1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef _LINUX_BLKDEV_H
3 #define _LINUX_BLKDEV_H
5 #include <linux/sched.h>
6 #include <linux/sched/clock.h>
7 #include <linux/major.h>
8 #include <linux/genhd.h>
9 #include <linux/list.h>
10 #include <linux/llist.h>
11 #include <linux/timer.h>
12 #include <linux/workqueue.h>
13 #include <linux/pagemap.h>
14 #include <linux/backing-dev-defs.h>
15 #include <linux/wait.h>
16 #include <linux/mempool.h>
17 #include <linux/pfn.h>
18 #include <linux/bio.h>
19 #include <linux/stringify.h>
20 #include <linux/gfp.h>
21 #include <linux/bsg.h>
22 #include <linux/smp.h>
23 #include <linux/rcupdate.h>
24 #include <linux/percpu-refcount.h>
25 #include <linux/scatterlist.h>
26 #include <linux/blkzoned.h>
29 struct scsi_ioctl_command;
32 struct elevator_queue;
38 struct blk_flush_queue;
41 struct blk_queue_stats;
42 struct blk_stat_callback;
43 struct blk_keyslot_manager;
45 #define BLKDEV_MIN_RQ 4
46 #define BLKDEV_MAX_RQ 128 /* Default maximum */
48 /* Must be consistent with blk_mq_poll_stats_bkt() */
49 #define BLK_MQ_POLL_STATS_BKTS 16
51 /* Doing classic polling */
52 #define BLK_MQ_POLL_CLASSIC -1
55 * Maximum number of blkcg policies allowed to be registered concurrently.
56 * Defined here to simplify include dependency.
58 #define BLKCG_MAX_POLS 5
60 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
64 typedef __u32 __bitwise req_flags_t;
66 /* elevator knows about this request */
67 #define RQF_SORTED ((__force req_flags_t)(1 << 0))
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 /* set for "ide_preempt" requests and also for requests for which the SCSI
81 "quiesce" state must be ignored. */
82 #define RQF_PREEMPT ((__force req_flags_t)(1 << 8))
83 /* vaguely specified driver internal error. Ignored by the block layer */
84 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
85 /* don't warn about errors */
86 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
87 /* elevator private data attached */
88 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
89 /* account into disk and partition IO statistics */
90 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
91 /* request came from our alloc pool */
92 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
93 /* runtime pm request */
94 #define RQF_PM ((__force req_flags_t)(1 << 15))
95 /* on IO scheduler merge hash */
96 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
97 /* track IO completion time */
98 #define RQF_STATS ((__force req_flags_t)(1 << 17))
99 /* Look at ->special_vec for the actual data payload instead of the
101 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
102 /* The per-zone write lock is held for this request */
103 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
104 /* already slept for hybrid poll */
105 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
106 /* ->timeout has been called, don't expire again */
107 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
109 /* flags that prevent us from merging requests: */
110 #define RQF_NOMERGE_FLAGS \
111 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
114 * Request state for blk-mq.
123 * Try to put the fields that are referenced together in the same cacheline.
125 * If you modify this structure, make sure to update blk_rq_init() and
126 * especially blk_mq_rq_ctx_init() to take care of the added fields.
129 struct request_queue *q;
130 struct blk_mq_ctx *mq_ctx;
131 struct blk_mq_hw_ctx *mq_hctx;
133 unsigned int cmd_flags; /* op and common flags */
134 req_flags_t rq_flags;
139 /* the following two fields are internal, NEVER access directly */
140 unsigned int __data_len; /* total data len */
141 sector_t __sector; /* sector cursor */
146 struct list_head queuelist;
149 * The hash is used inside the scheduler, and killed once the
150 * request reaches the dispatch list. The ipi_list is only used
151 * to queue the request for softirq completion, which is long
152 * after the request has been unhashed (and even removed from
153 * the dispatch list).
156 struct hlist_node hash; /* merge hash */
157 struct list_head ipi_list;
161 * The rb_node is only used inside the io scheduler, requests
162 * are pruned when moved to the dispatch queue. So let the
163 * completion_data share space with the rb_node.
166 struct rb_node rb_node; /* sort/lookup */
167 struct bio_vec special_vec;
168 void *completion_data;
169 int error_count; /* for legacy drivers, don't use */
173 * Three pointers are available for the IO schedulers, if they need
174 * more they have to dynamically allocate it. Flush requests are
175 * never put on the IO scheduler. So let the flush fields share
176 * space with the elevator data.
186 struct list_head list;
187 rq_end_io_fn *saved_end_io;
191 struct gendisk *rq_disk;
192 struct hd_struct *part;
193 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
194 /* Time that the first bio started allocating this request. */
197 /* Time that this request was allocated for this IO. */
199 /* Time that I/O was submitted to the device. */
200 u64 io_start_time_ns;
202 #ifdef CONFIG_BLK_WBT
203 unsigned short wbt_flags;
206 * rq sectors used for blk stats. It has the same value
207 * with blk_rq_sectors(rq), except that it never be zeroed
210 unsigned short stats_sectors;
213 * Number of scatter-gather DMA addr+len pairs after
214 * physical address coalescing is performed.
