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/pagemap.h>
15 #include <linux/backing-dev-defs.h>
16 #include <linux/wait.h>
17 #include <linux/mempool.h>
18 #include <linux/pfn.h>
19 #include <linux/bio.h>
20 #include <linux/stringify.h>
21 #include <linux/gfp.h>
22 #include <linux/bsg.h>
23 #include <linux/smp.h>
24 #include <linux/rcupdate.h>
25 #include <linux/percpu-refcount.h>
26 #include <linux/scatterlist.h>
27 #include <linux/blkzoned.h>
31 struct scsi_ioctl_command;
34 struct elevator_queue;
40 struct blk_flush_queue;
43 struct blk_queue_stats;
44 struct blk_stat_callback;
45 struct blk_keyslot_manager;
47 #define BLKDEV_MIN_RQ 4
48 #define BLKDEV_MAX_RQ 128 /* Default maximum */
50 /* Must be consistent with blk_mq_poll_stats_bkt() */
51 #define BLK_MQ_POLL_STATS_BKTS 16
53 /* Doing classic polling */
54 #define BLK_MQ_POLL_CLASSIC -1
57 * Maximum number of blkcg policies allowed to be registered concurrently.
58 * Defined here to simplify include dependency.
60 #define BLKCG_MAX_POLS 5
62 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
66 typedef __u32 __bitwise req_flags_t;
68 /* drive already may have started this one */
69 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
70 /* may not be passed by ioscheduler */
71 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
72 /* request for flush sequence */
73 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
74 /* merge of different types, fail separately */
75 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
76 /* track inflight for MQ */
77 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
78 /* don't call prep for this one */
79 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
80 /* vaguely specified driver internal error. Ignored by the block layer */
81 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
82 /* don't warn about errors */
83 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
84 /* elevator private data attached */
85 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
86 /* account into disk and partition IO statistics */
87 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
88 /* runtime pm request */
89 #define RQF_PM ((__force req_flags_t)(1 << 15))
90 /* on IO scheduler merge hash */
91 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
92 /* track IO completion time */
93 #define RQF_STATS ((__force req_flags_t)(1 << 17))
94 /* Look at ->special_vec for the actual data payload instead of the
96 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
97 /* The per-zone write lock is held for this request */
98 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
99 /* already slept for hybrid poll */
100 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
101 /* ->timeout has been called, don't expire again */
102 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
104 /* flags that prevent us from merging requests: */
105 #define RQF_NOMERGE_FLAGS \
106 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
109 * Request state for blk-mq.
118 * Try to put the fields that are referenced together in the same cacheline.
120 * If you modify this structure, make sure to update blk_rq_init() and
121 * especially blk_mq_rq_ctx_init() to take care of the added fields.
124 struct request_queue *q;
125 struct blk_mq_ctx *mq_ctx;
126 struct blk_mq_hw_ctx *mq_hctx;
128 unsigned int cmd_flags; /* op and common flags */
129 req_flags_t rq_flags;
134 /* the following two fields are internal, NEVER access directly */
135 unsigned int __data_len; /* total data len */
136 sector_t __sector; /* sector cursor */
141 struct list_head queuelist;
144 * The hash is used inside the scheduler, and killed once the
145 * request reaches the dispatch list. The ipi_list is only used
146 * to queue the request for softirq completion, which is long
147 * after the request has been unhashed (and even removed from
148 * the dispatch list).
151 struct hlist_node hash; /* merge hash */
152 struct llist_node ipi_list;
156 * The rb_node is only used inside the io scheduler, requests
157 * are pruned when moved to the dispatch queue. So let the
158 * completion_data share space with the rb_node.
161 struct rb_node rb_node; /* sort/lookup */
162 struct bio_vec special_vec;
163 void *completion_data;
164 int error_count; /* for legacy drivers, don't use */
168 * Three pointers are available for the IO schedulers, if they need
169 * more they have to dynamically allocate it. Flush requests are
170 * never put on the IO scheduler. So let the flush fields share
171 * space with the elevator data.
181 struct list_head list;
182 rq_end_io_fn *saved_end_io;
186 struct gendisk *rq_disk;
187 struct block_device *part;
188 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
189 /* Time that the first bio started allocating this request. */
192 /* Time that this request was allocated for this IO. */
194 /* Time that I/O was submitted to the device. */
195 u64 io_start_time_ns;
197 #ifdef CONFIG_BLK_WBT
198 unsigned short wbt_flags;
201 * rq sectors used for blk stats. It has the same value
202 * with blk_rq_sectors(rq), except that it never be zeroed
205 unsigned short stats_sectors;
208 * Number of scatter-gather DMA addr+len pairs after
209 * physical address coalescing is performed.
211 unsigned short nr_phys_segments;
213 #if defined(CONFIG_BLK_DEV_INTEGRITY)
214 unsigned short nr_integrity_segments;
217 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
218 struct bio_crypt_ctx *crypt_ctx;
219 struct blk_ksm_keyslot *crypt_keyslot;
222 unsigned short write_hint;
223 unsigned short ioprio;
225 enum mq_rq_state state;
228 unsigned int timeout;
229 unsigned long deadline;
232 struct __call_single_data csd;
237 * completion callback.
