1 /* SPDX-License-Identifier: GPL-2.0 */
3 * Portions Copyright (C) 1992 Drew Eckhardt
5 #ifndef _LINUX_BLKDEV_H
6 #define _LINUX_BLKDEV_H
8 #include <linux/types.h>
9 #include <linux/blk_types.h>
10 #include <linux/device.h>
11 #include <linux/list.h>
12 #include <linux/llist.h>
13 #include <linux/minmax.h>
14 #include <linux/timer.h>
15 #include <linux/workqueue.h>
16 #include <linux/wait.h>
17 #include <linux/bio.h>
18 #include <linux/gfp.h>
19 #include <linux/kdev_t.h>
20 #include <linux/rcupdate.h>
21 #include <linux/percpu-refcount.h>
22 #include <linux/blkzoned.h>
23 #include <linux/sched.h>
24 #include <linux/sbitmap.h>
25 #include <linux/uuid.h>
26 #include <linux/xarray.h>
27 #include <linux/file.h>
28 #include <linux/lockdep.h>
32 struct elevator_queue;
37 struct blk_flush_queue;
41 struct blk_queue_stats;
42 struct blk_stat_callback;
43 struct blk_crypto_profile;
45 extern const struct device_type disk_type;
46 extern const struct device_type part_type;
47 extern const struct class block_class;
50 * Maximum number of blkcg policies allowed to be registered concurrently.
51 * Defined here to simplify include dependency.
53 #define BLKCG_MAX_POLS 6
55 #define DISK_MAX_PARTS 256
56 #define DISK_NAME_LEN 32
58 #define PARTITION_META_INFO_VOLNAMELTH 64
60 * Enough for the string representation of any kind of UUID plus NULL.
61 * EFI UUID is 36 characters. MSDOS UUID is 11 characters.
63 #define PARTITION_META_INFO_UUIDLTH (UUID_STRING_LEN + 1)
65 struct partition_meta_info {
66 char uuid[PARTITION_META_INFO_UUIDLTH];
67 u8 volname[PARTITION_META_INFO_VOLNAMELTH];
71 * DOC: genhd capability flags
73 * ``GENHD_FL_REMOVABLE``: indicates that the block device gives access to
74 * removable media. When set, the device remains present even when media is not
75 * inserted. Shall not be set for devices which are removed entirely when the
78 * ``GENHD_FL_HIDDEN``: the block device is hidden; it doesn't produce events,
79 * doesn't appear in sysfs, and can't be opened from userspace or using
80 * blkdev_get*. Used for the underlying components of multipath devices.
82 * ``GENHD_FL_NO_PART``: partition support is disabled. The kernel will not
83 * scan for partitions from add_disk, and users can't add partitions manually.
87 GENHD_FL_REMOVABLE = 1 << 0,
88 GENHD_FL_HIDDEN = 1 << 1,
89 GENHD_FL_NO_PART = 1 << 2,
93 DISK_EVENT_MEDIA_CHANGE = 1 << 0, /* media changed */
94 DISK_EVENT_EJECT_REQUEST = 1 << 1, /* eject requested */
98 /* Poll even if events_poll_msecs is unset */
99 DISK_EVENT_FLAG_POLL = 1 << 0,
100 /* Forward events to udev */
101 DISK_EVENT_FLAG_UEVENT = 1 << 1,
102 /* Block event polling when open for exclusive write */
103 DISK_EVENT_FLAG_BLOCK_ON_EXCL_WRITE = 1 << 2,
109 enum blk_integrity_checksum {
110 BLK_INTEGRITY_CSUM_NONE = 0,
111 BLK_INTEGRITY_CSUM_IP = 1,
112 BLK_INTEGRITY_CSUM_CRC = 2,
113 BLK_INTEGRITY_CSUM_CRC64 = 3,
116 struct blk_integrity {
118 enum blk_integrity_checksum csum_type;
119 unsigned char tuple_size;
120 unsigned char pi_offset;
121 unsigned char interval_exp;
122 unsigned char tag_size;
125 typedef unsigned int __bitwise blk_mode_t;
127 /* open for reading */
128 #define BLK_OPEN_READ ((__force blk_mode_t)(1 << 0))
129 /* open for writing */
130 #define BLK_OPEN_WRITE ((__force blk_mode_t)(1 << 1))
131 /* open exclusively (vs other exclusive openers */
132 #define BLK_OPEN_EXCL ((__force blk_mode_t)(1 << 2))
133 /* opened with O_NDELAY */
134 #define BLK_OPEN_NDELAY ((__force blk_mode_t)(1 << 3))
135 /* open for "writes" only for ioctls (specialy hack for floppy.c) */
136 #define BLK_OPEN_WRITE_IOCTL ((__force blk_mode_t)(1 << 4))
137 /* open is exclusive wrt all other BLK_OPEN_WRITE opens to the device */
138 #define BLK_OPEN_RESTRICT_WRITES ((__force blk_mode_t)(1 << 5))
139 /* return partition scanning errors */
140 #define BLK_OPEN_STRICT_SCAN ((__force blk_mode_t)(1 << 6))
144 * major/first_minor/minors should not be set by any new driver, the
145 * block core will take care of allocating them automatically.
