer size which
1030 * is different from the size of the request. Any driver that supports such
1031 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1032 * calculate the data transfer size.
1034 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1036 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1037 return rq->special_vec.bv_len;
1038 return blk_rq_bytes(rq);
1042 * Return the first full biovec in the request. The caller needs to check that
1043 * there are any bvecs before calling this helper.
1045 static inline struct bio_vec req_bvec(struct request *rq)
1047 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1048 return rq->special_vec;
1049 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1052 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1055 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1056 return min(q->limits.max_discard_sectors,
1057 UINT_MAX >> SECTOR_SHIFT);
1059 if (unlikely(op == REQ_OP_WRITE_SAME))
1060 return q->limits.max_write_same_sectors;
1062 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1063 return q->limits.max_write_zeroes_sectors;
1065 return q->limits.max_sectors;
1069 * Return maximum size of a request at given offset. Only valid for
1070 * file system requests.
1072 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1075 unsigned int chunk_sectors = q->limits.chunk_sectors;
1078 return q->limits.max_sectors;
1080 if (likely(is_power_of_2(chunk_sectors)))
1081 chunk_sectors -= offset & (chunk_sectors - 1);
1083 chunk_sectors -= sector_div(offset, chunk_sectors);
1085 return min(q->limits.max_sectors, chunk_sectors);
1088 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1091 struct request_queue *q = rq->q;
1093 if (blk_rq_is_passthrough(rq))
1094 return q->limits.max_hw_sectors;
1096 if (!q->limits.chunk_sectors ||
1097 req_op(rq) == REQ_OP_DISCARD ||
1098 req_op(rq) == REQ_OP_SECURE_ERASE)
1099 return blk_queue_get_max_sectors(q, req_op(rq));
1101 return min(blk_max_size_offset(q, offset),
1102 blk_queue_get_max_sectors(q, req_op(rq)));
1105 static inline unsigned int blk_rq_count_bios(struct request *rq)
1107 unsigned int nr_bios = 0;
1110 __rq_for_each_bio(bio, rq)
1116 void blk_steal_bios(struct bio_list *list, struct request *rq);
1119 * Request completion related functions.
1121 * blk_update_request() completes given number of bytes and updates
1122 * the request without completing it.
1124 extern bool blk_update_request(struct request *rq, blk_status_t error,
1125 unsigned int nr_bytes);
1127 extern void blk_abort_request(struct request *);
1130 * Access functions for manipulating queue properties
1132 extern void blk_cleanup_queue(struct request_queue *);
1133 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1134 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1135 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1136 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1137 extern void blk_queue_max_discard_segments(struct request_queue *,
1139 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1140 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1141 unsigned int max_discard_sectors);
1142 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1143 unsigned int max_write_same_sectors);
1144 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1145 unsigned int max_write_same_sectors);
1146 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1147 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1148 unsigned int max_zone_append_sectors);
1149 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1150 extern void blk_queue_alignment_offset(struct request_queue *q,
1151 unsigned int alignment);
1152 void blk_queue_update_readahead(struct request_queue *q);
1153 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1154 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1155 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1156 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1157 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1158 extern void blk_set_default_limits(struct queue_limits *lim);
1159 extern void blk_set_stacking_limits(struct queue_limits *lim);
1160 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1162 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1164 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1165 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1166 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1167 extern void blk_queue_dma_alignment(struct request_queue *, int);
1168 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1169 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1170 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1171 extern void blk_queue_required_elevator_features(struct request_queue *q,
1172 unsigned int features);
1173 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1174 struct device *dev);
1177 * Number of physical segments as sent to the device.
1179 * Normally this is the number of discontiguous data segments sent by the
1180 * submitter. But for data-less command like discard we might have no
1181 * actual data segments submitted, but the driver might have to add it's
1182 * own special payload. In that case we still return 1 here so that this
1183 * special payload will be mapped.
1185 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1187 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1189 return rq->nr_phys_segments;
1193 * Number of discard segments (or ranges) the driver needs to fill in.
1194 * Each discard bio merged into a request is counted as one segment.