239 rq_end_io_fn *end_io;
243 static inline bool blk_op_is_scsi(unsigned int op)
245 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
248 static inline bool blk_op_is_private(unsigned int op)
250 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
253 static inline bool blk_rq_is_scsi(struct request *rq)
255 return blk_op_is_scsi(req_op(rq));
258 static inline bool blk_rq_is_private(struct request *rq)
260 return blk_op_is_private(req_op(rq));
263 static inline bool blk_rq_is_passthrough(struct request *rq)
265 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
268 static inline bool bio_is_passthrough(struct bio *bio)
270 unsigned op = bio_op(bio);
272 return blk_op_is_scsi(op) || blk_op_is_private(op);
275 static inline bool blk_op_is_passthrough(unsigned int op)
277 return (blk_op_is_scsi(op & REQ_OP_MASK) ||
278 blk_op_is_private(op & REQ_OP_MASK));
281 static inline unsigned short req_get_ioprio(struct request *req)
286 #include <linux/elevator.h>
288 struct blk_queue_ctx;
292 enum blk_eh_timer_return {
293 BLK_EH_DONE, /* drivers has completed the command */
294 BLK_EH_RESET_TIMER, /* reset timer and try again */
297 enum blk_queue_state {
302 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
303 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
305 #define BLK_SCSI_MAX_CMDS (256)
306 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
309 * Zoned block device models (zoned limit).
311 * Note: This needs to be ordered from the least to the most severe
312 * restrictions for the inheritance in blk_stack_limits() to work.
314 enum blk_zoned_model {
315 BLK_ZONED_NONE = 0, /* Regular block device */
316 BLK_ZONED_HA, /* Host-aware zoned block device */
317 BLK_ZONED_HM, /* Host-managed zoned block device */
321 * BLK_BOUNCE_NONE: never bounce (default)
322 * BLK_BOUNCE_HIGH: bounce all highmem pages
329 struct queue_limits {
330 enum blk_bounce bounce;
331 unsigned long seg_boundary_mask;
332 unsigned long virt_boundary_mask;
334 unsigned int max_hw_sectors;
335 unsigned int max_dev_sectors;
336 unsigned int chunk_sectors;
337 unsigned int max_sectors;
338 unsigned int max_segment_size;
339 unsigned int physical_block_size;
340 unsigned int logical_block_size;
341 unsigned int alignment_offset;
344 unsigned int max_discard_sectors;
345 unsigned int max_hw_discard_sectors;
346 unsigned int max_write_same_sectors;
347 unsigned int max_write_zeroes_sectors;
348 unsigned int max_zone_append_sectors;
349 unsigned int discard_granularity;
350 unsigned int discard_alignment;
351 unsigned int zone_write_granularity;
353 unsigned short max_segments;
354 unsigned short max_integrity_segments;
355 unsigned short max_discard_segments;
357 unsigned char misaligned;
358 unsigned char discard_misaligned;
359 unsigned char raid_partial_stripes_expensive;
360 enum blk_zoned_model zoned;
363 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
366 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
368 #ifdef CONFIG_BLK_DEV_ZONED
370 #define BLK_ALL_ZONES ((unsigned int)-1)
371 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
372 unsigned int nr_zones, report_zones_cb cb, void *data);
373 unsigned int blkdev_nr_zones(struct gendisk *disk);
374 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
375 sector_t sectors, sector_t nr_sectors,
377 int blk_revalidate_disk_zones(struct gendisk *disk,
378 void (*update_driver_data)(struct gendisk *disk));
380 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
381 unsigned int cmd, unsigned long arg);
382 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
383 unsigned int cmd, unsigned long arg);
385 #else /* CONFIG_BLK_DEV_ZONED */
387 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
392 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
393 fmode_t mode, unsigned int cmd,
399 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
400 fmode_t mode, unsigned int cmd,
406 #endif /* CONFIG_BLK_DEV_ZONED */
408 struct request_queue {
409 struct request *last_merge;
410 struct elevator_queue *elevator;
412 struct percpu_ref q_usage_counter;
414 struct blk_queue_stats *stats;
415 struct rq_qos *rq_qos;
417 const struct blk_mq_ops *mq_ops;
420 struct blk_mq_ctx __percpu *queue_ctx;
422 unsigned int queue_depth;
424 /* hw dispatch queues */
425 struct blk_mq_hw_ctx **queue_hw_ctx;
426 unsigned int nr_hw_queues;
428 struct backing_dev_info *backing_dev_info;
431 * The queue owner gets to use this for whatever they like.
432 * ll_rw_blk doesn't touch it.
437 * various queue flags, see QUEUE_* below
439 unsigned long queue_flags;
441 * Number of contexts that have called blk_set_pm_only(). If this
442 * counter is above zero then only RQF_PM requests are processed.
