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 6
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_passthrough(unsigned int op)
246 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
249 static inline bool blk_rq_is_passthrough(struct request *rq)
251 return blk_op_is_passthrough(req_op(rq));
254 static inline unsigned short req_get_ioprio(struct request *req)
259 #include <linux/elevator.h>
261 struct blk_queue_ctx;
265 enum blk_eh_timer_return {
266 BLK_EH_DONE, /* drivers has completed the command */
267 BLK_EH_RESET_TIMER, /* reset timer and try again */
270 enum blk_queue_state {
275 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
276 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
278 #define BLK_SCSI_MAX_CMDS (256)
279 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
282 * Zoned block device models (zoned limit).
284 * Note: This needs to be ordered from the least to the most severe
285 * restrictions for the inheritance in blk_stack_limits() to work.
287 enum blk_zoned_model {
288 BLK_ZONED_NONE = 0, /* Regular block device */
289 BLK_ZONED_HA, /* Host-aware zoned block device */
290 BLK_ZONED_HM, /* Host-managed zoned block device */
294 * BLK_BOUNCE_NONE: never bounce (default)
295 * BLK_BOUNCE_HIGH: bounce all highmem pages
302 struct queue_limits {
303 enum blk_bounce bounce;
304 unsigned long seg_boundary_mask;
305 unsigned long virt_boundary_mask;
307 unsigned int max_hw_sectors;
308 unsigned int max_dev_sectors;
309 unsigned int chunk_sectors;
310 unsigned int max_sectors;
311 unsigned int max_segment_size;
312 unsigned int physical_block_size;
313 unsigned int logical_block_size;
314 unsigned int alignment_offset;
317 unsigned int max_discard_sectors;
318 unsigned int max_hw_discard_sectors;
319 unsigned int max_write_same_sectors;
320 unsigned int max_write_zeroes_sectors;
321 unsigned int max_zone_append_sectors;
322 unsigned int discard_granularity;
323 unsigned int discard_alignment;
324 unsigned int zone_write_granularity;
326 unsigned short max_segments;
327 unsigned short max_integrity_segments;
328 unsigned short max_discard_segments;
330 unsigned char misaligned;
331 unsigned char discard_misaligned;
332 unsigned char raid_partial_stripes_expensive;
333 enum blk_zoned_model zoned;
336 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
339 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
341 #ifdef CONFIG_BLK_DEV_ZONED
343 #define BLK_ALL_ZONES ((unsigned int)-1)
344 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
345 unsigned int nr_zones, report_zones_cb cb, void *data);
346 unsigned int blkdev_nr_zones(struct gendisk *disk);
347 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
348 sector_t sectors, sector_t nr_sectors,
350 int blk_revalidate_disk_zones(struct gendisk *disk,
351 void (*update_driver_data)(struct gendisk *disk));
353 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
354 unsigned int cmd, unsigned long arg);
355 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
356 unsigned int cmd, unsigned long arg);
358 #else /* CONFIG_BLK_DEV_ZONED */
360 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
365 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
366 fmode_t mode, unsigned int cmd,
372 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
373 fmode_t mode, unsigned int cmd,
379 #endif /* CONFIG_BLK_DEV_ZONED */
381 struct request_queue {
382 struct request *last_merge;
383 struct elevator_queue *elevator;
385 struct percpu_ref q_usage_counter;
387 struct blk_queue_stats *stats;
388 struct rq_qos *rq_qos;
390 const struct blk_mq_ops *mq_ops;
393 struct blk_mq_ctx __percpu *queue_ctx;
395 unsigned int queue_depth;
397 /* hw dispatch queues */
398 struct blk_mq_hw_ctx **queue_hw_ctx;
399 unsigned int nr_hw_queues;
401 struct backing_dev_info *backing_dev_info;
404 * The queue owner gets to use this for whatever they like.
405 * ll_rw_blk doesn't touch it.
410 * various queue flags, see QUEUE_* below
412 unsigned long queue_flags;
414 * Number of contexts that have called blk_set_pm_only(). If this
415 * counter is above zero then only RQF_PM requests are processed.
