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 unsigned short req_get_ioprio(struct request *req)
280 #include <linux/elevator.h>
282 struct blk_queue_ctx;
286 enum blk_eh_timer_return {
287 BLK_EH_DONE, /* drivers has completed the command */
288 BLK_EH_RESET_TIMER, /* reset timer and try again */
291 enum blk_queue_state {
296 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
297 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
299 #define BLK_SCSI_MAX_CMDS (256)
300 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
303 * Zoned block device models (zoned limit).
305 * Note: This needs to be ordered from the least to the most severe
306 * restrictions for the inheritance in blk_stack_limits() to work.
308 enum blk_zoned_model {
309 BLK_ZONED_NONE = 0, /* Regular block device */
310 BLK_ZONED_HA, /* Host-aware zoned block device */
311 BLK_ZONED_HM, /* Host-managed zoned block device */
314 struct queue_limits {
315 unsigned long bounce_pfn;
316 unsigned long seg_boundary_mask;
317 unsigned long virt_boundary_mask;
319 unsigned int max_hw_sectors;
320 unsigned int max_dev_sectors;
321 unsigned int chunk_sectors;
322 unsigned int max_sectors;
323 unsigned int max_segment_size;
324 unsigned int physical_block_size;
325 unsigned int logical_block_size;
326 unsigned int alignment_offset;
329 unsigned int max_discard_sectors;
330 unsigned int max_hw_discard_sectors;
331 unsigned int max_write_same_sectors;
332 unsigned int max_write_zeroes_sectors;
333 unsigned int max_zone_append_sectors;
334 unsigned int discard_granularity;
335 unsigned int discard_alignment;
336 unsigned int zone_write_granularity;
338 unsigned short max_segments;
339 unsigned short max_integrity_segments;
340 unsigned short max_discard_segments;
342 unsigned char misaligned;
343 unsigned char discard_misaligned;
344 unsigned char raid_partial_stripes_expensive;
345 enum blk_zoned_model zoned;
348 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
351 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
353 #ifdef CONFIG_BLK_DEV_ZONED
355 #define BLK_ALL_ZONES ((unsigned int)-1)
356 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
357 unsigned int nr_zones, report_zones_cb cb, void *data);
358 unsigned int blkdev_nr_zones(struct gendisk *disk);
359 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
360 sector_t sectors, sector_t nr_sectors,
362 int blk_revalidate_disk_zones(struct gendisk *disk,
363 void (*update_driver_data)(struct gendisk *disk));
365 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
366 unsigned int cmd, unsigned long arg);
367 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
368 unsigned int cmd, unsigned long arg);
370 #else /* CONFIG_BLK_DEV_ZONED */
372 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
377 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
378 fmode_t mode, unsigned int cmd,
384 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
385 fmode_t mode, unsigned int cmd,
391 #endif /* CONFIG_BLK_DEV_ZONED */
393 struct request_queue {
394 struct request *last_merge;
395 struct elevator_queue *elevator;
397 struct percpu_ref q_usage_counter;
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 requests are processed.