1196 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1198 return max_t(unsigned short, rq->nr_phys_segments, 1);
1201 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1202 struct scatterlist *sglist, struct scatterlist **last_sg);
1203 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1204 struct scatterlist *sglist)
1206 struct scatterlist *last_sg = NULL;
1208 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1210 extern void blk_dump_rq_flags(struct request *, char *);
1212 bool __must_check blk_get_queue(struct request_queue *);
1213 struct request_queue *blk_alloc_queue(int node_id);
1214 extern void blk_put_queue(struct request_queue *);
1215 extern void blk_set_queue_dying(struct request_queue *);
1219 * blk_plug permits building a queue of related requests by holding the I/O
1220 * fragments for a short period. This allows merging of sequential requests
1221 * into single larger request. As the requests are moved from a per-task list to
1222 * the device's request_queue in a batch, this results in improved scalability
1223 * as the lock contention for request_queue lock is reduced.
1225 * It is ok not to disable preemption when adding the request to the plug list
1226 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1227 * the plug list when the task sleeps by itself. For details, please see
1228 * schedule() where blk_schedule_flush_plug() is called.
1231 struct list_head mq_list; /* blk-mq requests */
1232 struct list_head cb_list; /* md requires an unplug callback */
1233 unsigned short rq_count;
1234 bool multiple_queues;
1237 #define BLK_MAX_REQUEST_COUNT 16
1238 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1241 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1242 struct blk_plug_cb {
1243 struct list_head list;
1244 blk_plug_cb_fn callback;
1247 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1248 void *data, int size);
1249 extern void blk_start_plug(struct blk_plug *);
1250 extern void blk_finish_plug(struct blk_plug *);
1251 extern void blk_flush_plug_list(struct blk_plug *, bool);
1253 static inline void blk_flush_plug(struct task_struct *tsk)
1255 struct blk_plug *plug = tsk->plug;
1258 blk_flush_plug_list(plug, false);
1261 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1263 struct blk_plug *plug = tsk->plug;
1266 blk_flush_plug_list(plug, true);
1269 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1271 struct blk_plug *plug = tsk->plug;
1274 (!list_empty(&plug->mq_list) ||
1275 !list_empty(&plug->cb_list));
1278 int blkdev_issue_flush(struct block_device *, gfp_t);
1279 long nr_blockdev_pages(void);
1280 #else /* CONFIG_BLOCK */
1284 static inline void blk_start_plug(struct blk_plug *plug)
1288 static inline void blk_finish_plug(struct blk_plug *plug)
1292 static inline void blk_flush_plug(struct task_struct *task)
1296 static inline void blk_schedule_flush_plug(struct task_struct *task)
1301 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1306 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask)
1311 static inline long nr_blockdev_pages(void)
1315 #endif /* CONFIG_BLOCK */
1317 extern void blk_io_schedule(void);
1319 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1320 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1322 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1324 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1325 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1326 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1327 sector_t nr_sects, gfp_t gfp_mask, int flags,
1330 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1331 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1333 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1334 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1336 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1337 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1339 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1340 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1342 return blkdev_issue_discard(sb->s_bdev,
1343 block << (sb->s_blocksize_bits -
1345 nr_blocks << (sb->s_blocksize_bits -
1349 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1350 sector_t nr_blocks, gfp_t gfp_mask)
1352 return blkdev_issue_zeroout(sb->s_bdev,
1353 block << (sb->s_blocksize_bits -
1355 nr_blocks << (sb->s_blocksize_bits -
1360 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1362 static inline bool bdev_is_partition(struct block_device *bdev)
1364 return bdev->bd_partno;
1367 enum blk_default_limits {
1368 BLK_MAX_SEGMENTS = 128,
1369 BLK_SAFE_MAX_SECTORS = 255,
1370 BLK_DEF_MAX_SECTORS = 2560,
1371 BLK_MAX_SEGMENT_SIZE = 65536,
1372 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1375 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1377 return q->limits.seg_boundary_mask;
1380 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1382 return q->limits.virt_boundary_mask;
1385 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1387 return q->limits.max_sectors;