420 * ida allocated id for this queue. Used to index queues from
425 spinlock_t queue_lock;
435 struct kobject *mq_kobj;
437 #ifdef CONFIG_BLK_DEV_INTEGRITY
438 struct blk_integrity integrity;
439 #endif /* CONFIG_BLK_DEV_INTEGRITY */
443 enum rpm_status rpm_status;
449 unsigned long nr_requests; /* Max # of requests */
451 unsigned int dma_pad_mask;
452 unsigned int dma_alignment;
454 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
455 /* Inline crypto capabilities */
456 struct blk_keyslot_manager *ksm;
459 unsigned int rq_timeout;
462 struct blk_stat_callback *poll_cb;
463 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
465 struct timer_list timeout;
466 struct work_struct timeout_work;
468 atomic_t nr_active_requests_shared_sbitmap;
470 struct sbitmap_queue sched_bitmap_tags;
471 struct sbitmap_queue sched_breserved_tags;
473 struct list_head icq_list;
474 #ifdef CONFIG_BLK_CGROUP
475 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
476 struct blkcg_gq *root_blkg;
477 struct list_head blkg_list;
480 struct queue_limits limits;
482 unsigned int required_elevator_features;
484 #ifdef CONFIG_BLK_DEV_ZONED
486 * Zoned block device information for request dispatch control.
487 * nr_zones is the total number of zones of the device. This is always
488 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
489 * bits which indicates if a zone is conventional (bit set) or
490 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
491 * bits which indicates if a zone is write locked, that is, if a write
492 * request targeting the zone was dispatched. All three fields are
493 * initialized by the low level device driver (e.g. scsi/sd.c).
494 * Stacking drivers (device mappers) may or may not initialize
497 * Reads of this information must be protected with blk_queue_enter() /
498 * blk_queue_exit(). Modifying this information is only allowed while
499 * no requests are being processed. See also blk_mq_freeze_queue() and
500 * blk_mq_unfreeze_queue().
502 unsigned int nr_zones;
503 unsigned long *conv_zones_bitmap;
504 unsigned long *seq_zones_wlock;
505 unsigned int max_open_zones;
506 unsigned int max_active_zones;
507 #endif /* CONFIG_BLK_DEV_ZONED */
512 unsigned int sg_timeout;
513 unsigned int sg_reserved_size;
515 struct mutex debugfs_mutex;
516 #ifdef CONFIG_BLK_DEV_IO_TRACE
517 struct blk_trace __rcu *blk_trace;
520 * for flush operations
522 struct blk_flush_queue *fq;
524 struct list_head requeue_list;
525 spinlock_t requeue_lock;
526 struct delayed_work requeue_work;
528 struct mutex sysfs_lock;
529 struct mutex sysfs_dir_lock;
532 * for reusing dead hctx instance in case of updating
535 struct list_head unused_hctx_list;
536 spinlock_t unused_hctx_lock;
540 #if defined(CONFIG_BLK_DEV_BSG)
541 struct bsg_class_device bsg_dev;
544 #ifdef CONFIG_BLK_DEV_THROTTLING
546 struct throtl_data *td;
548 struct rcu_head rcu_head;
549 wait_queue_head_t mq_freeze_wq;
551 * Protect concurrent access to q_usage_counter by
552 * percpu_ref_kill() and percpu_ref_reinit().
554 struct mutex mq_freeze_lock;
556 struct blk_mq_tag_set *tag_set;
557 struct list_head tag_set_list;
558 struct bio_set bio_split;
560 struct dentry *debugfs_dir;
562 #ifdef CONFIG_BLK_DEBUG_FS
563 struct dentry *sched_debugfs_dir;
564 struct dentry *rqos_debugfs_dir;
567 bool mq_sysfs_init_done;
571 #define BLK_MAX_WRITE_HINTS 5
572 u64 write_hints[BLK_MAX_WRITE_HINTS];
575 /* Keep blk_queue_flag_name[] in sync with the definitions below */
576 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
577 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
578 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
579 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
580 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
581 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
582 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
583 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
584 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
585 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
586 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
587 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
588 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
589 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
590 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
591 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
592 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
593 #define QUEUE_FLAG_WC 17 /* Write back caching */
594 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
595 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
596 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
597 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
598 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
599 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
600 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
601 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
602 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
603 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
604 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
605 #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */
607 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
608 (1 << QUEUE_FLAG_SAME_COMP) | \
609 (1 << QUEUE_FLAG_NOWAIT))
611 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
612 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
613 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
615 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
616 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
617 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
618 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
619 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
620 #define blk_queue_noxmerges(q) \
621 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
622 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
623 #define blk_queue_stable_writes(q) \
624 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
625 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
626 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
627 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
628 #define blk_queue_zone_resetall(q) \
629 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
630 #define blk_queue_secure_erase(q) \
631 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
632 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
633 #define blk_queue_scsi_passthrough(q) \
634 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
635 #define blk_queue_pci_p2pdma(q) \
636 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
637 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
638 #define blk_queue_rq_alloc_time(q) \
639 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
641 #define blk_queue_rq_alloc_time(q) false
644 #define blk_noretry_request(rq) \
645 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
646 REQ_FAILFAST_DRIVER))
647 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
648 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