432 * ida allocated id for this queue. Used to index queues from
438 * queue needs bounce pages for pages above this limit
442 spinlock_t queue_lock;
452 struct kobject *mq_kobj;
454 #ifdef CONFIG_BLK_DEV_INTEGRITY
455 struct blk_integrity integrity;
456 #endif /* CONFIG_BLK_DEV_INTEGRITY */
460 enum rpm_status rpm_status;
466 unsigned long nr_requests; /* Max # of requests */
468 unsigned int dma_pad_mask;
469 unsigned int dma_alignment;
471 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
472 /* Inline crypto capabilities */
473 struct blk_keyslot_manager *ksm;
476 unsigned int rq_timeout;
479 struct blk_stat_callback *poll_cb;
480 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
482 struct timer_list timeout;
483 struct work_struct timeout_work;
485 atomic_t nr_active_requests_shared_sbitmap;
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;
570 struct blk_mq_tag_set *tag_set;
571 struct list_head tag_set_list;
572 struct bio_set bio_split;
574 struct dentry *debugfs_dir;
576 #ifdef CONFIG_BLK_DEBUG_FS
577 struct dentry *sched_debugfs_dir;
578 struct dentry *rqos_debugfs_dir;
581 bool mq_sysfs_init_done;
585 #define BLK_MAX_WRITE_HINTS 5
586 u64 write_hints[BLK_MAX_WRITE_HINTS];
589 /* Keep blk_queue_flag_name[] in sync with the definitions below */
590 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
591 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
592 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
593 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
594 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
595 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
596 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
597 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
598 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
599 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
600 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
601 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
602 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
603 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
604 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
605 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
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 */
618 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
619 #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */
621 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
622 (1 << QUEUE_FLAG_SAME_COMP) | \
623 (1 << QUEUE_FLAG_NOWAIT))
625 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
626 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
627 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
629 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
630 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
631 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
632 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
633 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
634 #define blk_queue_noxmerges(q) \
635 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
636 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
637 #define blk_queue_stable_writes(q) \
638 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
639 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
640 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
641 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
642 #define blk_queue_zone_resetall(q) \
643 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
644 #define blk_queue_secure_erase(q) \
645 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
646 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
647 #define blk_queue_scsi_passthrough(q) \
648 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
649 #define blk_queue_pci_p2pdma(q) \
650 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
651 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
652 #define blk_queue_rq_alloc_time(q) \
653 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
655 #define blk_queue_rq_alloc_time(q) false
658 #define blk_noretry_request(rq) \
659 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
660 REQ_FAILFAST_DRIVER))
661 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
662 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
663 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
664 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
665 #define blk_queue_nowait(q) test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
667 extern void blk_set_pm_only(struct request_queue *q);
668 extern void blk_clear_pm_only(struct request_queue *q);
670 static inline bool blk_account_rq(struct request *rq)
672 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
675 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
677 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
679 #define rq_dma_dir(rq) \
680 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
682 #define dma_map_bvec(dev, bv, dir, attrs) \
683 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
686 static inline bool queue_is_mq(struct request_queue *q)
692 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
694 return q->rpm_status;
697 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
703 static inline enum blk_zoned_model
704 blk_queue_zoned_model(struct request_queue *q)
706 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
707 return q->limits.zoned;
708 return BLK_ZONED_NONE;
711 static inline bool blk_queue_is_zoned(struct request_queue *q)
713 switch (blk_queue_zoned_model(q)) {
722 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
724 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
727 #ifdef CONFIG_BLK_DEV_ZONED
728 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
730 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
733 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
736 if (!blk_queue_is_zoned(q))
738 return sector >> ilog2(q->limits.chunk_sectors);
741 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
744 if (!blk_queue_is_zoned(q))
746 if (!q->conv_zones_bitmap)
748 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
751 static inline void blk_queue_max_open_zones(struct request_queue *q,
752 unsigned int max_open_zones)
754 q->max_open_zones = max_open_zones;
757 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
759 return q->max_open_zones;
762 static inline void blk_queue_max_active_zones(struct request_queue *q,
763 unsigned int max_active_zones)
765 q->max_active_zones = max_active_zones;
768 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
770 return q->max_active_zones;
772 #else /* CONFIG_BLK_DEV_ZONED */
773 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
777 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
782 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
787 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