216 unsigned short nr_phys_segments;
218 #if defined(CONFIG_BLK_DEV_INTEGRITY)
219 unsigned short nr_integrity_segments;
222 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
223 struct bio_crypt_ctx *crypt_ctx;
224 struct blk_ksm_keyslot *crypt_keyslot;
227 unsigned short write_hint;
228 unsigned short ioprio;
230 enum mq_rq_state state;
233 unsigned int timeout;
234 unsigned long deadline;
237 struct __call_single_data csd;
242 * completion callback.
244 rq_end_io_fn *end_io;
248 static inline bool blk_op_is_scsi(unsigned int op)
250 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
253 static inline bool blk_op_is_private(unsigned int op)
255 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
258 static inline bool blk_rq_is_scsi(struct request *rq)
260 return blk_op_is_scsi(req_op(rq));
263 static inline bool blk_rq_is_private(struct request *rq)
265 return blk_op_is_private(req_op(rq));
268 static inline bool blk_rq_is_passthrough(struct request *rq)
270 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
273 static inline bool bio_is_passthrough(struct bio *bio)
275 unsigned op = bio_op(bio);
277 return blk_op_is_scsi(op) || blk_op_is_private(op);
280 static inline unsigned short req_get_ioprio(struct request *req)
285 #include <linux/elevator.h>
287 struct blk_queue_ctx;
291 enum blk_eh_timer_return {
292 BLK_EH_DONE, /* drivers has completed the command */
293 BLK_EH_RESET_TIMER, /* reset timer and try again */
296 enum blk_queue_state {
301 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
302 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
304 #define BLK_SCSI_MAX_CMDS (256)
305 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
308 * Zoned block device models (zoned limit).
310 * Note: This needs to be ordered from the least to the most severe
311 * restrictions for the inheritance in blk_stack_limits() to work.
313 enum blk_zoned_model {
314 BLK_ZONED_NONE = 0, /* Regular block device */
315 BLK_ZONED_HA, /* Host-aware zoned block device */
316 BLK_ZONED_HM, /* Host-managed zoned block device */
319 struct queue_limits {
320 unsigned long bounce_pfn;
321 unsigned long seg_boundary_mask;
322 unsigned long virt_boundary_mask;
324 unsigned int max_hw_sectors;
325 unsigned int max_dev_sectors;
326 unsigned int chunk_sectors;
327 unsigned int max_sectors;
328 unsigned int max_segment_size;
329 unsigned int physical_block_size;
330 unsigned int logical_block_size;
331 unsigned int alignment_offset;
334 unsigned int max_discard_sectors;
335 unsigned int max_hw_discard_sectors;
336 unsigned int max_write_same_sectors;
337 unsigned int max_write_zeroes_sectors;
338 unsigned int max_zone_append_sectors;
339 unsigned int discard_granularity;
340 unsigned int discard_alignment;
342 unsigned short max_segments;
343 unsigned short max_integrity_segments;
344 unsigned short max_discard_segments;
346 unsigned char misaligned;
347 unsigned char discard_misaligned;
348 unsigned char raid_partial_stripes_expensive;
349 enum blk_zoned_model zoned;
352 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
355 #ifdef CONFIG_BLK_DEV_ZONED
357 #define BLK_ALL_ZONES ((unsigned int)-1)
358 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
359 unsigned int nr_zones, report_zones_cb cb, void *data);
360 unsigned int blkdev_nr_zones(struct gendisk *disk);
361 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
362 sector_t sectors, sector_t nr_sectors,
364 int blk_revalidate_disk_zones(struct gendisk *disk,
365 void (*update_driver_data)(struct gendisk *disk));
367 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
368 unsigned int cmd, unsigned long arg);
369 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
370 unsigned int cmd, unsigned long arg);
372 #else /* CONFIG_BLK_DEV_ZONED */
374 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
379 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
380 fmode_t mode, unsigned int cmd,
386 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
387 fmode_t mode, unsigned int cmd,
393 #endif /* CONFIG_BLK_DEV_ZONED */
395 struct request_queue {
396 struct request *last_merge;
397 struct elevator_queue *elevator;
399 struct blk_queue_stats *stats;
400 struct rq_qos *rq_qos;
402 const struct blk_mq_ops *mq_ops;
405 struct blk_mq_ctx __percpu *queue_ctx;
407 unsigned int queue_depth;
409 /* hw dispatch queues */
410 struct blk_mq_hw_ctx **queue_hw_ctx;
411 unsigned int nr_hw_queues;
413 struct backing_dev_info *backing_dev_info;
416 * The queue owner gets to use this for whatever they like.
417 * ll_rw_blk doesn't touch it.
422 * various queue flags, see QUEUE_* below
424 unsigned long queue_flags;
426 * Number of contexts that have called blk_set_pm_only(). If this
427 * counter is above zero then only RQF_PM and RQF_PREEMPT requests are
433 * ida allocated id for this queue. Used to index queues from
439 * queue needs bounce pages for pages above this limit
443 spinlock_t queue_lock;
453 struct kobject *mq_kobj;
455 #ifdef CONFIG_BLK_DEV_INTEGRITY
456 struct blk_integrity integrity;
457 #endif /* CONFIG_BLK_DEV_INTEGRITY */
462 unsigned int nr_pending;
468 unsigned long nr_requests; /* Max # of requests */
470 unsigned int dma_pad_mask;
471 unsigned int dma_alignment;
473 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
474 /* Inline crypto capabilities */
475 struct blk_keyslot_manager *ksm;
478 unsigned int rq_timeout;
481 struct blk_stat_callback *poll_cb;
482 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
484 struct timer_list timeout;
485 struct work_struct timeout_work;
487 struct list_head icq_list;
488 #ifdef CONFIG_BLK_CGROUP
489 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
490 struct blkcg_gq *root_blkg;
491 struct list_head blkg_list;
494 struct queue_limits limits;
496 unsigned int required_elevator_features;
498 #ifdef CONFIG_BLK_DEV_ZONED
500 * Zoned block device information for request dispatch control.