649 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
650 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
651 #define blk_queue_nowait(q) test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
653 extern void blk_set_pm_only(struct request_queue *q);
654 extern void blk_clear_pm_only(struct request_queue *q);
656 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
658 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
660 #define rq_dma_dir(rq) \
661 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
663 #define dma_map_bvec(dev, bv, dir, attrs) \
664 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
667 #define queue_to_disk(q) (dev_to_disk(kobj_to_dev((q)->kobj.parent)))
669 static inline bool queue_is_mq(struct request_queue *q)
675 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
677 return q->rpm_status;
680 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
686 static inline enum blk_zoned_model
687 blk_queue_zoned_model(struct request_queue *q)
689 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
690 return q->limits.zoned;
691 return BLK_ZONED_NONE;
694 static inline bool blk_queue_is_zoned(struct request_queue *q)
696 switch (blk_queue_zoned_model(q)) {
705 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
707 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
710 #ifdef CONFIG_BLK_DEV_ZONED
711 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
713 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
716 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
719 if (!blk_queue_is_zoned(q))
721 return sector >> ilog2(q->limits.chunk_sectors);
724 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
727 if (!blk_queue_is_zoned(q))
729 if (!q->conv_zones_bitmap)
731 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
734 static inline void blk_queue_max_open_zones(struct request_queue *q,
735 unsigned int max_open_zones)
737 q->max_open_zones = max_open_zones;
740 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
742 return q->max_open_zones;
745 static inline void blk_queue_max_active_zones(struct request_queue *q,
746 unsigned int max_active_zones)
748 q->max_active_zones = max_active_zones;
751 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
753 return q->max_active_zones;
755 #else /* CONFIG_BLK_DEV_ZONED */
756 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
760 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
765 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
770 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
774 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
778 #endif /* CONFIG_BLK_DEV_ZONED */
780 static inline bool rq_is_sync(struct request *rq)
782 return op_is_sync(rq->cmd_flags);
785 static inline bool rq_mergeable(struct request *rq)
787 if (blk_rq_is_passthrough(rq))
790 if (req_op(rq) == REQ_OP_FLUSH)
793 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
796 if (req_op(rq) == REQ_OP_ZONE_APPEND)
799 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
801 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
807 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
809 if (bio_page(a) == bio_page(b) &&
810 bio_offset(a) == bio_offset(b))
816 static inline unsigned int blk_queue_depth(struct request_queue *q)
819 return q->queue_depth;
821 return q->nr_requests;
825 * default timeout for SG_IO if none specified
827 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
828 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
834 unsigned long offset;
839 struct req_iterator {
840 struct bvec_iter iter;
844 /* This should not be used directly - use rq_for_each_segment */
845 #define for_each_bio(_bio) \
846 for (; _bio; _bio = _bio->bi_next)
847 #define __rq_for_each_bio(_bio, rq) \
849 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
851 #define rq_for_each_segment(bvl, _rq, _iter) \
852 __rq_for_each_bio(_iter.bio, _rq) \
853 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
855 #define rq_for_each_bvec(bvl, _rq, _iter) \
856 __rq_for_each_bio(_iter.bio, _rq) \
857 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
859 #define rq_iter_last(bvec, _iter) \
860 (_iter.bio->bi_next == NULL && \
861 bio_iter_last(bvec, _iter.iter))
863 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
864 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
866 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
867 extern void rq_flush_dcache_pages(struct request *rq);
869 static inline void rq_flush_dcache_pages(struct request *rq)
874 extern int blk_register_queue(struct gendisk *disk);
875 extern void blk_unregister_queue(struct gendisk *disk);
876 blk_qc_t submit_bio_noacct(struct bio *bio);
877 extern void blk_rq_init(struct request_queue *q, struct request *rq);
878 extern void blk_put_request(struct request *);
879 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
880 blk_mq_req_flags_t flags);
881 extern int blk_lld_busy(struct request_queue *q);
882 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
883 struct bio_set *bs, gfp_t gfp_mask,
884 int (*bio_ctr)(struct bio *, struct bio *, void *),
886 extern void blk_rq_unprep_clone(struct request *rq);
887 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
889 int blk_rq_append_bio(struct request *rq, struct bio *bio);
890 extern void blk_queue_split(struct bio **);
891 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
892 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
893 unsigned int, void __user *);
894 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
895 unsigned int, void __user *);
896 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
897 struct scsi_ioctl_command __user *);
898 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
899 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
901 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
902 extern void blk_queue_exit(struct request_queue *q);
903 extern void blk_sync_queue(struct request_queue *q);
904 extern int blk_rq_map_user(struct request_queue *, struct request *,
905 struct rq_map_data *, void __user *, unsigned long,
907 extern int blk_rq_unmap_user(struct bio *);
908 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
909 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
910 struct rq_map_data *, const struct iov_iter *,
912 extern void blk_execute_rq_nowait(struct gendisk *,
913 struct request *, int, rq_end_io_fn *);
915 blk_status_t blk_execute_rq(struct gendisk *bd_disk, struct request *rq,
918 /* Helper to convert REQ_OP_XXX to its string format XXX */
919 extern const char *blk_op_str(unsigned int op);
921 int blk_status_to_errno(blk_status_t status);
922 blk_status_t errno_to_blk_status(int errno);
924 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
926 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
928 return bdev->bd_disk->queue; /* this is never NULL */
932 * The basic unit of block I/O is a sector. It is used in a number of contexts
933 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
934 * bytes. Variables of type sector_t represent an offset or size that is a
935 * multiple of 512 bytes. Hence these two constants.