791 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
795 #endif /* CONFIG_BLK_DEV_ZONED */
797 static inline bool rq_is_sync(struct request *rq)
799 return op_is_sync(rq->cmd_flags);
802 static inline bool rq_mergeable(struct request *rq)
804 if (blk_rq_is_passthrough(rq))
807 if (req_op(rq) == REQ_OP_FLUSH)
810 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
813 if (req_op(rq) == REQ_OP_ZONE_APPEND)
816 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
818 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
824 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
826 if (bio_page(a) == bio_page(b) &&
827 bio_offset(a) == bio_offset(b))
833 static inline unsigned int blk_queue_depth(struct request_queue *q)
836 return q->queue_depth;
838 return q->nr_requests;
841 extern unsigned long blk_max_low_pfn, blk_max_pfn;
844 * standard bounce addresses:
846 * BLK_BOUNCE_HIGH : bounce all highmem pages
847 * BLK_BOUNCE_ANY : don't bounce anything
848 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
851 #if BITS_PER_LONG == 32
852 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
854 #define BLK_BOUNCE_HIGH -1ULL
856 #define BLK_BOUNCE_ANY (-1ULL)
857 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
860 * default timeout for SG_IO if none specified
862 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
863 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
869 unsigned long offset;
874 struct req_iterator {
875 struct bvec_iter iter;
879 /* This should not be used directly - use rq_for_each_segment */
880 #define for_each_bio(_bio) \
881 for (; _bio; _bio = _bio->bi_next)
882 #define __rq_for_each_bio(_bio, rq) \
884 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
886 #define rq_for_each_segment(bvl, _rq, _iter) \
887 __rq_for_each_bio(_iter.bio, _rq) \
888 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
890 #define rq_for_each_bvec(bvl, _rq, _iter) \
891 __rq_for_each_bio(_iter.bio, _rq) \
892 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
894 #define rq_iter_last(bvec, _iter) \
895 (_iter.bio->bi_next == NULL && \
896 bio_iter_last(bvec, _iter.iter))
898 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
899 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
901 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
902 extern void rq_flush_dcache_pages(struct request *rq);
904 static inline void rq_flush_dcache_pages(struct request *rq)
909 extern int blk_register_queue(struct gendisk *disk);
910 extern void blk_unregister_queue(struct gendisk *disk);
911 blk_qc_t submit_bio_noacct(struct bio *bio);
912 extern void blk_rq_init(struct request_queue *q, struct request *rq);
913 extern void blk_put_request(struct request *);
914 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
915 blk_mq_req_flags_t flags);
916 extern int blk_lld_busy(struct request_queue *q);
917 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
918 struct bio_set *bs, gfp_t gfp_mask,
919 int (*bio_ctr)(struct bio *, struct bio *, void *),
921 extern void blk_rq_unprep_clone(struct request *rq);
922 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
924 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
925 extern void blk_queue_split(struct bio **);
926 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
927 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
928 unsigned int, void __user *);
929 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
930 unsigned int, void __user *);
931 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
932 struct scsi_ioctl_command __user *);
933 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
934 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
936 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
937 extern void blk_queue_exit(struct request_queue *q);
938 extern void blk_sync_queue(struct request_queue *q);
939 extern int blk_rq_map_user(struct request_queue *, struct request *,
940 struct rq_map_data *, void __user *, unsigned long,
942 extern int blk_rq_unmap_user(struct bio *);
943 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
944 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
945 struct rq_map_data *, const struct iov_iter *,
947 extern void blk_execute_rq(struct gendisk *, struct request *, int);
948 extern void blk_execute_rq_nowait(struct gendisk *,
949 struct request *, int, rq_end_io_fn *);
951 /* Helper to convert REQ_OP_XXX to its string format XXX */
952 extern const char *blk_op_str(unsigned int op);
954 int blk_status_to_errno(blk_status_t status);
955 blk_status_t errno_to_blk_status(int errno);
957 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
959 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
961 return bdev->bd_disk->queue; /* this is never NULL */
965 * The basic unit of block I/O is a sector. It is used in a number of contexts
966 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
967 * bytes. Variables of type sector_t represent an offset or size that is a
968 * multiple of 512 bytes. Hence these two constants.
971 #define SECTOR_SHIFT 9
974 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
978 * blk_rq_pos() : the current sector
979 * blk_rq_bytes() : bytes left in the entire request
980 * blk_rq_cur_bytes() : bytes left in the current segment
981 * blk_rq_err_bytes() : bytes left till the next error boundary
982 * blk_rq_sectors() : sectors left in the entire request
983 * blk_rq_cur_sectors() : sectors left in the current segment
984 * blk_rq_stats_sectors() : sectors of the entire request used for stats
986 static inline sector_t blk_rq_pos(const struct request *rq)
991 static inline unsigned int blk_rq_bytes(const struct request *rq)
993 return rq->__data_len;
996 static inline int blk_rq_cur_bytes(const struct request *rq)
998 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
1001 extern unsigned int blk_rq_err_bytes(const struct request *rq);
1003 static inline unsigned int blk_rq_sectors(const struct request *rq)
1005 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
1008 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
1010 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
1013 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
1015 return rq->stats_sectors;
1018 #ifdef CONFIG_BLK_DEV_ZONED
1020 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
1021 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
1023 static inline unsigned int blk_rq_zone_no(struct request *rq)
1025 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1028 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1030 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1032 #endif /* CONFIG_BLK_DEV_ZONED */
1035 * Some commands like WRITE SAME have a payload or data transfer size which
1036 * is different from the size of the request. Any driver that supports such
1037 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1038 * calculate the data transfer size.