501 * nr_zones is the total number of zones of the device. This is always
502 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
503 * bits which indicates if a zone is conventional (bit set) or
504 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
505 * bits which indicates if a zone is write locked, that is, if a write
506 * request targeting the zone was dispatched. All three fields are
507 * initialized by the low level device driver (e.g. scsi/sd.c).
508 * Stacking drivers (device mappers) may or may not initialize
511 * Reads of this information must be protected with blk_queue_enter() /
512 * blk_queue_exit(). Modifying this information is only allowed while
513 * no requests are being processed. See also blk_mq_freeze_queue() and
514 * blk_mq_unfreeze_queue().
516 unsigned int nr_zones;
517 unsigned long *conv_zones_bitmap;
518 unsigned long *seq_zones_wlock;
519 unsigned int max_open_zones;
520 unsigned int max_active_zones;
521 #endif /* CONFIG_BLK_DEV_ZONED */
526 unsigned int sg_timeout;
527 unsigned int sg_reserved_size;
529 struct mutex debugfs_mutex;
530 #ifdef CONFIG_BLK_DEV_IO_TRACE
531 struct blk_trace __rcu *blk_trace;
534 * for flush operations
536 struct blk_flush_queue *fq;
538 struct list_head requeue_list;
539 spinlock_t requeue_lock;
540 struct delayed_work requeue_work;
542 struct mutex sysfs_lock;
543 struct mutex sysfs_dir_lock;
546 * for reusing dead hctx instance in case of updating
549 struct list_head unused_hctx_list;
550 spinlock_t unused_hctx_lock;
554 #if defined(CONFIG_BLK_DEV_BSG)
555 struct bsg_class_device bsg_dev;
558 #ifdef CONFIG_BLK_DEV_THROTTLING
560 struct throtl_data *td;
562 struct rcu_head rcu_head;
563 wait_queue_head_t mq_freeze_wq;
565 * Protect concurrent access to q_usage_counter by
566 * percpu_ref_kill() and percpu_ref_reinit().
568 struct mutex mq_freeze_lock;
569 struct percpu_ref q_usage_counter;
571 struct blk_mq_tag_set *tag_set;
572 struct list_head tag_set_list;
573 struct bio_set bio_split;
575 struct dentry *debugfs_dir;
577 #ifdef CONFIG_BLK_DEBUG_FS
578 struct dentry *sched_debugfs_dir;
579 struct dentry *rqos_debugfs_dir;
582 bool mq_sysfs_init_done;
586 #define BLK_MAX_WRITE_HINTS 5
587 u64 write_hints[BLK_MAX_WRITE_HINTS];
590 /* Keep blk_queue_flag_name[] in sync with the definitions below */
591 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
592 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
593 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
594 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
595 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
596 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
597 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
598 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
599 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
600 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
601 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
602 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
603 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
604 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
605 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
606 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
607 #define QUEUE_FLAG_WC 17 /* Write back caching */
608 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
609 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
610 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
611 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
612 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
613 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
614 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
615 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
616 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
617 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
619 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
620 (1 << QUEUE_FLAG_SAME_COMP))
622 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
623 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
624 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
626 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
627 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
628 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
629 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
630 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
631 #define blk_queue_noxmerges(q) \
632 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
633 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
634 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
635 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
636 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
637 #define blk_queue_zone_resetall(q) \
638 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
639 #define blk_queue_secure_erase(q) \
640 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
641 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
642 #define blk_queue_scsi_passthrough(q) \
643 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
644 #define blk_queue_pci_p2pdma(q) \
645 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
646 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
647 #define blk_queue_rq_alloc_time(q) \
648 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
650 #define blk_queue_rq_alloc_time(q) false
653 #define blk_noretry_request(rq) \
654 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
655 REQ_FAILFAST_DRIVER))
656 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
657 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
658 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
659 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
661 extern void blk_set_pm_only(struct request_queue *q);
662 extern void blk_clear_pm_only(struct request_queue *q);
664 static inline bool blk_account_rq(struct request *rq)
666 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
669 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
671 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
673 #define rq_dma_dir(rq) \
674 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
676 #define dma_map_bvec(dev, bv, dir, attrs) \
677 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
680 static inline bool queue_is_mq(struct request_queue *q)
685 static inline enum blk_zoned_model
686 blk_queue_zoned_model(struct request_queue *q)
688 return q->limits.zoned;