938 #define SECTOR_SHIFT 9
941 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
945 * blk_rq_pos() : the current sector
946 * blk_rq_bytes() : bytes left in the entire request
947 * blk_rq_cur_bytes() : bytes left in the current segment
948 * blk_rq_err_bytes() : bytes left till the next error boundary
949 * blk_rq_sectors() : sectors left in the entire request
950 * blk_rq_cur_sectors() : sectors left in the current segment
951 * blk_rq_stats_sectors() : sectors of the entire request used for stats
953 static inline sector_t blk_rq_pos(const struct request *rq)
958 static inline unsigned int blk_rq_bytes(const struct request *rq)
960 return rq->__data_len;
963 static inline int blk_rq_cur_bytes(const struct request *rq)
965 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
968 extern unsigned int blk_rq_err_bytes(const struct request *rq);
970 static inline unsigned int blk_rq_sectors(const struct request *rq)
972 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
975 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
977 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
980 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
982 return rq->stats_sectors;
985 #ifdef CONFIG_BLK_DEV_ZONED
987 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
988 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
990 static inline unsigned int bio_zone_no(struct bio *bio)
992 return blk_queue_zone_no(bdev_get_queue(bio->bi_bdev),
993 bio->bi_iter.bi_sector);
996 static inline unsigned int bio_zone_is_seq(struct bio *bio)
998 return blk_queue_zone_is_seq(bdev_get_queue(bio->bi_bdev),
999 bio->bi_iter.bi_sector);
1002 static inline unsigned int blk_rq_zone_no(struct request *rq)
1004 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1007 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1009 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1011 #endif /* CONFIG_BLK_DEV_ZONED */
1014 * Some commands like WRITE SAME have a payload or data transfer size which
1015 * is different from the size of the request. Any driver that supports such
1016 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1017 * calculate the data transfer size.
1019 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1021 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1022 return rq->special_vec.bv_len;
1023 return blk_rq_bytes(rq);
1027 * Return the first full biovec in the request. The caller needs to check that
1028 * there are any bvecs before calling this helper.
1030 static inline struct bio_vec req_bvec(struct request *rq)
1032 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1033 return rq->special_vec;
1034 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1037 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1040 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1041 return min(q->limits.max_discard_sectors,
1042 UINT_MAX >> SECTOR_SHIFT);
1044 if (unlikely(op == REQ_OP_WRITE_SAME))
1045 return q->limits.max_write_same_sectors;
1047 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1048 return q->limits.max_write_zeroes_sectors;
1050 return q->limits.max_sectors;
1054 * Return maximum size of a request at given offset. Only valid for
1055 * file system requests.
1057 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1059 unsigned int chunk_sectors)
1061 if (!chunk_sectors) {
1062 if (q->limits.chunk_sectors)
1063 chunk_sectors = q->limits.chunk_sectors;
1065 return q->limits.max_sectors;
1068 if (likely(is_power_of_2(chunk_sectors)))
1069 chunk_sectors -= offset & (chunk_sectors - 1);
1071 chunk_sectors -= sector_div(offset, chunk_sectors);
1073 return min(q->limits.max_sectors, chunk_sectors);
1076 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1079 struct request_queue *q = rq->q;
1081 if (blk_rq_is_passthrough(rq))
1082 return q->limits.max_hw_sectors;
1084 if (!q->limits.chunk_sectors ||
1085 req_op(rq) == REQ_OP_DISCARD ||
1086 req_op(rq) == REQ_OP_SECURE_ERASE)
1087 return blk_queue_get_max_sectors(q, req_op(rq));
1089 return min(blk_max_size_offset(q, offset, 0),
1090 blk_queue_get_max_sectors(q, req_op(rq)));
1093 static inline unsigned int blk_rq_count_bios(struct request *rq)
1095 unsigned int nr_bios = 0;
1098 __rq_for_each_bio(bio, rq)
1104 void blk_steal_bios(struct bio_list *list, struct request *rq);
1107 * Request completion related functions.