1040 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1042 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1043 return rq->special_vec.bv_len;
1044 return blk_rq_bytes(rq);
1048 * Return the first full biovec in the request. The caller needs to check that
1049 * there are any bvecs before calling this helper.
1051 static inline struct bio_vec req_bvec(struct request *rq)
1053 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1054 return rq->special_vec;
1055 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1058 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1061 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1062 return min(q->limits.max_discard_sectors,
1063 UINT_MAX >> SECTOR_SHIFT);
1065 if (unlikely(op == REQ_OP_WRITE_SAME))
1066 return q->limits.max_write_same_sectors;
1068 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1069 return q->limits.max_write_zeroes_sectors;
1071 return q->limits.max_sectors;
1075 * Return maximum size of a request at given offset. Only valid for
1076 * file system requests.
1078 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1080 unsigned int chunk_sectors)
1082 if (!chunk_sectors) {
1083 if (q->limits.chunk_sectors)
1084 chunk_sectors = q->limits.chunk_sectors;
1086 return q->limits.max_sectors;
1089 if (likely(is_power_of_2(chunk_sectors)))
1090 chunk_sectors -= offset & (chunk_sectors - 1);
1092 chunk_sectors -= sector_div(offset, chunk_sectors);
1094 return min(q->limits.max_sectors, chunk_sectors);
1097 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1100 struct request_queue *q = rq->q;
1102 if (blk_rq_is_passthrough(rq))
1103 return q->limits.max_hw_sectors;
1105 if (!q->limits.chunk_sectors ||
1106 req_op(rq) == REQ_OP_DISCARD ||
1107 req_op(rq) == REQ_OP_SECURE_ERASE)
1108 return blk_queue_get_max_sectors(q, req_op(rq));
1110 return min(blk_max_size_offset(q, offset, 0),
1111 blk_queue_get_max_sectors(q, req_op(rq)));
1114 static inline unsigned int blk_rq_count_bios(struct request *rq)
1116 unsigned int nr_bios = 0;
1119 __rq_for_each_bio(bio, rq)
1125 void blk_steal_bios(struct bio_list *list, struct request *rq);
1128 * Request completion related functions.
1130 * blk_update_request() completes given number of bytes and updates
1131 * the request without completing it.
1133 extern bool blk_update_request(struct request *rq, blk_status_t error,
1134 unsigned int nr_bytes);
1136 extern void blk_abort_request(struct request *);
1139 * Access functions for manipulating queue properties
1141 extern void blk_cleanup_queue(struct request_queue *);
1142 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1143 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1144 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1145 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1146 extern void blk_queue_max_discard_segments(struct request_queue *,
1148 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1149 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1150 unsigned int max_discard_sectors);
1151 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1152 unsigned int max_write_same_sectors);
1153 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1154 unsigned int max_write_same_sectors);
1155 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1156 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1157 unsigned int max_zone_append_sectors);
1158 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1159 void blk_queue_zone_write_granularity(struct request_queue *q,
1161 extern void blk_queue_alignment_offset(struct request_queue *q,
1162 unsigned int alignment);
1163 void blk_queue_update_readahead(struct request_queue *q);
1164 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1165 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1166 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1167 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1168 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1169 extern void blk_set_default_limits(struct queue_limits *lim);
1170 extern void blk_set_stacking_limits(struct queue_limits *lim);
1171 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1173 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1175 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1176 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1177 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1178 extern void blk_queue_dma_alignment(struct request_queue *, int);
1179 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1180 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1181 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1182 extern void blk_queue_required_elevator_features(struct request_queue *q,
1183 unsigned int features);
1184 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1185 struct device *dev);
1188 * Number of physical segments as sent to the device.
1190 * Normally this is the number of discontiguous data segments sent by the
1191 * submitter. But for data-less command like discard we might have no
1192 * actual data segments submitted, but the driver might have to add it's
1193 * own special payload. In that case we still return 1 here so that this
1194 * special payload will be mapped.