691 static inline bool blk_queue_is_zoned(struct request_queue *q)
693 switch (blk_queue_zoned_model(q)) {
702 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
704 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
707 #ifdef CONFIG_BLK_DEV_ZONED
708 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
710 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
713 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
716 if (!blk_queue_is_zoned(q))
718 return sector >> ilog2(q->limits.chunk_sectors);
721 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
724 if (!blk_queue_is_zoned(q))
726 if (!q->conv_zones_bitmap)
728 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
731 static inline void blk_queue_max_open_zones(struct request_queue *q,
732 unsigned int max_open_zones)
734 q->max_open_zones = max_open_zones;
737 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
739 return q->max_open_zones;
742 static inline void blk_queue_max_active_zones(struct request_queue *q,
743 unsigned int max_active_zones)
745 q->max_active_zones = max_active_zones;
748 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
750 return q->max_active_zones;
752 #else /* CONFIG_BLK_DEV_ZONED */
753 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
757 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
762 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
767 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
771 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
775 #endif /* CONFIG_BLK_DEV_ZONED */
777 static inline bool rq_is_sync(struct request *rq)
779 return op_is_sync(rq->cmd_flags);
782 static inline bool rq_mergeable(struct request *rq)
784 if (blk_rq_is_passthrough(rq))
787 if (req_op(rq) == REQ_OP_FLUSH)
790 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
793 if (req_op(rq) == REQ_OP_ZONE_APPEND)
796 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
798 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
804 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
806 if (bio_page(a) == bio_page(b) &&
807 bio_offset(a) == bio_offset(b))
813 static inline unsigned int blk_queue_depth(struct request_queue *q)
816 return q->queue_depth;
818 return q->nr_requests;
821 extern unsigned long blk_max_low_pfn, blk_max_pfn;
824 * standard bounce addresses:
826 * BLK_BOUNCE_HIGH : bounce all highmem pages
827 * BLK_BOUNCE_ANY : don't bounce anything
828 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
831 #if BITS_PER_LONG == 32
832 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
834 #define BLK_BOUNCE_HIGH -1ULL
836 #define BLK_BOUNCE_ANY (-1ULL)
837 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
840 * default timeout for SG_IO if none specified
842 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
843 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
849 unsigned long offset;
854 struct req_iterator {
855 struct bvec_iter iter;
859 /* This should not be used directly - use rq_for_each_segment */
860 #define for_each_bio(_bio) \
861 for (; _bio; _bio = _bio->bi_next)
862 #define __rq_for_each_bio(_bio, rq) \
864 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
866 #define rq_for_each_segment(bvl, _rq, _iter) \
867 __rq_for_each_bio(_iter.bio, _rq) \
868 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
870 #define rq_for_each_bvec(bvl, _rq, _iter) \
871 __rq_for_each_bio(_iter.bio, _rq) \
872 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
874 #define rq_iter_last(bvec, _iter) \
875 (_iter.bio->bi_next == NULL && \
876 bio_iter_last(bvec, _iter.iter))
878 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
879 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
881 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
882 extern void rq_flush_dcache_pages(struct request *rq);
884 static inline void rq_flush_dcache_pages(struct request *rq)
889 extern int blk_register_queue(struct gendisk *disk);
890 extern void blk_unregister_queue(struct gendisk *disk);
891 blk_qc_t submit_bio_noacct(struct bio *bio);
892 extern void blk_rq_init(struct request_queue *q, struct request *rq);
893 extern void blk_put_request(struct request *);
894 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
895 blk_mq_req_flags_t flags);
896 extern int blk_lld_busy(struct request_queue *q);
897 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
898 struct bio_set *bs, gfp_t gfp_mask,
899 int (*bio_ctr)(struct bio *, struct bio *, void *),
901 extern void blk_rq_unprep_clone(struct request *rq);
902 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
904 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
905 extern void blk_queue_split(struct bio **);
906 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
907 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
908 unsigned int, void __user *);
909 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
910 unsigned int, void __user *);
911 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
912 struct scsi_ioctl_command __user *);
913 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
914 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
916 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
917 extern void blk_queue_exit(struct request_queue *q);
918 extern void blk_sync_queue(struct request_queue *q);
919 extern int blk_rq_map_user(struct request_queue *, struct request *,
920 struct rq_map_data *, void __user *, unsigned long,
922 extern int blk_rq_unmap_user(struct bio *);
923 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
924 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
925 struct rq_map_data *, const struct iov_iter *,
927 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
928 struct request *, int);
929 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
930 struct request *, int, rq_end_io_fn *);
932 /* Helper to convert REQ_OP_XXX to its string format XXX */
933 extern const char *blk_op_str(unsigned int op);
935 int blk_status_to_errno(blk_status_t status);
936 blk_status_t errno_to_blk_status(int errno);
938 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
940 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
942 return bdev->bd_disk->queue; /* this is never NULL */
946 * The basic unit of block I/O is a sector. It is used in a number of contexts
947 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
948 * bytes. Variables of type sector_t represent an offset or size that is a
949 * multiple of 512 bytes. Hence these two constants.