1109 * blk_update_request() completes given number of bytes and updates
1110 * the request without completing it.
1112 extern bool blk_update_request(struct request *rq, blk_status_t error,
1113 unsigned int nr_bytes);
1115 extern void blk_abort_request(struct request *);
1118 * Access functions for manipulating queue properties
1120 extern void blk_cleanup_queue(struct request_queue *);
1121 void blk_queue_bounce_limit(struct request_queue *q, enum blk_bounce limit);
1122 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1123 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1124 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1125 extern void blk_queue_max_discard_segments(struct request_queue *,
1127 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1128 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1129 unsigned int max_discard_sectors);
1130 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1131 unsigned int max_write_same_sectors);
1132 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1133 unsigned int max_write_same_sectors);
1134 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1135 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1136 unsigned int max_zone_append_sectors);
1137 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1138 void blk_queue_zone_write_granularity(struct request_queue *q,
1140 extern void blk_queue_alignment_offset(struct request_queue *q,
1141 unsigned int alignment);
1142 void blk_queue_update_readahead(struct request_queue *q);
1143 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1144 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1145 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1146 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1147 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1148 extern void blk_set_default_limits(struct queue_limits *lim);
1149 extern void blk_set_stacking_limits(struct queue_limits *lim);
1150 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1152 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1154 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1155 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1156 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1157 extern void blk_queue_dma_alignment(struct request_queue *, int);
1158 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1159 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1160 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1161 extern void blk_queue_required_elevator_features(struct request_queue *q,
1162 unsigned int features);
1163 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1164 struct device *dev);
1167 * Number of physical segments as sent to the device.
1169 * Normally this is the number of discontiguous data segments sent by the
1170 * submitter. But for data-less command like discard we might have no
1171 * actual data segments submitted, but the driver might have to add it's
1172 * own special payload. In that case we still return 1 here so that this
1173 * special payload will be mapped.
1175 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1177 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1179 return rq->nr_phys_segments;
1183 * Number of discard segments (or ranges) the driver needs to fill in.
1184 * Each discard bio merged into a request is counted as one segment.
1186 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1188 return max_t(unsigned short, rq->nr_phys_segments, 1);
1191 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1192 struct scatterlist *sglist, struct scatterlist **last_sg);
1193 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1194 struct scatterlist *sglist)
1196 struct scatterlist *last_sg = NULL;
1198 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1200 extern void blk_dump_rq_flags(struct request *, char *);
1202 bool __must_check blk_get_queue(struct request_queue *);
1203 extern void blk_put_queue(struct request_queue *);
1204 extern void blk_set_queue_dying(struct request_queue *);
1208 * blk_plug permits building a queue of related requests by holding the I/O
1209 * fragments for a short period. This allows merging of sequential requests
1210 * into single larger request. As the requests are moved from a per-task list to
1211 * the device's request_queue in a batch, this results in improved scalability
1212 * as the lock contention for request_queue lock is reduced.
1214 * It is ok not to disable preemption when adding the request to the plug list
1215 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1216 * the plug list when the task sleeps by itself. For details, please see
1217 * schedule() where blk_schedule_flush_plug() is called.
1220 struct list_head mq_list; /* blk-mq requests */
1221 struct list_head cb_list; /* md requires an unplug callback */
1222 unsigned short rq_count;
1223 bool multiple_queues;
1226 #define BLK_MAX_REQUEST_COUNT 16
1227 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1230 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1231 struct blk_plug_cb {
1232 struct list_head list;
1233 blk_plug_cb_fn callback;
1236 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1237 void *data, int size);
1238 extern void blk_start_plug(struct blk_plug *);
1239 extern void blk_finish_plug(struct blk_plug *);
1240 extern void blk_flush_plug_list(struct blk_plug *, bool);
1242 static inline void blk_flush_plug(struct task_struct *tsk)
1244 struct blk_plug *plug = tsk->plug;
1247 blk_flush_plug_list(plug, false);