447 * ida allocated id for this queue. Used to index queues from
452 spinlock_t queue_lock;
462 struct kobject *mq_kobj;
464 #ifdef CONFIG_BLK_DEV_INTEGRITY
465 struct blk_integrity integrity;
466 #endif /* CONFIG_BLK_DEV_INTEGRITY */
470 enum rpm_status rpm_status;
476 unsigned long nr_requests; /* Max # of requests */
478 unsigned int dma_pad_mask;
479 unsigned int dma_alignment;
481 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
482 /* Inline crypto capabilities */
483 struct blk_keyslot_manager *ksm;
486 unsigned int rq_timeout;
489 struct blk_stat_callback *poll_cb;
490 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
492 struct timer_list timeout;
493 struct work_struct timeout_work;
495 atomic_t nr_active_requests_shared_sbitmap;
497 struct list_head icq_list;
498 #ifdef CONFIG_BLK_CGROUP
499 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
500 struct blkcg_gq *root_blkg;
501 struct list_head blkg_list;
504 struct queue_limits limits;
506 unsigned int required_elevator_features;
508 #ifdef CONFIG_BLK_DEV_ZONED
510 * Zoned block device information for request dispatch control.
511 * nr_zones is the total number of zones of the device. This is always
512 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
513 * bits which indicates if a zone is conventional (bit set) or
514 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
515 * bits which indicates if a zone is write locked, that is, if a write
516 * request targeting the zone was dispatched. All three fields are
517 * initialized by the low level device driver (e.g. scsi/sd.c).
518 * Stacking drivers (device mappers) may or may not initialize
521 * Reads of this information must be protected with blk_queue_enter() /
522 * blk_queue_exit(). Modifying this information is only allowed while
523 * no requests are being processed. See also blk_mq_freeze_queue() and
524 * blk_mq_unfreeze_queue().
526 unsigned int nr_zones;
527 unsigned long *conv_zones_bitmap;
528 unsigned long *seq_zones_wlock;
529 unsigned int max_open_zones;
530 unsigned int max_active_zones;
531 #endif /* CONFIG_BLK_DEV_ZONED */
536 unsigned int sg_timeout;
537 unsigned int sg_reserved_size;
539 struct mutex debugfs_mutex;
540 #ifdef CONFIG_BLK_DEV_IO_TRACE
541 struct blk_trace __rcu *blk_trace;
544 * for flush operations
546 struct blk_flush_queue *fq;
548 struct list_head requeue_list;
549 spinlock_t requeue_lock;
550 struct delayed_work requeue_work;
552 struct mutex sysfs_lock;
553 struct mutex sysfs_dir_lock;
556 * for reusing dead hctx instance in case of updating
559 struct list_head unused_hctx_list;
560 spinlock_t unused_hctx_lock;
564 #if defined(CONFIG_BLK_DEV_BSG)
565 struct bsg_class_device bsg_dev;
568 #ifdef CONFIG_BLK_DEV_THROTTLING
570 struct throtl_data *td;
572 struct rcu_head rcu_head;
573 wait_queue_head_t mq_freeze_wq;
575 * Protect concurrent access to q_usage_counter by
576 * percpu_ref_kill() and percpu_ref_reinit().
578 struct mutex mq_freeze_lock;
580 struct blk_mq_tag_set *tag_set;
581 struct list_head tag_set_list;
582 struct bio_set bio_split;
584 struct dentry *debugfs_dir;
586 #ifdef CONFIG_BLK_DEBUG_FS
587 struct dentry *sched_debugfs_dir;
588 struct dentry *rqos_debugfs_dir;
591 bool mq_sysfs_init_done;
595 #define BLK_MAX_WRITE_HINTS 5
596 u64 write_hints[BLK_MAX_WRITE_HINTS];
599 /* Keep blk_queue_flag_name[] in sync with the definitions below */
600 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
601 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
602 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
603 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
604 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
605 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
606 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
607 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
608 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
609 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
610 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
611 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
612 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
613 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
614 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
615 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
616 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
617 #define QUEUE_FLAG_WC 17 /* Write back caching */
618 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
619 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
620 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
621 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
622 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
623 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
624 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
625 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
626 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
627 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
628 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
629 #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */
631 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
632 (1 << QUEUE_FLAG_SAME_COMP) | \
633 (1 << QUEUE_FLAG_NOWAIT))
635 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
636 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
637 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
639 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
640 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
641 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
642 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
643 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
644 #define blk_queue_noxmerges(q) \
645 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
646 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
647 #define blk_queue_stable_writes(q) \
648 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
649 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
650 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
651 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
652 #define blk_queue_zone_resetall(q) \
653 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
654 #define blk_queue_secure_erase(q) \
655 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
656 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
657 #define blk_queue_scsi_passthrough(q) \
658 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
659 #define blk_queue_pci_p2pdma(q) \
660 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
661 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
662 #define blk_queue_rq_alloc_time(q) \
663 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
665 #define blk_queue_rq_alloc_time(q) false
668 #define blk_noretry_request(rq) \
669 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
670 REQ_FAILFAST_DRIVER))
671 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
672 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
673 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
674 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
675 #define blk_queue_nowait(q) test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
677 extern void blk_set_pm_only(struct request_queue *q);
678 extern void blk_clear_pm_only(struct request_queue *q);