952 #define SECTOR_SHIFT 9
955 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
959 * blk_rq_pos() : the current sector
960 * blk_rq_bytes() : bytes left in the entire request
961 * blk_rq_cur_bytes() : bytes left in the current segment
962 * blk_rq_err_bytes() : bytes left till the next error boundary
963 * blk_rq_sectors() : sectors left in the entire request
964 * blk_rq_cur_sectors() : sectors left in the current segment
965 * blk_rq_stats_sectors() : sectors of the entire request used for stats
967 static inline sector_t blk_rq_pos(const struct request *rq)
972 static inline unsigned int blk_rq_bytes(const struct request *rq)
974 return rq->__data_len;
977 static inline int blk_rq_cur_bytes(const struct request *rq)
979 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
982 extern unsigned int blk_rq_err_bytes(const struct request *rq);
984 static inline unsigned int blk_rq_sectors(const struct request *rq)
986 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
989 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
991 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
994 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
996 return rq->stats_sectors;
999 #ifdef CONFIG_BLK_DEV_ZONED
1001 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
1002 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
1004 static inline unsigned int blk_rq_zone_no(struct request *rq)
1006 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1009 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1011 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1013 #endif /* CONFIG_BLK_DEV_ZONED */
1016 * Some commands like WRITE SAME have a payload or data transfer size which
1017 * is different from the size of the request. Any driver that supports such
1018 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1019 * calculate the data transfer size.
1021 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1023 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1024 return rq->special_vec.bv_len;
1025 return blk_rq_bytes(rq);
1029 * Return the first full biovec in the request. The caller needs to check that
1030 * there are any bvecs before calling this helper.
1032 static inline struct bio_vec req_bvec(struct request *rq)
1034 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1035 return rq->special_vec;
1036 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1039 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1042 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1043 return min(q->limits.max_discard_sectors,
1044 UINT_MAX >> SECTOR_SHIFT);
1046 if (unlikely(op == REQ_OP_WRITE_SAME))
1047 return q->limits.max_write_same_sectors;
1049 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1050 return q->limits.max_write_zeroes_sectors;
1052 return q->limits.max_sectors;
1056 * Return maximum size of a request at given offset. Only valid for
1057 * file system requests.
1059 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1062 if (!q->limits.chunk_sectors)
1063 return q->limits.max_sectors;
1065 return min(q->limits.max_sectors, (unsigned int)(q->limits.chunk_sectors -
1066 (offset & (q->limits.chunk_sectors - 1))));
1069 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1072 struct request_queue *q = rq->q;
1074 if (blk_rq_is_passthrough(rq))
1075 return q->limits.max_hw_sectors;
1077 if (!q->limits.chunk_sectors ||
1078 req_op(rq) == REQ_OP_DISCARD ||
1079 req_op(rq) == REQ_OP_SECURE_ERASE)
1080 return blk_queue_get_max_sectors(q, req_op(rq));
1082 return min(blk_max_size_offset(q, offset),
1083 blk_queue_get_max_sectors(q, req_op(rq)));
1086 static inline unsigned int blk_rq_count_bios(struct request *rq)
1088 unsigned int nr_bios = 0;
1091 __rq_for_each_bio(bio, rq)
1097 void blk_steal_bios(struct bio_list *list, struct request *rq);
1100 * Request completion related functions.
1102 * blk_update_request() completes given number of bytes and updates
1103 * the request without completing it.
1105 extern bool blk_update_request(struct request *rq, blk_status_t error,
1106 unsigned int nr_bytes);
1108 extern void blk_abort_request(struct request *);
1111 * Access functions for manipulating queue properties
1113 extern void blk_cleanup_queue(struct request_queue *);
1114 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1115 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1116 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1117 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1118 extern void blk_queue_max_discard_segments(struct request_queue *,
1120 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1121 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1122 unsigned int max_discard_sectors);
1123 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1124 unsigned int max_write_same_sectors);
1125 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1126 unsigned int max_write_same_sectors);
1127 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1128 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1129 unsigned int max_zone_append_sectors);
1130 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1131 extern void blk_queue_alignment_offset(struct request_queue *q,
1132 unsigned int alignment);
1133 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1134 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1135 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1136 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1137 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1138 extern void blk_set_default_limits(struct queue_limits *lim);
1139 extern void blk_set_stacking_limits(struct queue_limits *lim);
1140 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1142 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1144 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1145 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1146 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1147 extern void blk_queue_dma_alignment(struct request_queue *, int);
1148 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1149 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1150 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1151 extern void blk_queue_required_elevator_features(struct request_queue *q,
1152 unsigned int features);
1153 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1154 struct device *dev);
1157 * Number of physical segments as sent to the device.
1159 * Normally this is the number of discontiguous data segments sent by the
1160 * submitter. But for data-less command like discard we might have no
1161 * actual data segments submitted, but the driver might have to add it's
1162 * own special payload. In that case we still return 1 here so that this
1163 * special payload will be mapped.
1165 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1167 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1169 return rq->nr_phys_segments;
1173 * Number of discard segments (or ranges) the driver needs to fill in.
1174 * Each discard bio merged into a request is counted as one segment.
1176 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1178 return max_t(unsigned short, rq->nr_phys_segments, 1);
1181 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1182 struct scatterlist *sglist, struct scatterlist **last_sg);
1183 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1184 struct scatterlist *sglist)
1186 struct scatterlist *last_sg = NULL;
1188 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1190 extern void blk_dump_rq_flags(struct request *, char *);
1192 bool __must_check blk_get_queue(struct request_queue *);
1193 struct request_queue *blk_alloc_queue(int node_id);
1194 extern void blk_put_queue(struct request_queue *);
1195 extern void blk_set_queue_dying(struct request_queue *);
1199 * blk_plug permits building a queue of related requests by holding the I/O
1200 * fragments for a short period. This allows merging of sequential requests
1201 * into single larger request. As the requests are moved from a per-task list to
1202 * the device's request_queue in a batch, this results in improved scalability
1203 * as the lock contention for request_queue lock is reduced.