1250 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1252 struct blk_plug *plug = tsk->plug;
1255 blk_flush_plug_list(plug, true);
1258 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1260 struct blk_plug *plug = tsk->plug;
1263 (!list_empty(&plug->mq_list) ||
1264 !list_empty(&plug->cb_list));
1267 int blkdev_issue_flush(struct block_device *bdev);
1268 long nr_blockdev_pages(void);
1269 #else /* CONFIG_BLOCK */
1273 static inline void blk_start_plug(struct blk_plug *plug)
1277 static inline void blk_finish_plug(struct blk_plug *plug)
1281 static inline void blk_flush_plug(struct task_struct *task)
1285 static inline void blk_schedule_flush_plug(struct task_struct *task)
1290 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1295 static inline int blkdev_issue_flush(struct block_device *bdev)
1300 static inline long nr_blockdev_pages(void)
1304 #endif /* CONFIG_BLOCK */
1306 extern void blk_io_schedule(void);
1308 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1309 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1311 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1313 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1314 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1315 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1316 sector_t nr_sects, gfp_t gfp_mask, int flags,
1319 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1320 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1322 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1323 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1325 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1326 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1328 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1329 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1331 return blkdev_issue_discard(sb->s_bdev,
1332 block << (sb->s_blocksize_bits -
1334 nr_blocks << (sb->s_blocksize_bits -
1338 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1339 sector_t nr_blocks, gfp_t gfp_mask)
1341 return blkdev_issue_zeroout(sb->s_bdev,
1342 block << (sb->s_blocksize_bits -
1344 nr_blocks << (sb->s_blocksize_bits -
1349 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1351 static inline bool bdev_is_partition(struct block_device *bdev)
1353 return bdev->bd_partno;
1356 enum blk_default_limits {
1357 BLK_MAX_SEGMENTS = 128,
1358 BLK_SAFE_MAX_SECTORS = 255,
1359 BLK_DEF_MAX_SECTORS = 2560,
1360 BLK_MAX_SEGMENT_SIZE = 65536,
1361 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1364 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1366 return q->limits.seg_boundary_mask;
1369 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1371 return q->limits.virt_boundary_mask;
1374 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1376 return q->limits.max_sectors;
1379 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1381 return q->limits.max_hw_sectors;
1384 static inline unsigned short queue_max_segments(const struct request_queue *q)
1386 return q->limits.max_segments;
1389 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1391 return q->limits.max_discard_segments;
1394 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1396 return q->limits.max_segment_size;
1399 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1402 const struct queue_limits *l = &q->limits;
1404 return min(l->max_zone_append_sectors, l->max_sectors);
1407 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1411 if (q && q->limits.logical_block_size)
1412 retval = q->limits.logical_block_size;
1417 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1419 return queue_logical_block_size(bdev_get_queue(bdev));
1422 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1424 return q->limits.physical_block_size;
1427 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1429 return queue_physical_block_size(bdev_get_queue(bdev));
1432 static inline unsigned int queue_io_min(const struct request_queue *q)
1434 return q->limits.io_min;
1437 static inline int bdev_io_min(struct block_device *bdev)
1439 return queue_io_min(bdev_get_queue(bdev));
1442 static inline unsigned int queue_io_opt(const struct request_queue *q)
1444 return q->limits.io_opt;
1447 static inline int bdev_io_opt(struct block_device *bdev)
1449 return queue_io_opt(bdev_get_queue(bdev));
1452 static inline unsigned int
1453 queue_zone_write_granularity(const struct request_queue *q)
1455 return q->limits.zone_write_granularity;
1458 static inline unsigned int
1459 bdev_zone_write_granularity(struct block_device *bdev)
1461 return queue_zone_write_granularity(bdev_get_queue(bdev));
1464 static inline int queue_alignment_offset(const struct request_queue *q)
1466 if (q->limits.misaligned)
1469 return q->limits.alignment_offset;
1472 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1474 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1475 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1478 return (granularity + lim->alignment_offset - alignment) % granularity;
1481 static inline int bdev_alignment_offset(struct block_device *bdev)
1483 struct request_queue *q = bdev_get_queue(bdev);
1485 if (q->limits.misaligned)
1487 if (bdev_is_partition(bdev))
1488 return queue_limit_alignment_offset(&q->limits,
1489 bdev->bd_start_sect);
1490 return q->limits.alignment_offset;
1493 static inline int queue_discard_alignment(const struct request_queue *q)
1495 if (q->limits.discard_misaligned)
1498 return q->limits.discard_alignment;