1390 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1392 return q->limits.max_hw_sectors;
1395 static inline unsigned short queue_max_segments(const struct request_queue *q)
1397 return q->limits.max_segments;
1400 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1402 return q->limits.max_discard_segments;
1405 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1407 return q->limits.max_segment_size;
1410 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1413 const struct queue_limits *l = &q->limits;
1415 return min(l->max_zone_append_sectors, l->max_sectors);
1418 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1422 if (q && q->limits.logical_block_size)
1423 retval = q->limits.logical_block_size;
1428 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1430 return queue_logical_block_size(bdev_get_queue(bdev));
1433 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1435 return q->limits.physical_block_size;
1438 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1440 return queue_physical_block_size(bdev_get_queue(bdev));
1443 static inline unsigned int queue_io_min(const struct request_queue *q)
1445 return q->limits.io_min;
1448 static inline int bdev_io_min(struct block_device *bdev)
1450 return queue_io_min(bdev_get_queue(bdev));
1453 static inline unsigned int queue_io_opt(const struct request_queue *q)
1455 return q->limits.io_opt;
1458 static inline int bdev_io_opt(struct block_device *bdev)
1460 return queue_io_opt(bdev_get_queue(bdev));
1463 static inline int queue_alignment_offset(const struct request_queue *q)
1465 if (q->limits.misaligned)
1468 return q->limits.alignment_offset;
1471 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1473 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1474 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1477 return (granularity + lim->alignment_offset - alignment) % granularity;
1480 static inline int bdev_alignment_offset(struct block_device *bdev)
1482 struct request_queue *q = bdev_get_queue(bdev);
1484 if (q->limits.misaligned)
1486 if (bdev_is_partition(bdev))
1487 return queue_limit_alignment_offset(&q->limits,
1488 bdev->bd_part->start_sect);
1489 return q->limits.alignment_offset;
1492 static inline int queue_discard_alignment(const struct request_queue *q)
1494 if (q->limits.discard_misaligned)
1497 return q->limits.discard_alignment;
1500 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1502 unsigned int alignment, granularity, offset;
1504 if (!lim->max_discard_sectors)
1507 /* Why are these in bytes, not sectors? */
1508 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1509 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1513 /* Offset of the partition start in 'granularity' sectors */
1514 offset = sector_div(sector, granularity);
1516 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1517 offset = (granularity + alignment - offset) % granularity;
1519 /* Turn it back into bytes, gaah */
1520 return offset << SECTOR_SHIFT;
1523 static inline int bdev_discard_alignment(struct block_device *bdev)
1525 struct request_queue *q = bdev_get_queue(bdev);
1527 if (bdev_is_partition(bdev))
1528 return queue_limit_discard_alignment(&q->limits,
1529 bdev->bd_part->start_sect);
1530 return q->limits.discard_alignment;
1533 static inline unsigned int bdev_write_same(struct block_device *bdev)
1535 struct request_queue *q = bdev_get_queue(bdev);
1538 return q->limits.max_write_same_sectors;
1543 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1545 struct request_queue *q = bdev_get_queue(bdev);
1548 return q->limits.max_write_zeroes_sectors;
1553 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1555 struct request_queue *q = bdev_get_queue(bdev);
1558 return blk_queue_zoned_model(q);
1560 return BLK_ZONED_NONE;
1563 static inline bool bdev_is_zoned(struct block_device *bdev)
1565 struct request_queue *q = bdev_get_queue(bdev);
1568 return blk_queue_is_zoned(q);
1573 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1575 struct request_queue *q = bdev_get_queue(bdev);
1578 return blk_queue_zone_sectors(q);
1582 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1584 struct request_queue *q = bdev_get_queue(bdev);
1587 return queue_max_open_zones(q);
1591 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1593 struct request_queue *q = bdev_get_queue(bdev);
1596 return queue_max_active_zones(q);
1600 static inline int queue_dma_alignment(const struct request_queue *q)
1602 return q ? q->dma_alignment : 511;
1605 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1608 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1609 return !(addr & alignment) && !(len & alignment);
1612 /* assumes size > 256 */
1613 static inline unsigned int blksize_bits(unsigned int size)
1615 unsigned int bits = 8;
1619 } while (size > 256);
1623 static inline unsigned int block_size(struct block_device *bdev)
1625 return 1 << bdev->bd_inode->i_blkbits;
1628 int kblockd_schedule_work(struct work_struct *work);