680 static inline bool blk_account_rq(struct request *rq)
682 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
685 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
687 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
689 #define rq_dma_dir(rq) \
690 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
692 #define dma_map_bvec(dev, bv, dir, attrs) \
693 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
696 #define queue_to_disk(q) (dev_to_disk(kobj_to_dev((q)->kobj.parent)))
698 static inline bool queue_is_mq(struct request_queue *q)
704 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
706 return q->rpm_status;
709 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
715 static inline enum blk_zoned_model
716 blk_queue_zoned_model(struct request_queue *q)
718 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
719 return q->limits.zoned;
720 return BLK_ZONED_NONE;
723 static inline bool blk_queue_is_zoned(struct request_queue *q)
725 switch (blk_queue_zoned_model(q)) {
734 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
736 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
739 #ifdef CONFIG_BLK_DEV_ZONED
740 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
742 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
745 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
748 if (!blk_queue_is_zoned(q))
750 return sector >> ilog2(q->limits.chunk_sectors);
753 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
756 if (!blk_queue_is_zoned(q))
758 if (!q->conv_zones_bitmap)
760 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
763 static inline void blk_queue_max_open_zones(struct request_queue *q,
764 unsigned int max_open_zones)
766 q->max_open_zones = max_open_zones;
769 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
771 return q->max_open_zones;
774 static inline void blk_queue_max_active_zones(struct request_queue *q,
775 unsigned int max_active_zones)
777 q->max_active_zones = max_active_zones;
780 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
782 return q->max_active_zones;
784 #else /* CONFIG_BLK_DEV_ZONED */
785 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
789 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
794 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
799 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
803 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
807 #endif /* CONFIG_BLK_DEV_ZONED */
809 static inline bool rq_is_sync(struct request *rq)
811 return op_is_sync(rq->cmd_flags);
814 static inline bool rq_mergeable(struct request *rq)
816 if (blk_rq_is_passthrough(rq))
819 if (req_op(rq) == REQ_OP_FLUSH)
822 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
825 if (req_op(rq) == REQ_OP_ZONE_APPEND)
828 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
830 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
836 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
838 if (bio_page(a) == bio_page(b) &&
839 bio_offset(a) == bio_offset(b))
845 static inline unsigned int blk_queue_depth(struct request_queue *q)
848 return q->queue_depth;
850 return q->nr_requests;
854 * default timeout for SG_IO if none specified
856 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
857 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
863 unsigned long offset;
868 struct req_iterator {
869 struct bvec_iter iter;
873 /* This should not be used directly - use rq_for_each_segment */
874 #define for_each_bio(_bio) \
875 for (; _bio; _bio = _bio->bi_next)
876 #define __rq_for_each_bio(_bio, rq) \
878 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
880 #define rq_for_each_segment(bvl, _rq, _iter) \
881 __rq_for_each_bio(_iter.bio, _rq) \
882 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
884 #define rq_for_each_bvec(bvl, _rq, _iter) \
885 __rq_for_each_bio(_iter.bio, _rq) \
886 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
888 #define rq_iter_last(bvec, _iter) \
889 (_iter.bio->bi_next == NULL && \
890 bio_iter_last(bvec, _iter.iter))
892 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
893 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
895 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
896 extern void rq_flush_dcache_pages(struct request *rq);
898 static inline void rq_flush_dcache_pages(struct request *rq)
903 extern int blk_register_queue(struct gendisk *disk);
904 extern void blk_unregister_queue(struct gendisk *disk);
905 blk_qc_t submit_bio_noacct(struct bio *bio);
906 extern void blk_rq_init(struct request_queue *q, struct request *rq);
907 extern void blk_put_request(struct request *);
908 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
909 blk_mq_req_flags_t flags);
910 extern int blk_lld_busy(struct request_queue *q);
911 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
912 struct bio_set *bs, gfp_t gfp_mask,
913 int (*bio_ctr)(struct bio *, struct bio *, void *),
915 extern void blk_rq_unprep_clone(struct request *rq);
916 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
918 int blk_rq_append_bio(struct request *rq, struct bio *bio);
919 extern void blk_queue_split(struct bio **);
920 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
921 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
922 unsigned int, void __user *);
923 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
924 unsigned int, void __user *);
925 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
926 struct scsi_ioctl_command __user *);
927 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
928 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
930 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
931 extern void blk_queue_exit(struct request_queue *q);
932 extern void blk_sync_queue(struct request_queue *q);
933 extern int blk_rq_map_user(struct request_queue *, struct request *,
934 struct rq_map_data *, void __user *, unsigned long,
936 extern int blk_rq_unmap_user(struct bio *);
937 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
938 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
939 struct rq_map_data *, const struct iov_iter *,
941 extern void blk_execute_rq(struct gendisk *, struct request *, int);
942 extern void blk_execute_rq_nowait(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,
1074 unsigned int chunk_sectors)
1076 if (!chunk_sectors) {
1077 if (q->limits.chunk_sectors)
1078 chunk_sectors = q->limits.chunk_sectors;
1080 return q->limits.max_sectors;
1083 if (likely(is_power_of_2(chunk_sectors)))
1084 chunk_sectors -= offset & (chunk_sectors - 1);
1086 chunk_sectors -= sector_div(offset, chunk_sectors);
1088 return min(q->limits.max_sectors, chunk_sectors);
1091 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1094 struct request_queue *q = rq->q;
1096 if (blk_rq_is_passthrough(rq))
1097 return q->limits.max_hw_sectors;
1099 if (!q->limits.chunk_sectors ||
1100 req_op(rq) == REQ_OP_DISCARD ||
1101 req_op(rq) == REQ_OP_SECURE_ERASE)
1102 return blk_queue_get_max_sectors(q, req_op(rq));
1104 return min(blk_max_size_offset(q, offset, 0),
1105 blk_queue_get_max_sectors(q, req_op(rq)));
1108 static inline unsigned int blk_rq_count_bios(struct request *rq)
1110 unsigned int nr_bios = 0;
1113 __rq_for_each_bio(bio, rq)
1119 void blk_steal_bios(struct bio_list *list, struct request *rq);
1122 * Request completion related functions.