1501 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1503 unsigned int alignment, granularity, offset;
1505 if (!lim->max_discard_sectors)
1508 /* Why are these in bytes, not sectors? */
1509 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1510 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1514 /* Offset of the partition start in 'granularity' sectors */
1515 offset = sector_div(sector, granularity);
1517 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1518 offset = (granularity + alignment - offset) % granularity;
1520 /* Turn it back into bytes, gaah */
1521 return offset << SECTOR_SHIFT;
1524 static inline int bdev_discard_alignment(struct block_device *bdev)
1526 struct request_queue *q = bdev_get_queue(bdev);
1528 if (bdev_is_partition(bdev))
1529 return queue_limit_discard_alignment(&q->limits,
1530 bdev->bd_start_sect);
1531 return q->limits.discard_alignment;
1534 static inline unsigned int bdev_write_same(struct block_device *bdev)
1536 struct request_queue *q = bdev_get_queue(bdev);
1539 return q->limits.max_write_same_sectors;
1544 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1546 struct request_queue *q = bdev_get_queue(bdev);
1549 return q->limits.max_write_zeroes_sectors;
1554 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1556 struct request_queue *q = bdev_get_queue(bdev);
1559 return blk_queue_zoned_model(q);
1561 return BLK_ZONED_NONE;
1564 static inline bool bdev_is_zoned(struct block_device *bdev)
1566 struct request_queue *q = bdev_get_queue(bdev);
1569 return blk_queue_is_zoned(q);
1574 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1576 struct request_queue *q = bdev_get_queue(bdev);
1579 return blk_queue_zone_sectors(q);
1583 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1585 struct request_queue *q = bdev_get_queue(bdev);
1588 return queue_max_open_zones(q);
1592 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1594 struct request_queue *q = bdev_get_queue(bdev);
1597 return queue_max_active_zones(q);
1601 static inline int queue_dma_alignment(const struct request_queue *q)
1603 return q ? q->dma_alignment : 511;
1606 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1609 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1610 return !(addr & alignment) && !(len & alignment);
1613 /* assumes size > 256 */
1614 static inline unsigned int blksize_bits(unsigned int size)
1616 unsigned int bits = 8;
1620 } while (size > 256);
1624 static inline unsigned int block_size(struct block_device *bdev)
1626 return 1 << bdev->bd_inode->i_blkbits;
1629 int kblockd_schedule_work(struct work_struct *work);
1630 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1632 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1633 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1634 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1635 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1637 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1639 enum blk_integrity_flags {
1640 BLK_INTEGRITY_VERIFY = 1 << 0,
1641 BLK_INTEGRITY_GENERATE = 1 << 1,
1642 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1643 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1646 struct blk_integrity_iter {
1650 unsigned int data_size;
1651 unsigned short interval;
1652 const char *disk_name;
1655 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1656 typedef void (integrity_prepare_fn) (struct request *);
1657 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1659 struct blk_integrity_profile {
1660 integrity_processing_fn *generate_fn;
1661 integrity_processing_fn *verify_fn;
1662 integrity_prepare_fn *prepare_fn;
1663 integrity_complete_fn *complete_fn;
1667 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1668 extern void blk_integrity_unregister(struct gendisk *);
1669 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1670 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1671 struct scatterlist *);
1672 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1674 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1676 struct blk_integrity *bi = &disk->queue->integrity;
1685 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1687 return blk_get_integrity(bdev->bd_disk);
1691 blk_integrity_queue_supports_integrity(struct request_queue *q)
1693 return q->integrity.profile;
1696 static inline bool blk_integrity_rq(struct request *rq)
1698 return rq->cmd_flags & REQ_INTEGRITY;
1701 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1704 q->limits.max_integrity_segments = segs;
1707 static inline unsigned short
1708 queue_max_integrity_segments(const struct request_queue *q)
1710 return q->limits.max_integrity_segments;
1714 * bio_integrity_intervals - Return number of integrity intervals for a bio
1715 * @bi: blk_integrity profile for device
1716 * @sectors: Size of the bio in 512-byte sectors
1718 * Description: The block layer calculates everything in 512 byte
1719 * sectors but integrity metadata is done in terms of the data integrity
1720 * interval size of the storage device. Convert the block layer sectors
1721 * to the appropriate number of integrity intervals.
1723 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1724 unsigned int sectors)
1726 return sectors >> (bi->interval_exp - 9);
1729 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1730 unsigned int sectors)
1732 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1736 * Return the first bvec that contains integrity data. Only drivers that are
1737 * limited to a single integrity segment should use this helper.