1629 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1631 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1632 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1633 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1634 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1636 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1638 enum blk_integrity_flags {
1639 BLK_INTEGRITY_VERIFY = 1 << 0,
1640 BLK_INTEGRITY_GENERATE = 1 << 1,
1641 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1642 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1645 struct blk_integrity_iter {
1649 unsigned int data_size;
1650 unsigned short interval;
1651 const char *disk_name;
1654 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1655 typedef void (integrity_prepare_fn) (struct request *);
1656 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1658 struct blk_integrity_profile {
1659 integrity_processing_fn *generate_fn;
1660 integrity_processing_fn *verify_fn;
1661 integrity_prepare_fn *prepare_fn;
1662 integrity_complete_fn *complete_fn;
1666 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1667 extern void blk_integrity_unregister(struct gendisk *);
1668 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1669 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1670 struct scatterlist *);
1671 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1673 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1675 struct blk_integrity *bi = &disk->queue->integrity;
1684 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1686 return blk_get_integrity(bdev->bd_disk);
1690 blk_integrity_queue_supports_integrity(struct request_queue *q)
1692 return q->integrity.profile;
1695 static inline bool blk_integrity_rq(struct request *rq)
1697 return rq->cmd_flags & REQ_INTEGRITY;
1700 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1703 q->limits.max_integrity_segments = segs;
1706 static inline unsigned short
1707 queue_max_integrity_segments(const struct request_queue *q)
1709 return q->limits.max_integrity_segments;
1713 * bio_integrity_intervals - Return number of integrity intervals for a bio
1714 * @bi: blk_integrity profile for device
1715 * @sectors: Size of the bio in 512-byte sectors
1717 * Description: The block layer calculates everything in 512 byte
1718 * sectors but integrity metadata is done in terms of the data integrity
1719 * interval size of the storage device. Convert the block layer sectors
1720 * to the appropriate number of integrity intervals.
1722 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1723 unsigned int sectors)
1725 return sectors >> (bi->interval_exp - 9);
1728 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1729 unsigned int sectors)
1731 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1735 * Return the first bvec that contains integrity data. Only drivers that are
1736 * limited to a single integrity segment should use this helper.
1738 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1740 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1742 return rq->bio->bi_integrity->bip_vec;
1745 #else /* CONFIG_BLK_DEV_INTEGRITY */
1748 struct block_device;
1750 struct blk_integrity;
1752 static inline int blk_integrity_rq(struct request *rq)
1756 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1761 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1763 struct scatterlist *s)
1767 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1771 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1776 blk_integrity_queue_supports_integrity(struct request_queue *q)
1780 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1784 static inline void blk_integrity_register(struct gendisk *d,
1785 struct blk_integrity *b)
1788 static inline void blk_integrity_unregister(struct gendisk *d)
1791 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1795 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1800 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1801 unsigned int sectors)
1806 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1807 unsigned int sectors)
1812 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1817 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1819 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1821 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1823 void blk_ksm_unregister(struct request_queue *q);
1825 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1827 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1828 struct request_queue *q)
1833 static inline void blk_ksm_unregister(struct request_queue *q) { }
1835 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1838 struct block_device_operations {
1839 blk_qc_t (*submit_bio) (struct bio *bio);
1840 int (*open) (struct block_device *, fmode_t);
1841 void (*release) (struct gendisk *, fmode_t);
1842 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1843 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1844 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1845 unsigned int (*check_events) (struct gendisk *disk,
1846 unsigned int clearing);
1847 void (*unlock_native_capacity) (struct gendisk *);