1205 * It is ok not to disable preemption when adding the request to the plug list
1206 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1207 * the plug list when the task sleeps by itself. For details, please see
1208 * schedule() where blk_schedule_flush_plug() is called.
1211 struct list_head mq_list; /* blk-mq requests */
1212 struct list_head cb_list; /* md requires an unplug callback */
1213 unsigned short rq_count;
1214 bool multiple_queues;
1217 #define BLK_MAX_REQUEST_COUNT 16
1218 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1221 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1222 struct blk_plug_cb {
1223 struct list_head list;
1224 blk_plug_cb_fn callback;
1227 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1228 void *data, int size);
1229 extern void blk_start_plug(struct blk_plug *);
1230 extern void blk_finish_plug(struct blk_plug *);
1231 extern void blk_flush_plug_list(struct blk_plug *, bool);
1233 static inline void blk_flush_plug(struct task_struct *tsk)
1235 struct blk_plug *plug = tsk->plug;
1238 blk_flush_plug_list(plug, false);
1241 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1243 struct blk_plug *plug = tsk->plug;
1246 blk_flush_plug_list(plug, true);
1249 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1251 struct blk_plug *plug = tsk->plug;
1254 (!list_empty(&plug->mq_list) ||
1255 !list_empty(&plug->cb_list));
1258 int blkdev_issue_flush(struct block_device *, gfp_t);
1259 long nr_blockdev_pages(void);
1260 #else /* CONFIG_BLOCK */
1264 static inline void blk_start_plug(struct blk_plug *plug)
1268 static inline void blk_finish_plug(struct blk_plug *plug)
1272 static inline void blk_flush_plug(struct task_struct *task)
1276 static inline void blk_schedule_flush_plug(struct task_struct *task)
1281 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1286 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask)
1291 static inline long nr_blockdev_pages(void)
1295 #endif /* CONFIG_BLOCK */
1297 extern void blk_io_schedule(void);
1299 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1300 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1302 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1304 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1305 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1306 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1307 sector_t nr_sects, gfp_t gfp_mask, int flags,
1310 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1311 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1313 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1314 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1316 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1317 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1319 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1320 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1322 return blkdev_issue_discard(sb->s_bdev,
1323 block << (sb->s_blocksize_bits -
1325 nr_blocks << (sb->s_blocksize_bits -
1329 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1330 sector_t nr_blocks, gfp_t gfp_mask)
1332 return blkdev_issue_zeroout(sb->s_bdev,
1333 block << (sb->s_blocksize_bits -
1335 nr_blocks << (sb->s_blocksize_bits -
1340 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1342 enum blk_default_limits {
1343 BLK_MAX_SEGMENTS = 128,
1344 BLK_SAFE_MAX_SECTORS = 255,
1345 BLK_DEF_MAX_SECTORS = 2560,
1346 BLK_MAX_SEGMENT_SIZE = 65536,
1347 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1350 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1352 return q->limits.seg_boundary_mask;
1355 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1357 return q->limits.virt_boundary_mask;
1360 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1362 return q->limits.max_sectors;
1365 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1367 return q->limits.max_hw_sectors;
1370 static inline unsigned short queue_max_segments(const struct request_queue *q)
1372 return q->limits.max_segments;
1375 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1377 return q->limits.max_discard_segments;
1380 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1382 return q->limits.max_segment_size;
1385 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1387 return q->limits.max_zone_append_sectors;
1390 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1394 if (q && q->limits.logical_block_size)
1395 retval = q->limits.logical_block_size;
1400 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1402 return queue_logical_block_size(bdev_get_queue(bdev));
1405 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1407 return q->limits.physical_block_size;
1410 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1412 return queue_physical_block_size(bdev_get_queue(bdev));
1415 static inline unsigned int queue_io_min(const struct request_queue *q)
1417 return q->limits.io_min;
1420 static inline int bdev_io_min(struct block_device *bdev)
1422 return queue_io_min(bdev_get_queue(bdev));
1425 static inline unsigned int queue_io_opt(const struct request_queue *q)
1427 return q->limits.io_opt;
1430 static inline int bdev_io_opt(struct block_device *bdev)
1432 return queue_io_opt(bdev_get_queue(bdev));
1435 static inline int queue_alignment_offset(const struct request_queue *q)
1437 if (q->limits.misaligned)
1440 return q->limits.alignment_offset;
1443 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1445 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1446 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1449 return (granularity + lim->alignment_offset - alignment) % granularity;
1452 static inline int bdev_alignment_offset(struct block_device *bdev)
1454 struct request_queue *q = bdev_get_queue(bdev);
1456 if (q->limits.misaligned)
1459 if (bdev != bdev->bd_contains)
1460 return bdev->bd_part->alignment_offset;