1196 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1198 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1200 return rq->nr_phys_segments;
1204 * Number of discard segments (or ranges) the driver needs to fill in.
1205 * Each discard bio merged into a request is counted as one segment.
1207 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1209 return max_t(unsigned short, rq->nr_phys_segments, 1);
1212 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1213 struct scatterlist *sglist, struct scatterlist **last_sg);
1214 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1215 struct scatterlist *sglist)
1217 struct scatterlist *last_sg = NULL;
1219 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1221 extern void blk_dump_rq_flags(struct request *, char *);
1223 bool __must_check blk_get_queue(struct request_queue *);
1224 struct request_queue *blk_alloc_queue(int node_id);
1225 extern void blk_put_queue(struct request_queue *);
1226 extern void blk_set_queue_dying(struct request_queue *);
1230 * blk_plug permits building a queue of related requests by holding the I/O
1231 * fragments for a short period. This allows merging of sequential requests
1232 * into single larger request. As the requests are moved from a per-task list to
1233 * the device's request_queue in a batch, this results in improved scalability
1234 * as the lock contention for request_queue lock is reduced.
1236 * It is ok not to disable preemption when adding the request to the plug list
1237 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1238 * the plug list when the task sleeps by itself. For details, please see
1239 * schedule() where blk_schedule_flush_plug() is called.
1242 struct list_head mq_list; /* blk-mq requests */
1243 struct list_head cb_list; /* md requires an unplug callback */
1244 unsigned short rq_count;
1245 bool multiple_queues;
1248 #define BLK_MAX_REQUEST_COUNT 16
1249 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1252 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1253 struct blk_plug_cb {
1254 struct list_head list;
1255 blk_plug_cb_fn callback;
1258 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1259 void *data, int size);
1260 extern void blk_start_plug(struct blk_plug *);
1261 extern void blk_finish_plug(struct blk_plug *);
1262 extern void blk_flush_plug_list(struct blk_plug *, bool);
1264 static inline void blk_flush_plug(struct task_struct *tsk)
1266 struct blk_plug *plug = tsk->plug;
1269 blk_flush_plug_list(plug, false);
1272 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1274 struct blk_plug *plug = tsk->plug;
1277 blk_flush_plug_list(plug, true);
1280 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1282 struct blk_plug *plug = tsk->plug;
1285 (!list_empty(&plug->mq_list) ||
1286 !list_empty(&plug->cb_list));
1289 int blkdev_issue_flush(struct block_device *bdev);
1290 long nr_blockdev_pages(void);
1291 #else /* CONFIG_BLOCK */
1295 static inline void blk_start_plug(struct blk_plug *plug)
1299 static inline void blk_finish_plug(struct blk_plug *plug)
1303 static inline void blk_flush_plug(struct task_struct *task)
1307 static inline void blk_schedule_flush_plug(struct task_struct *task)
1312 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1317 static inline int blkdev_issue_flush(struct block_device *bdev)
1322 static inline long nr_blockdev_pages(void)
1326 #endif /* CONFIG_BLOCK */
1328 extern void blk_io_schedule(void);
1330 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1331 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1333 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1335 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1336 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1337 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1338 sector_t nr_sects, gfp_t gfp_mask, int flags,
1341 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1342 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1344 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1345 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1347 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1348 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1350 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1351 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1353 return blkdev_issue_discard(sb->s_bdev,
1354 block << (sb->s_blocksize_bits -
1356 nr_blocks << (sb->s_blocksize_bits -
1360 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1361 sector_t nr_blocks, gfp_t gfp_mask)
1363 return blkdev_issue_zeroout(sb->s_bdev,
1364 block << (sb->s_blocksize_bits -
1366 nr_blocks << (sb->s_blocksize_bits -
1371 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1373 static inline bool bdev_is_partition(struct block_device *bdev)
1375 return bdev->bd_partno;
1378 enum blk_default_limits {
1379 BLK_MAX_SEGMENTS = 128,
1380 BLK_SAFE_MAX_SECTORS = 255,
1381 BLK_DEF_MAX_SECTORS = 2560,
1382 BLK_MAX_SEGMENT_SIZE = 65536,
1383 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1386 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1388 return q->limits.seg_boundary_mask;