1739 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1741 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1743 return rq->bio->bi_integrity->bip_vec;
1746 #else /* CONFIG_BLK_DEV_INTEGRITY */
1749 struct block_device;
1751 struct blk_integrity;
1753 static inline int blk_integrity_rq(struct request *rq)
1757 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1762 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1764 struct scatterlist *s)
1768 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1772 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1777 blk_integrity_queue_supports_integrity(struct request_queue *q)
1781 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1785 static inline void blk_integrity_register(struct gendisk *d,
1786 struct blk_integrity *b)
1789 static inline void blk_integrity_unregister(struct gendisk *d)
1792 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1796 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1801 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1802 unsigned int sectors)
1807 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1808 unsigned int sectors)
1813 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1818 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1820 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1822 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1824 void blk_ksm_unregister(struct request_queue *q);
1826 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1828 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1829 struct request_queue *q)
1834 static inline void blk_ksm_unregister(struct request_queue *q) { }
1836 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1839 struct block_device_operations {
1840 blk_qc_t (*submit_bio) (struct bio *bio);
1841 int (*open) (struct block_device *, fmode_t);
1842 void (*release) (struct gendisk *, fmode_t);
1843 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1844 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1845 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1846 unsigned int (*check_events) (struct gendisk *disk,
1847 unsigned int clearing);
1848 void (*unlock_native_capacity) (struct gendisk *);
1849 int (*getgeo)(struct block_device *, struct hd_geometry *);
1850 int (*set_read_only)(struct block_device *bdev, bool ro);
1851 /* this callback is with swap_lock and sometimes page table lock held */
1852 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1853 int (*report_zones)(struct gendisk *, sector_t sector,
1854 unsigned int nr_zones, report_zones_cb cb, void *data);
1855 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1856 struct module *owner;
1857 const struct pr_ops *pr_ops;
1860 #ifdef CONFIG_COMPAT
1861 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1862 unsigned int, unsigned long);
1864 #define blkdev_compat_ptr_ioctl NULL
1867 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1868 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1869 struct writeback_control *);
1871 #ifdef CONFIG_BLK_DEV_ZONED
1872 bool blk_req_needs_zone_write_lock(struct request *rq);
1873 bool blk_req_zone_write_trylock(struct request *rq);
1874 void __blk_req_zone_write_lock(struct request *rq);
1875 void __blk_req_zone_write_unlock(struct request *rq);
1877 static inline void blk_req_zone_write_lock(struct request *rq)
1879 if (blk_req_needs_zone_write_lock(rq))
1880 __blk_req_zone_write_lock(rq);
1883 static inline void blk_req_zone_write_unlock(struct request *rq)
1885 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1886 __blk_req_zone_write_unlock(rq);
1889 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1891 return rq->q->seq_zones_wlock &&
1892 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1895 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1897 if (!blk_req_needs_zone_write_lock(rq))
1899 return !blk_req_zone_is_write_locked(rq);
1902 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1907 static inline void blk_req_zone_write_lock(struct request *rq)
1911 static inline void blk_req_zone_write_unlock(struct request *rq)
1914 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1919 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1923 #endif /* CONFIG_BLK_DEV_ZONED */
1925 static inline void blk_wake_io_task(struct task_struct *waiter)
1928 * If we're polling, the task itself is doing the completions. For
1929 * that case, we don't need to signal a wakeup, it's enough to just
1930 * mark us as RUNNING.
1932 if (waiter == current)
1933 __set_current_state(TASK_RUNNING);
1935 wake_up_process(waiter);
1938 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1940 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1941 unsigned long start_time);
1943 unsigned long bio_start_io_acct(struct bio *bio);
1944 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1945 struct block_device *orig_bdev);
1948 * bio_end_io_acct - end I/O accounting for bio based drivers
1949 * @bio: bio to end account for
1950 * @start: start time returned by bio_start_io_acct()
1952 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1954 return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1957 int bdev_read_only(struct block_device *bdev);
1958 int set_blocksize(struct block_device *bdev, int size);
1960 const char *bdevname(struct block_device *bdev, char *buffer);
1961 int lookup_bdev(const char *pathname, dev_t *dev);
1963 void blkdev_show(struct seq_file *seqf, off_t offset);
1965 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1966 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1968 #define BLKDEV_MAJOR_MAX 512
1970 #define BLKDEV_MAJOR_MAX 0
1973 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1975 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1976 int bd_prepare_to_claim(struct block_device *bdev, void *holder);
1977 void bd_abort_claiming(struct block_device *bdev, void *holder);
1978 void blkdev_put(struct block_device *bdev, fmode_t mode);
1980 /* just for blk-cgroup, don't use elsewhere */
1981 struct block_device *blkdev_get_no_open(dev_t dev);
1982 void blkdev_put_no_open(struct block_device *bdev);
1984 struct block_device *bdev_alloc(struct gendisk *disk, u8 partno);
1985 void bdev_add(struct block_device *bdev, dev_t dev);
1986 struct block_device *I_BDEV(struct inode *inode);
1987 struct block_device *bdgrab(struct block_device *bdev);
1988 void bdput(struct block_device *);
1989 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
1993 void invalidate_bdev(struct block_device *bdev);
1994 int sync_blockdev(struct block_device *bdev);
1996 static inline void invalidate_bdev(struct block_device *bdev)
1999 static inline int sync_blockdev(struct block_device *bdev)
2004 int fsync_bdev(struct block_device *bdev);
2006 int freeze_bdev(struct block_device *bdev);
2007 int thaw_bdev(struct block_device *bdev);
2009 #endif /* _LINUX_BLKDEV_H */