1462 return q->limits.alignment_offset;
1465 static inline int queue_discard_alignment(const struct request_queue *q)
1467 if (q->limits.discard_misaligned)
1470 return q->limits.discard_alignment;
1473 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1475 unsigned int alignment, granularity, offset;
1477 if (!lim->max_discard_sectors)
1480 /* Why are these in bytes, not sectors? */
1481 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1482 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1486 /* Offset of the partition start in 'granularity' sectors */
1487 offset = sector_div(sector, granularity);
1489 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1490 offset = (granularity + alignment - offset) % granularity;
1492 /* Turn it back into bytes, gaah */
1493 return offset << SECTOR_SHIFT;
1496 static inline int bdev_discard_alignment(struct block_device *bdev)
1498 struct request_queue *q = bdev_get_queue(bdev);
1500 if (bdev != bdev->bd_contains)
1501 return bdev->bd_part->discard_alignment;
1503 return q->limits.discard_alignment;
1506 static inline unsigned int bdev_write_same(struct block_device *bdev)
1508 struct request_queue *q = bdev_get_queue(bdev);
1511 return q->limits.max_write_same_sectors;
1516 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1518 struct request_queue *q = bdev_get_queue(bdev);
1521 return q->limits.max_write_zeroes_sectors;
1526 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1528 struct request_queue *q = bdev_get_queue(bdev);
1531 return blk_queue_zoned_model(q);
1533 return BLK_ZONED_NONE;
1536 static inline bool bdev_is_zoned(struct block_device *bdev)
1538 struct request_queue *q = bdev_get_queue(bdev);
1541 return blk_queue_is_zoned(q);
1546 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1548 struct request_queue *q = bdev_get_queue(bdev);
1551 return blk_queue_zone_sectors(q);
1555 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1557 struct request_queue *q = bdev_get_queue(bdev);
1560 return queue_max_open_zones(q);
1564 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1566 struct request_queue *q = bdev_get_queue(bdev);
1569 return queue_max_active_zones(q);
1573 static inline int queue_dma_alignment(const struct request_queue *q)
1575 return q ? q->dma_alignment : 511;
1578 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1581 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1582 return !(addr & alignment) && !(len & alignment);
1585 /* assumes size > 256 */
1586 static inline unsigned int blksize_bits(unsigned int size)
1588 unsigned int bits = 8;
1592 } while (size > 256);
1596 static inline unsigned int block_size(struct block_device *bdev)
1598 return 1 << bdev->bd_inode->i_blkbits;
1601 int kblockd_schedule_work(struct work_struct *work);
1602 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1604 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1605 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1606 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1607 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1609 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1611 enum blk_integrity_flags {
1612 BLK_INTEGRITY_VERIFY = 1 << 0,
1613 BLK_INTEGRITY_GENERATE = 1 << 1,
1614 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1615 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1618 struct blk_integrity_iter {
1622 unsigned int data_size;
1623 unsigned short interval;
1624 const char *disk_name;
1627 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1628 typedef void (integrity_prepare_fn) (struct request *);
1629 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1631 struct blk_integrity_profile {
1632 integrity_processing_fn *generate_fn;
1633 integrity_processing_fn *verify_fn;
1634 integrity_prepare_fn *prepare_fn;
1635 integrity_complete_fn *complete_fn;
1639 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1640 extern void blk_integrity_unregister(struct gendisk *);
1641 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1642 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1643 struct scatterlist *);
1644 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1645 extern bool blk_integrity_merge_rq(struct request_queue *, struct request *,
1647 extern bool blk_integrity_merge_bio(struct request_queue *, struct request *,
1650 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1652 struct blk_integrity *bi = &disk->queue->integrity;
1661 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1663 return blk_get_integrity(bdev->bd_disk);
1667 blk_integrity_queue_supports_integrity(struct request_queue *q)
1669 return q->integrity.profile;
1672 static inline bool blk_integrity_rq(struct request *rq)
1674 return rq->cmd_flags & REQ_INTEGRITY;
1677 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1680 q->limits.max_integrity_segments = segs;
1683 static inline unsigned short
1684 queue_max_integrity_segments(const struct request_queue *q)
1686 return q->limits.max_integrity_segments;
1690 * bio_integrity_intervals - Return number of integrity intervals for a bio
1691 * @bi: blk_integrity profile for device
1692 * @sectors: Size of the bio in 512-byte sectors
1694 * Description: The block layer calculates everything in 512 byte
1695 * sectors but integrity metadata is done in terms of the data integrity
1696 * interval size of the storage device. Convert the block layer sectors
1697 * to the appropriate number of integrity intervals.
1699 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1700 unsigned int sectors)
1702 return sectors >> (bi->interval_exp - 9);
1705 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1706 unsigned int sectors)
1708 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1712 * Return the first bvec that contains integrity data. Only drivers that are
1713 * limited to a single integrity segment should use this helper.