1124 * blk_update_request() completes given number of bytes and updates
1125 * the request without completing it.
1127 extern bool blk_update_request(struct request *rq, blk_status_t error,
1128 unsigned int nr_bytes);
1130 extern void blk_abort_request(struct request *);
1133 * Access functions for manipulating queue properties
1135 extern void blk_cleanup_queue(struct request_queue *);
1136 void blk_queue_bounce_limit(struct request_queue *q, enum blk_bounce limit);
1137 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1138 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1139 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1140 extern void blk_queue_max_discard_segments(struct request_queue *,
1142 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1143 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1144 unsigned int max_discard_sectors);
1145 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1146 unsigned int max_write_same_sectors);
1147 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1148 unsigned int max_write_same_sectors);
1149 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1150 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1151 unsigned int max_zone_append_sectors);
1152 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1153 void blk_queue_zone_write_granularity(struct request_queue *q,
1155 extern void blk_queue_alignment_offset(struct request_queue *q,
1156 unsigned int alignment);
1157 void blk_queue_update_readahead(struct request_queue *q);
1158 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1159 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1160 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1161 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1162 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1163 extern void blk_set_default_limits(struct queue_limits *lim);
1164 extern void blk_set_stacking_limits(struct queue_limits *lim);
1165 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1167 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1169 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1170 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1171 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1172 extern void blk_queue_dma_alignment(struct request_queue *, int);
1173 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1174 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1175 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1176 extern void blk_queue_required_elevator_features(struct request_queue *q,
1177 unsigned int features);
1178 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1179 struct device *dev);
1182 * Number of physical segments as sent to the device.
1184 * Normally this is the number of discontiguous data segments sent by the
1185 * submitter. But for data-less command like discard we might have no
1186 * actual data segments submitted, but the driver might have to add it's
1187 * own special payload. In that case we still return 1 here so that this
1188 * special payload will be mapped.
1190 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1192 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1194 return rq->nr_phys_segments;
1198 * Number of discard segments (or ranges) the driver needs to fill in.
1199 * Each discard bio merged into a request is counted as one segment.
1201 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1203 return max_t(unsigned short, rq->nr_phys_segments, 1);
1206 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1207 struct scatterlist *sglist, struct scatterlist **last_sg);
1208 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1209 struct scatterlist *sglist)
1211 struct scatterlist *last_sg = NULL;
1213 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1215 extern void blk_dump_rq_flags(struct request *, char *);
1217 bool __must_check blk_get_queue(struct request_queue *);
1218 struct request_queue *blk_alloc_queue(int node_id);
1219 extern void blk_put_queue(struct request_queue *);
1220 extern void blk_set_queue_dying(struct request_queue *);
1224 * blk_plug permits building a queue of related requests by holding the I/O
1225 * fragments for a short period. This allows merging of sequential requests
1226 * into single larger request. As the requests are moved from a per-task list to
1227 * the device's request_queue in a batch, this results in improved scalability
1228 * as the lock contention for request_queue lock is reduced.
1230 * It is ok not to disable preemption when adding the request to the plug list
1231 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1232 * the plug list when the task sleeps by itself. For details, please see
1233 * schedule() where blk_schedule_flush_plug() is called.