1848 int (*revalidate_disk) (struct gendisk *);
1849 int (*getgeo)(struct block_device *, struct hd_geometry *);
1850 /* this callback is with swap_lock and sometimes page table lock held */
1851 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1852 int (*report_zones)(struct gendisk *, sector_t sector,
1853 unsigned int nr_zones, report_zones_cb cb, void *data);
1854 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1855 struct module *owner;
1856 const struct pr_ops *pr_ops;
1859 #ifdef CONFIG_COMPAT
1860 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1861 unsigned int, unsigned long);
1863 #define blkdev_compat_ptr_ioctl NULL
1866 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1868 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1869 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1870 struct writeback_control *);
1872 #ifdef CONFIG_BLK_DEV_ZONED
1873 bool blk_req_needs_zone_write_lock(struct request *rq);
1874 bool blk_req_zone_write_trylock(struct request *rq);
1875 void __blk_req_zone_write_lock(struct request *rq);
1876 void __blk_req_zone_write_unlock(struct request *rq);
1878 static inline void blk_req_zone_write_lock(struct request *rq)
1880 if (blk_req_needs_zone_write_lock(rq))
1881 __blk_req_zone_write_lock(rq);
1884 static inline void blk_req_zone_write_unlock(struct request *rq)
1886 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1887 __blk_req_zone_write_unlock(rq);
1890 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1892 return rq->q->seq_zones_wlock &&
1893 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1896 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1898 if (!blk_req_needs_zone_write_lock(rq))
1900 return !blk_req_zone_is_write_locked(rq);
1903 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1908 static inline void blk_req_zone_write_lock(struct request *rq)
1912 static inline void blk_req_zone_write_unlock(struct request *rq)
1915 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1920 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1924 #endif /* CONFIG_BLK_DEV_ZONED */
1926 static inline void blk_wake_io_task(struct task_struct *waiter)
1929 * If we're polling, the task itself is doing the completions. For
1930 * that case, we don't need to signal a wakeup, it's enough to just
1931 * mark us as RUNNING.
1933 if (waiter == current)
1934 __set_current_state(TASK_RUNNING);
1936 wake_up_process(waiter);
1939 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1941 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1942 unsigned long start_time);
1944 unsigned long part_start_io_acct(struct gendisk *disk, struct hd_struct **part,
1946 void part_end_io_acct(struct hd_struct *part, struct bio *bio,
1947 unsigned long start_time);
1950 * bio_start_io_acct - start I/O accounting for bio based drivers
1951 * @bio: bio to start account for
1953 * Returns the start time that should be passed back to bio_end_io_acct().
1955 static inline unsigned long bio_start_io_acct(struct bio *bio)
1957 return disk_start_io_acct(bio->bi_disk, bio_sectors(bio), bio_op(bio));
1961 * bio_end_io_acct - end I/O accounting for bio based drivers
1962 * @bio: bio to end account for
1963 * @start: start time returned by bio_start_io_acct()
1965 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1967 return disk_end_io_acct(bio->bi_disk, bio_op(bio), start_time);
1970 int bdev_read_only(struct block_device *bdev);
1971 int set_blocksize(struct block_device *bdev, int size);
1973 const char *bdevname(struct block_device *bdev, char *buffer);
1974 struct block_device *lookup_bdev(const char *);
1976 void blkdev_show(struct seq_file *seqf, off_t offset);
1978 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1979 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1981 #define BLKDEV_MAJOR_MAX 512
1983 #define BLKDEV_MAJOR_MAX 0
1986 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1988 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1989 int bd_prepare_to_claim(struct block_device *bdev, struct block_device *whole,
1991 void bd_abort_claiming(struct block_device *bdev, struct block_device *whole,
1993 void blkdev_put(struct block_device *bdev, fmode_t mode);
1995 struct block_device *I_BDEV(struct inode *inode);
1996 struct block_device *bdget_part(struct hd_struct *part);
1997 struct block_device *bdgrab(struct block_device *bdev);
1998 void bdput(struct block_device *);
2001 void invalidate_bdev(struct block_device *bdev);
2002 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
2004 int sync_blockdev(struct block_device *bdev);
2006 static inline void invalidate_bdev(struct block_device *bdev)
2009 static inline int truncate_bdev_range(struct block_device *bdev, fmode_t mode,
2010 loff_t lstart, loff_t lend)
2014 static inline int sync_blockdev(struct block_device *bdev)
2019 int fsync_bdev(struct block_device *bdev);
2021 struct super_block *freeze_bdev(struct block_device *bdev);
2022 int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2024 #endif /* _LINUX_BLKDEV_H */