1715 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1717 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1719 return rq->bio->bi_integrity->bip_vec;
1722 #else /* CONFIG_BLK_DEV_INTEGRITY */
1725 struct block_device;
1727 struct blk_integrity;
1729 static inline int blk_integrity_rq(struct request *rq)
1733 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1738 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1740 struct scatterlist *s)
1744 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1748 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1753 blk_integrity_queue_supports_integrity(struct request_queue *q)
1757 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1761 static inline void blk_integrity_register(struct gendisk *d,
1762 struct blk_integrity *b)
1765 static inline void blk_integrity_unregister(struct gendisk *d)
1768 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1772 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1776 static inline bool blk_integrity_merge_rq(struct request_queue *rq,
1782 static inline bool blk_integrity_merge_bio(struct request_queue *rq,
1789 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1790 unsigned int sectors)
1795 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1796 unsigned int sectors)
1801 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1806 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1808 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1810 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1812 void blk_ksm_unregister(struct request_queue *q);
1814 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1816 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1817 struct request_queue *q)
1822 static inline void blk_ksm_unregister(struct request_queue *q) { }
1824 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1827 struct block_device_operations {
1828 blk_qc_t (*submit_bio) (struct bio *bio);
1829 int (*open) (struct block_device *, fmode_t);
1830 void (*release) (struct gendisk *, fmode_t);
1831 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1832 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1833 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1834 unsigned int (*check_events) (struct gendisk *disk,
1835 unsigned int clearing);
1836 void (*unlock_native_capacity) (struct gendisk *);
1837 int (*revalidate_disk) (struct gendisk *);
1838 int (*getgeo)(struct block_device *, struct hd_geometry *);
1839 /* this callback is with swap_lock and sometimes page table lock held */
1840 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1841 int (*report_zones)(struct gendisk *, sector_t sector,
1842 unsigned int nr_zones, report_zones_cb cb, void *data);
1843 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1844 struct module *owner;
1845 const struct pr_ops *pr_ops;
1848 #ifdef CONFIG_COMPAT
1849 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1850 unsigned int, unsigned long);
1852 #define blkdev_compat_ptr_ioctl NULL
1855 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1857 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1858 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1859 struct writeback_control *);
1861 #ifdef CONFIG_BLK_DEV_ZONED
1862 bool blk_req_needs_zone_write_lock(struct request *rq);
1863 bool blk_req_zone_write_trylock(struct request *rq);
1864 void __blk_req_zone_write_lock(struct request *rq);
1865 void __blk_req_zone_write_unlock(struct request *rq);
1867 static inline void blk_req_zone_write_lock(struct request *rq)
1869 if (blk_req_needs_zone_write_lock(rq))
1870 __blk_req_zone_write_lock(rq);
1873 static inline void blk_req_zone_write_unlock(struct request *rq)
1875 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1876 __blk_req_zone_write_unlock(rq);
1879 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1881 return rq->q->seq_zones_wlock &&
1882 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1885 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1887 if (!blk_req_needs_zone_write_lock(rq))
1889 return !blk_req_zone_is_write_locked(rq);
1892 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1897 static inline void blk_req_zone_write_lock(struct request *rq)
1901 static inline void blk_req_zone_write_unlock(struct request *rq)
1904 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1909 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1913 #endif /* CONFIG_BLK_DEV_ZONED */
1915 static inline void blk_wake_io_task(struct task_struct *waiter)
1918 * If we're polling, the task itself is doing the completions. For
1919 * that case, we don't need to signal a wakeup, it's enough to just
1920 * mark us as RUNNING.
1922 if (waiter == current)
1923 __set_current_state(TASK_RUNNING);
1925 wake_up_process(waiter);
1928 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1930 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1931 unsigned long start_time);
1934 * bio_start_io_acct - start I/O accounting for bio based drivers
1935 * @bio: bio to start account for
1937 * Returns the start time that should be passed back to bio_end_io_acct().
1939 static inline unsigned long bio_start_io_acct(struct bio *bio)
1941 return disk_start_io_acct(bio->bi_disk, bio_sectors(bio), bio_op(bio));
1945 * bio_end_io_acct - end I/O accounting for bio based drivers
1946 * @bio: bio to end account for
1947 * @start: start time returned by bio_start_io_acct()
1949 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1951 return disk_end_io_acct(bio->bi_disk, bio_op(bio), start_time);
1954 int bdev_read_only(struct block_device *bdev);
1955 int set_blocksize(struct block_device *bdev, int size);
1957 const char *bdevname(struct block_device *bdev, char *buffer);
1958 struct block_device *lookup_bdev(const char *);
1960 void blkdev_show(struct seq_file *seqf, off_t offset);
1962 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1963 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1965 #define BLKDEV_MAJOR_MAX 512
1967 #define BLKDEV_MAJOR_MAX 0
1970 int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
1971 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1973 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1974 int bd_prepare_to_claim(struct block_device *bdev, struct block_device *whole,
1976 void bd_abort_claiming(struct block_device *bdev, struct block_device *whole,
1978 void blkdev_put(struct block_device *bdev, fmode_t mode);
1980 struct block_device *I_BDEV(struct inode *inode);
1981 struct block_device *bdget(dev_t);
1982 struct block_device *bdgrab(struct block_device *bdev);
1983 void bdput(struct block_device *);
1986 void invalidate_bdev(struct block_device *bdev);
1987 int sync_blockdev(struct block_device *bdev);
1989 static inline void invalidate_bdev(struct block_device *bdev)
1992 static inline int sync_blockdev(struct block_device *bdev)
1997 int fsync_bdev(struct block_device *bdev);
1999 struct super_block *freeze_bdev(struct block_device *bdev);
2000 int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2002 #endif /* _LINUX_BLKDEV_H */