1236 struct list_head mq_list; /* blk-mq requests */
1237 struct list_head cb_list; /* md requires an unplug callback */
1238 unsigned short rq_count;
1239 bool multiple_queues;
1242 #define BLK_MAX_REQUEST_COUNT 16
1243 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1246 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1247 struct blk_plug_cb {
1248 struct list_head list;
1249 blk_plug_cb_fn callback;
1252 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1253 void *data, int size);
1254 extern void blk_start_plug(struct blk_plug *);
1255 extern void blk_finish_plug(struct blk_plug *);
1256 extern void blk_flush_plug_list(struct blk_plug *, bool);
1258 static inline void blk_flush_plug(struct task_struct *tsk)
1260 struct blk_plug *plug = tsk->plug;
1263 blk_flush_plug_list(plug, false);
1266 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1268 struct blk_plug *plug = tsk->plug;
1271 blk_flush_plug_list(plug, true);
1274 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1276 struct blk_plug *plug = tsk->plug;
1279 (!list_empty(&plug->mq_list) ||
1280 !list_empty(&plug->cb_list));
1283 int blkdev_issue_flush(struct block_device *bdev);
1284 long nr_blockdev_pages(void);
1285 #else /* CONFIG_BLOCK */
1289 static inline void blk_start_plug(struct blk_plug *plug)
1293 static inline void blk_finish_plug(struct blk_plug *plug)
1297 static inline void blk_flush_plug(struct task_struct *task)
1301 static inline void blk_schedule_flush_plug(struct task_struct *task)
1306 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1311 static inline int blkdev_issue_flush(struct block_device *bdev)
1316 static inline long nr_blockdev_pages(void)
1320 #endif /* CONFIG_BLOCK */
1322 extern void blk_io_schedule(void);
1324 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1325 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1327 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1329 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1330 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1331 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1332 sector_t nr_sects, gfp_t gfp_mask, int flags,
1335 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1336 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1338 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1339 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1341 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1342 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1344 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1345 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1347 return blkdev_issue_discard(sb->s_bdev,
1348 block << (sb->s_blocksize_bits -
1350 nr_blocks << (sb->s_blocksize_bits -
1354 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1355 sector_t nr_blocks, gfp_t gfp_mask)
1357 return blkdev_issue_zeroout(sb->s_bdev,
1358 block << (sb->s_blocksize_bits -
1360 nr_blocks << (sb->s_blocksize_bits -
1365 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1367 static inline bool bdev_is_partition(struct block_device *bdev)
1369 return bdev->bd_partno;
1372 enum blk_default_limits {
1373 BLK_MAX_SEGMENTS = 128,
1374 BLK_SAFE_MAX_SECTORS = 255,
1375 BLK_DEF_MAX_SECTORS = 2560,
1376 BLK_MAX_SEGMENT_SIZE = 65536,
1377 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1380 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1382 return q->limits.seg_boundary_mask;
1385 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1387 return q->limits.virt_boundary_mask;
1390 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1392 return q->limits.max_sectors;
1395 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1397 return q->limits.max_hw_sectors;
1400 static inline unsigned short queue_max_segments(const struct request_queue *q)
1402 return q->limits.max_segments;
1405 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1407 return q->limits.max_discard_segments;
1410 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1412 return q->limits.max_segment_size;
1415 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1418 const struct queue_limits *l = &q->limits;
1420 return min(l->max_zone_append_sectors, l->max_sectors);
1423 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1427 if (q && q->limits.logical_block_size)
1428 retval = q->limits.logical_block_size;
1433 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1435 return queue_logical_block_size(bdev_get_queue(bdev));
1438 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1440 return q->limits.physical_block_size;
1443 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1445 return queue_physical_block_size(bdev_get_queue(bdev));
1448 static inline unsigned int queue_io_min(const struct request_queue *q)
1450 return q->limits.io_min;
1453 static inline int bdev_io_min(struct block_device *bdev)
1455 return queue_io_min(bdev_get_queue(bdev));
1458 static inline unsigned int queue_io_opt(const struct request_queue *q)
1460 return q->limits.io_opt;
1463 static inline int bdev_io_opt(struct block_device *bdev)
1465 return queue_io_opt(bdev_get_queue(bdev));
1468 static inline unsigned int
1469 queue_zone_write_granularity(const struct request_queue *q)
1471 return q->limits.zone_write_granularity;
1474 static inline unsigned int
1475 bdev_zone_write_granularity(struct block_device *bdev)
1477 return queue_zone_write_granularity(bdev_get_queue(bdev));
1480 static inline int queue_alignment_offset(const struct request_queue *q)
1482 if (q->limits.misaligned)
1485 return q->limits.alignment_offset;
1488 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1490 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1491 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1494 return (granularity + lim->alignment_offset - alignment) % granularity;
1497 static inline int bdev_alignment_offset(struct block_device *bdev)
1499 struct request_queue *q = bdev_get_queue(bdev);