1391 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1393 return q->limits.virt_boundary_mask;
1396 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1398 return q->limits.max_sectors;
1401 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1403 return q->limits.max_hw_sectors;
1406 static inline unsigned short queue_max_segments(const struct request_queue *q)
1408 return q->limits.max_segments;
1411 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1413 return q->limits.max_discard_segments;
1416 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1418 return q->limits.max_segment_size;
1421 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1424 const struct queue_limits *l = &q->limits;
1426 return min(l->max_zone_append_sectors, l->max_sectors);
1429 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1433 if (q && q->limits.logical_block_size)
1434 retval = q->limits.logical_block_size;
1439 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1441 return queue_logical_block_size(bdev_get_queue(bdev));
1444 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1446 return q->limits.physical_block_size;
1449 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1451 return queue_physical_block_size(bdev_get_queue(bdev));
1454 static inline unsigned int queue_io_min(const struct request_queue *q)
1456 return q->limits.io_min;
1459 static inline int bdev_io_min(struct block_device *bdev)
1461 return queue_io_min(bdev_get_queue(bdev));
1464 static inline unsigned int queue_io_opt(const struct request_queue *q)
1466 return q->limits.io_opt;
1469 static inline int bdev_io_opt(struct block_device *bdev)
1471 return queue_io_opt(bdev_get_queue(bdev));
1474 static inline unsigned int
1475 queue_zone_write_granularity(const struct request_queue *q)
1477 return q->limits.zone_write_granularity;
1480 static inline unsigned int
1481 bdev_zone_write_granularity(struct block_device *bdev)
1483 return queue_zone_write_granularity(bdev_get_queue(bdev));
1486 static inline int queue_alignment_offset(const struct request_queue *q)
1488 if (q->limits.misaligned)
1491 return q->limits.alignment_offset;
1494 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1496 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1497 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1500 return (granularity + lim->alignment_offset - alignment) % granularity;
1503 static inline int bdev_alignment_offset(struct block_device *bdev)
1505 struct request_queue *q = bdev_get_queue(bdev);
1507 if (q->limits.misaligned)
1509 if (bdev_is_partition(bdev))
1510 return queue_limit_alignment_offset(&q->limits,
1511 bdev->bd_start_sect);
1512 return q->limits.alignment_offset;
1515 static inline int queue_discard_alignment(const struct request_queue *q)
1517 if (q->limits.discard_misaligned)
1520 return q->limits.discard_alignment;
1523 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1525 unsigned int alignment, granularity, offset;
1527 if (!lim->max_discard_sectors)
1530 /* Why are these in bytes, not sectors? */
1531 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1532 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1536 /* Offset of the partition start in 'granularity' sectors */
1537 offset = sector_div(sector, granularity);
1539 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1540 offset = (granularity + alignment - offset) % granularity;
1542 /* Turn it back into bytes, gaah */
1543 return offset << SECTOR_SHIFT;
1546 static inline int bdev_discard_alignment(struct block_device *bdev)
1548 struct request_queue *q = bdev_get_queue(bdev);
1550 if (bdev_is_partition(bdev))
1551 return queue_limit_discard_alignment(&q->limits,
1552 bdev->bd_start_sect);
1553 return q->limits.discard_alignment;
1556 static inline unsigned int bdev_write_same(struct block_device *bdev)
1558 struct request_queue *q = bdev_get_queue(bdev);
1561 return q->limits.max_write_same_sectors;
1566 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1568 struct request_queue *q = bdev_get_queue(bdev);
1571 return q->limits.max_write_zeroes_sectors;
1576 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1578 struct request_queue *q = bdev_get_queue(bdev);
1581 return blk_queue_zoned_model(q);
1583 return BLK_ZONED_NONE;
1586 static inline bool bdev_is_zoned(struct block_device *bdev)
1588 struct request_queue *q = bdev_get_queue(bdev);
1591 return blk_queue_is_zoned(q);
1596 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1598 struct request_queue *q = bdev_get_queue(bdev);
1601 return blk_queue_zone_sectors(q);
1605 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1607 struct request_queue *q = bdev_get_queue(bdev);
1610 return queue_max_open_zones(q);
1614 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1616 struct request_queue *q = bdev_get_queue(bdev);
1619 return queue_max_active_zones(q);
1623 static inline int queue_dma_alignment(const struct request_queue *q)
1625 return q ? q->dma_alignment : 511;
1628 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1631 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1632 return !(addr & alignment) && !(len & alignment);