1501 if (q->limits.misaligned)
1503 if (bdev_is_partition(bdev))
1504 return queue_limit_alignment_offset(&q->limits,
1505 bdev->bd_start_sect);
1506 return q->limits.alignment_offset;
1509 static inline int queue_discard_alignment(const struct request_queue *q)
1511 if (q->limits.discard_misaligned)
1514 return q->limits.discard_alignment;
1517 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1519 unsigned int alignment, granularity, offset;
1521 if (!lim->max_discard_sectors)
1524 /* Why are these in bytes, not sectors? */
1525 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1526 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1530 /* Offset of the partition start in 'granularity' sectors */
1531 offset = sector_div(sector, granularity);
1533 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1534 offset = (granularity + alignment - offset) % granularity;
1536 /* Turn it back into bytes, gaah */
1537 return offset << SECTOR_SHIFT;
1540 static inline int bdev_discard_alignment(struct block_device *bdev)
1542 struct request_queue *q = bdev_get_queue(bdev);
1544 if (bdev_is_partition(bdev))
1545 return queue_limit_discard_alignment(&q->limits,
1546 bdev->bd_start_sect);
1547 return q->limits.discard_alignment;
1550 static inline unsigned int bdev_write_same(struct block_device *bdev)
1552 struct request_queue *q = bdev_get_queue(bdev);
1555 return q->limits.max_write_same_sectors;
1560 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1562 struct request_queue *q = bdev_get_queue(bdev);
1565 return q->limits.max_write_zeroes_sectors;
1570 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1572 struct request_queue *q = bdev_get_queue(bdev);
1575 return blk_queue_zoned_model(q);
1577 return BLK_ZONED_NONE;
1580 static inline bool bdev_is_zoned(struct block_device *bdev)
1582 struct request_queue *q = bdev_get_queue(bdev);
1585 return blk_queue_is_zoned(q);
1590 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1592 struct request_queue *q = bdev_get_queue(bdev);
1595 return blk_queue_zone_sectors(q);
1599 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1601 struct request_queue *q = bdev_get_queue(bdev);
1604 return queue_max_open_zones(q);
1608 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1610 struct request_queue *q = bdev_get_queue(bdev);
1613 return queue_max_active_zones(q);
1617 static inline int queue_dma_alignment(const struct request_queue *q)
1619 return q ? q->dma_alignment : 511;
1622 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1625 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1626 return !(addr & alignment) && !(len & alignment);
1629 /* assumes size > 256 */
1630 static inline unsigned int blksize_bits(unsigned int size)
1632 unsigned int bits = 8;
1636 } while (size > 256);
1640 static inline unsigned int block_size(struct block_device *bdev)
1642 return 1 << bdev->bd_inode->i_blkbits;
1645 int kblockd_schedule_work(struct work_struct *work);
1646 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1648 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1649 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1650 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1651 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1653 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1655 enum blk_integrity_flags {
1656 BLK_INTEGRITY_VERIFY = 1 << 0,
1657 BLK_INTEGRITY_GENERATE = 1 << 1,
1658 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1659 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1662 struct blk_integrity_iter {
1666 unsigned int data_size;
1667 unsigned short interval;
1668 const char *disk_name;
1671 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1672 typedef void (integrity_prepare_fn) (struct request *);
1673 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1675 struct blk_integrity_profile {
1676 integrity_processing_fn *generate_fn;
1677 integrity_processing_fn *verify_fn;
1678 integrity_prepare_fn *prepare_fn;
1679 integrity_complete_fn *complete_fn;
1683 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1684 extern void blk_integrity_unregister(struct gendisk *);
1685 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1686 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1687 struct scatterlist *);
1688 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1690 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1692 struct blk_integrity *bi = &disk->queue->integrity;
1701 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1703 return blk_get_integrity(bdev->bd_disk);
1707 blk_integrity_queue_supports_integrity(struct request_queue *q)
1709 return q->integrity.profile;
1712 static inline bool blk_integrity_rq(struct request *rq)
1714 return rq->cmd_flags & REQ_INTEGRITY;
1717 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1720 q->limits.max_integrity_segments = segs;
1723 static inline unsigned short
1724 queue_max_integrity_segments(const struct request_queue *q)
1726 return q->limits.max_integrity_segments;
1730 * bio_integrity_intervals - Return number of integrity intervals for a bio
1731 * @bi: blk_integrity profile for device
1732 * @sectors: Size of the bio in 512-byte sectors
1734 * Description: The block layer calculates everything in 512 byte
1735 * sectors but integrity metadata is done in terms of the data integrity
1736 * interval size of the storage device. Convert the block layer sectors
1737 * to the appropriate number of integrity intervals.
1739 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1740 unsigned int sectors)
1742 return sectors >> (bi->interval_exp - 9);
1745 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1746 unsigned int sectors)
1748 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1752 * Return the first bvec that contains integrity data. Only drivers that are
1753 * limited to a single integrity segment should use this helper.