1635 /* assumes size > 256 */
1636 static inline unsigned int blksize_bits(unsigned int size)
1638 unsigned int bits = 8;
1642 } while (size > 256);
1646 static inline unsigned int block_size(struct block_device *bdev)
1648 return 1 << bdev->bd_inode->i_blkbits;
1651 int kblockd_schedule_work(struct work_struct *work);
1652 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1654 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1655 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1656 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1657 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1659 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1661 enum blk_integrity_flags {
1662 BLK_INTEGRITY_VERIFY = 1 << 0,
1663 BLK_INTEGRITY_GENERATE = 1 << 1,
1664 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1665 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1668 struct blk_integrity_iter {
1672 unsigned int data_size;
1673 unsigned short interval;
1674 const char *disk_name;
1677 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1678 typedef void (integrity_prepare_fn) (struct request *);
1679 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1681 struct blk_integrity_profile {
1682 integrity_processing_fn *generate_fn;
1683 integrity_processing_fn *verify_fn;
1684 integrity_prepare_fn *prepare_fn;
1685 integrity_complete_fn *complete_fn;
1689 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1690 extern void blk_integrity_unregister(struct gendisk *);
1691 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1692 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1693 struct scatterlist *);
1694 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1696 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1698 struct blk_integrity *bi = &disk->queue->integrity;
1707 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1709 return blk_get_integrity(bdev->bd_disk);
1713 blk_integrity_queue_supports_integrity(struct request_queue *q)
1715 return q->integrity.profile;
1718 static inline bool blk_integrity_rq(struct request *rq)
1720 return rq->cmd_flags & REQ_INTEGRITY;
1723 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1726 q->limits.max_integrity_segments = segs;
1729 static inline unsigned short
1730 queue_max_integrity_segments(const struct request_queue *q)
1732 return q->limits.max_integrity_segments;
1736 * bio_integrity_intervals - Return number of integrity intervals for a bio
1737 * @bi: blk_integrity profile for device
1738 * @sectors: Size of the bio in 512-byte sectors
1740 * Description: The block layer calculates everything in 512 byte
1741 * sectors but integrity metadata is done in terms of the data integrity
1742 * interval size of the storage device. Convert the block layer sectors
1743 * to the appropriate number of integrity intervals.
1745 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1746 unsigned int sectors)
1748 return sectors >> (bi->interval_exp - 9);
1751 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1752 unsigned int sectors)
1754 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1758 * Return the first bvec that contains integrity data. Only drivers that are
1759 * limited to a single integrity segment should use this helper.
1761 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1763 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1765 return rq->bio->bi_integrity->bip_vec;
1768 #else /* CONFIG_BLK_DEV_INTEGRITY */
1771 struct block_device;
1773 struct blk_integrity;
1775 static inline int blk_integrity_rq(struct request *rq)
1779 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1784 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1786 struct scatterlist *s)
1790 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1794 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1799 blk_integrity_queue_supports_integrity(struct request_queue *q)
1803 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1807 static inline void blk_integrity_register(struct gendisk *d,
1808 struct blk_integrity *b)
1811 static inline void blk_integrity_unregister(struct gendisk *d)
1814 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1818 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1823 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1824 unsigned int sectors)
1829 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1830 unsigned int sectors)
1835 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1840 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1842 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1844 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1846 void blk_ksm_unregister(struct request_queue *q);
1848 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1850 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1851 struct request_queue *q)
1856 static inline void blk_ksm_unregister(struct request_queue *q) { }
1858 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1861 struct block_device_operations {
1862 blk_qc_t (*submit_bio) (struct bio *bio);
1863 int (*open) (struct block_device *, fmode_t);