1755 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1757 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1759 return rq->bio->bi_integrity->bip_vec;
1762 #else /* CONFIG_BLK_DEV_INTEGRITY */
1765 struct block_device;
1767 struct blk_integrity;
1769 static inline int blk_integrity_rq(struct request *rq)
1773 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1778 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1780 struct scatterlist *s)
1784 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1788 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1793 blk_integrity_queue_supports_integrity(struct request_queue *q)
1797 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1801 static inline void blk_integrity_register(struct gendisk *d,
1802 struct blk_integrity *b)
1805 static inline void blk_integrity_unregister(struct gendisk *d)
1808 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1812 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1817 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1818 unsigned int sectors)
1823 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1824 unsigned int sectors)
1829 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1834 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1836 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1838 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1840 void blk_ksm_unregister(struct request_queue *q);
1842 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1844 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1845 struct request_queue *q)
1850 static inline void blk_ksm_unregister(struct request_queue *q) { }
1852 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1855 struct block_device_operations {
1856 blk_qc_t (*submit_bio) (struct bio *bio);
1857 int (*open) (struct block_device *, fmode_t);
1858 void (*release) (struct gendisk *, fmode_t);
1859 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1860 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1861 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1862 unsigned int (*check_events) (struct gendisk *disk,
1863 unsigned int clearing);
1864 void (*unlock_native_capacity) (struct gendisk *);
1865 int (*getgeo)(struct block_device *, struct hd_geometry *);
1866 int (*set_read_only)(struct block_device *bdev, bool ro);
1867 /* this callback is with swap_lock and sometimes page table lock held */
1868 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1869 int (*report_zones)(struct gendisk *, sector_t sector,
1870 unsigned int nr_zones, report_zones_cb cb, void *data);
1871 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1872 struct module *owner;
1873 const struct pr_ops *pr_ops;
1876 #ifdef CONFIG_COMPAT
1877 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1878 unsigned int, unsigned long);
1880 #define blkdev_compat_ptr_ioctl NULL
1883 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1884 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1885 struct writeback_control *);
1887 #ifdef CONFIG_BLK_DEV_ZONED
1888 bool blk_req_needs_zone_write_lock(struct request *rq);
1889 bool blk_req_zone_write_trylock(struct request *rq);
1890 void __blk_req_zone_write_lock(struct request *rq);
1891 void __blk_req_zone_write_unlock(struct request *rq);
1893 static inline void blk_req_zone_write_lock(struct request *rq)
1895 if (blk_req_needs_zone_write_lock(rq))
1896 __blk_req_zone_write_lock(rq);
1899 static inline void blk_req_zone_write_unlock(struct request *rq)
1901 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1902 __blk_req_zone_write_unlock(rq);
1905 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1907 return rq->q->seq_zones_wlock &&
1908 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1911 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1913 if (!blk_req_needs_zone_write_lock(rq))
1915 return !blk_req_zone_is_write_locked(rq);
1918 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1923 static inline void blk_req_zone_write_lock(struct request *rq)
1927 static inline void blk_req_zone_write_unlock(struct request *rq)
1930 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1935 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1939 #endif /* CONFIG_BLK_DEV_ZONED */
1941 static inline void blk_wake_io_task(struct task_struct *waiter)
1944 * If we're polling, the task itself is doing the completions. For
1945 * that case, we don't need to signal a wakeup, it's enough to just
1946 * mark us as RUNNING.
1948 if (waiter == current)
1949 __set_current_state(TASK_RUNNING);
1951 wake_up_process(waiter);
1954 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1956 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1957 unsigned long start_time);
1959 unsigned long bio_start_io_acct(struct bio *bio);
1960 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1961 struct block_device *orig_bdev);
1964 * bio_end_io_acct - end I/O accounting for bio based drivers
1965 * @bio: bio to end account for
1966 * @start: start time returned by bio_start_io_acct()
1968 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1970 return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1973 int bdev_read_only(struct block_device *bdev);
1974 int set_blocksize(struct block_device *bdev, int size);
1976 const char *bdevname(struct block_device *bdev, char *buffer);
1977 int lookup_bdev(const char *pathname, dev_t *dev);
1979 void blkdev_show(struct seq_file *seqf, off_t offset);
1981 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1982 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1984 #define BLKDEV_MAJOR_MAX 512
1986 #define BLKDEV_MAJOR_MAX 0
1989 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1991 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1992 int bd_prepare_to_claim(struct block_device *bdev, void *holder);
1993 void bd_abort_claiming(struct block_device *bdev, void *holder);
1994 void blkdev_put(struct block_device *bdev, fmode_t mode);
1996 /* just for blk-cgroup, don't use elsewhere */
1997 struct block_device *blkdev_get_no_open(dev_t dev);
1998 void blkdev_put_no_open(struct block_device *bdev);
2000 struct block_device *bdev_alloc(struct gendisk *disk, u8 partno);
2001 void bdev_add(struct block_device *bdev, dev_t dev);
2002 struct block_device *I_BDEV(struct inode *inode);
2003 struct block_device *bdgrab(struct block_device *bdev);
2004 void bdput(struct block_device *);
2005 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
2009 void invalidate_bdev(struct block_device *bdev);
2010 int sync_blockdev(struct block_device *bdev);
2012 static inline void invalidate_bdev(struct block_device *bdev)
2015 static inline int sync_blockdev(struct block_device *bdev)
2020 int fsync_bdev(struct block_device *bdev);
2022 int freeze_bdev(struct block_device *bdev);
2023 int thaw_bdev(struct block_device *bdev);
2025 #endif /* _LINUX_BLKDEV_H */