1864 void (*release) (struct gendisk *, fmode_t);
1865 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1866 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1867 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1868 unsigned int (*check_events) (struct gendisk *disk,
1869 unsigned int clearing);
1870 void (*unlock_native_capacity) (struct gendisk *);
1871 int (*revalidate_disk) (struct gendisk *);
1872 int (*getgeo)(struct block_device *, struct hd_geometry *);
1873 int (*set_read_only)(struct block_device *bdev, bool ro);
1874 /* this callback is with swap_lock and sometimes page table lock held */
1875 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1876 int (*report_zones)(struct gendisk *, sector_t sector,
1877 unsigned int nr_zones, report_zones_cb cb, void *data);
1878 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1879 struct module *owner;
1880 const struct pr_ops *pr_ops;
1883 #ifdef CONFIG_COMPAT
1884 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1885 unsigned int, unsigned long);
1887 #define blkdev_compat_ptr_ioctl NULL
1890 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1891 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1892 struct writeback_control *);
1894 #ifdef CONFIG_BLK_DEV_ZONED
1895 bool blk_req_needs_zone_write_lock(struct request *rq);
1896 bool blk_req_zone_write_trylock(struct request *rq);
1897 void __blk_req_zone_write_lock(struct request *rq);
1898 void __blk_req_zone_write_unlock(struct request *rq);
1900 static inline void blk_req_zone_write_lock(struct request *rq)
1902 if (blk_req_needs_zone_write_lock(rq))
1903 __blk_req_zone_write_lock(rq);
1906 static inline void blk_req_zone_write_unlock(struct request *rq)
1908 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1909 __blk_req_zone_write_unlock(rq);
1912 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1914 return rq->q->seq_zones_wlock &&
1915 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1918 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1920 if (!blk_req_needs_zone_write_lock(rq))
1922 return !blk_req_zone_is_write_locked(rq);
1925 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1930 static inline void blk_req_zone_write_lock(struct request *rq)
1934 static inline void blk_req_zone_write_unlock(struct request *rq)
1937 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1942 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1946 #endif /* CONFIG_BLK_DEV_ZONED */
1948 static inline void blk_wake_io_task(struct task_struct *waiter)
1951 * If we're polling, the task itself is doing the completions. For
1952 * that case, we don't need to signal a wakeup, it's enough to just
1953 * mark us as RUNNING.
1955 if (waiter == current)
1956 __set_current_state(TASK_RUNNING);
1958 wake_up_process(waiter);
1961 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1963 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1964 unsigned long start_time);
1966 unsigned long bio_start_io_acct(struct bio *bio);
1967 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1968 struct block_device *orig_bdev);
1971 * bio_end_io_acct - end I/O accounting for bio based drivers
1972 * @bio: bio to end account for
1973 * @start: start time returned by bio_start_io_acct()
1975 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1977 return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1980 int bdev_read_only(struct block_device *bdev);
1981 int set_blocksize(struct block_device *bdev, int size);
1983 const char *bdevname(struct block_device *bdev, char *buffer);
1984 int lookup_bdev(const char *pathname, dev_t *dev);
1986 void blkdev_show(struct seq_file *seqf, off_t offset);
1988 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1989 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1991 #define BLKDEV_MAJOR_MAX 512
1993 #define BLKDEV_MAJOR_MAX 0
1996 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1998 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1999 int bd_prepare_to_claim(struct block_device *bdev, void *holder);
2000 void bd_abort_claiming(struct block_device *bdev, void *holder);
2001 void blkdev_put(struct block_device *bdev, fmode_t mode);
2003 /* just for blk-cgroup, don't use elsewhere */
2004 struct block_device *blkdev_get_no_open(dev_t dev);
2005 void blkdev_put_no_open(struct block_device *bdev);
2007 struct block_device *bdev_alloc(struct gendisk *disk, u8 partno);
2008 void bdev_add(struct block_device *bdev, dev_t dev);
2009 struct block_device *I_BDEV(struct inode *inode);
2010 struct block_device *bdgrab(struct block_device *bdev);
2011 void bdput(struct block_device *);
2012 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
2016 void invalidate_bdev(struct block_device *bdev);
2017 int sync_blockdev(struct block_device *bdev);
2019 static inline void invalidate_bdev(struct block_device *bdev)
2022 static inline int sync_blockdev(struct block_device *bdev)
2027 int fsync_bdev(struct block_device *bdev);
2029 int freeze_bdev(struct block_device *bdev);
2030 int thaw_bdev(struct block_device *bdev);
2032 #endif /* _LINUX_BLKDEV_H */