151 char disk_name[DISK_NAME_LEN]; /* name of major driver */
153 unsigned short events; /* supported events */
154 unsigned short event_flags; /* flags related to event processing */
156 struct xarray part_tbl;
157 struct block_device *part0;
159 const struct block_device_operations *fops;
160 struct request_queue *queue;
163 struct bio_set bio_split;
167 #define GD_NEED_PART_SCAN 0
168 #define GD_READ_ONLY 1
170 #define GD_NATIVE_CAPACITY 3
172 #define GD_SUPPRESS_PART_SCAN 5
173 #define GD_OWNS_QUEUE 6
175 struct mutex open_mutex; /* open/close mutex */
176 unsigned open_partitions; /* number of open partitions */
178 struct backing_dev_info *bdi;
179 struct kobject queue_kobj; /* the queue/ directory */
180 struct kobject *slave_dir;
181 #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED
182 struct list_head slave_bdevs;
184 struct timer_rand_state *random;
185 atomic_t sync_io; /* RAID */
186 struct disk_events *ev;
188 #ifdef CONFIG_BLK_DEV_ZONED
190 * Zoned block device information. Reads of this information must be
191 * protected with blk_queue_enter() / blk_queue_exit(). Modifying this
192 * information is only allowed while no requests are being processed.
193 * See also blk_mq_freeze_queue() and blk_mq_unfreeze_queue().
195 unsigned int nr_zones;
196 unsigned int zone_capacity;
197 unsigned int last_zone_capacity;
198 unsigned long __rcu *conv_zones_bitmap;
199 unsigned int zone_wplugs_hash_bits;
200 spinlock_t zone_wplugs_lock;
201 struct mempool_s *zone_wplugs_pool;
202 struct hlist_head *zone_wplugs_hash;
203 struct workqueue_struct *zone_wplugs_wq;
204 #endif /* CONFIG_BLK_DEV_ZONED */
206 #if IS_ENABLED(CONFIG_CDROM)
207 struct cdrom_device_info *cdi;
210 struct badblocks *bb;
211 struct lockdep_map lockdep_map;
213 blk_mode_t open_mode;
216 * Independent sector access ranges. This is always NULL for
217 * devices that do not have multiple independent access ranges.
219 struct blk_independent_access_ranges *ia_ranges;
223 * disk_openers - returns how many openers are there for a disk
224 * @disk: disk to check
226 * This returns the number of openers for a disk. Note that this value is only
227 * stable if disk->open_mutex is held.
229 * Note: Due to a quirk in the block layer open code, each open partition is
230 * only counted once even if there are multiple openers.
232 static inline unsigned int disk_openers(struct gendisk *disk)
234 return atomic_read(&disk->part0->bd_openers);
238 * disk_has_partscan - return %true if partition scanning is enabled on a disk
239 * @disk: disk to check
241 * Returns %true if partitions scanning is enabled for @disk, or %false if
242 * partition scanning is disabled either permanently or temporarily.
244 static inline bool disk_has_partscan(struct gendisk *disk)
246 return !(disk->flags & (GENHD_FL_NO_PART | GENHD_FL_HIDDEN)) &&
247 !test_bit(GD_SUPPRESS_PART_SCAN, &disk->state);
251 * The gendisk is refcounted by the part0 block_device, and the bd_device
252 * therein is also used for device model presentation in sysfs.
254 #define dev_to_disk(device) \
255 (dev_to_bdev(device)->bd_disk)
256 #define disk_to_dev(disk) \
257 (&((disk)->part0->bd_device))
259 #if IS_REACHABLE(CONFIG_CDROM)
260 #define disk_to_cdi(disk) ((disk)->cdi)
262 #define disk_to_cdi(disk) NULL
265 static inline dev_t disk_devt(struct gendisk *disk)
267 return MKDEV(disk->major, disk->first_minor);
270 /* blk_validate_limits() validates bsize, so drivers don't usually need to */
271 static inline int blk_validate_block_size(unsigned long bsize)
273 if (bsize < 512 || bsize > PAGE_SIZE || !is_power_of_2(bsize))
279 static inline bool blk_op_is_passthrough(blk_opf_t op)
282 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
285 /* flags set by the driver in queue_limits.features */
286 typedef unsigned int __bitwise blk_features_t;
288 /* supports a volatile write cache */
289 #define BLK_FEAT_WRITE_CACHE ((__force blk_features_t)(1u << 0))
291 /* supports passing on the FUA bit */
292 #define BLK_FEAT_FUA ((__force blk_features_t)(1u << 1))
294 /* rotational device (hard drive or floppy) */
295 #define BLK_FEAT_ROTATIONAL ((__force blk_features_t)(1u << 2))
297 /* contributes to the random number pool */
298 #define BLK_FEAT_ADD_RANDOM ((__force blk_features_t)(1u << 3))
300 /* do disk/partitions IO accounting */
301 #define BLK_FEAT_IO_STAT ((__force blk_features_t)(1u << 4))
303 /* don't modify data until writeback is done */
304 #define BLK_FEAT_STABLE_WRITES ((__force blk_features_t)(1u << 5))
306 /* always completes in submit context */
307 #define BLK_FEAT_SYNCHRONOUS ((__force blk_features_t)(1u << 6))
309 /* supports REQ_NOWAIT */
310 #define BLK_FEAT_NOWAIT ((__force blk_features_t)(1u << 7))
313 #define BLK_FEAT_DAX ((__force blk_features_t)(1u << 8))
315 /* supports I/O polling */
316 #define BLK_FEAT_POLL ((__force blk_features_t)(1u << 9))
318 /* is a zoned device */
319 #define BLK_FEAT_ZONED ((__force blk_features_t)(1u << 10))
321 /* supports PCI(e) p2p requests */
322 #define BLK_FEAT_PCI_P2PDMA ((__force blk_features_t)(1u << 12))
324 /* skip this queue in blk_mq_(un)quiesce_tagset */
325 #define BLK_FEAT_SKIP_TAGSET_QUIESCE ((__force blk_features_t)(1u << 13))
327 /* bounce all highmem pages */
328 #define BLK_FEAT_BOUNCE_HIGH ((__force blk_features_t)(1u << 14))
330 /* undocumented magic for bcache */
331 #define BLK_FEAT_RAID_PARTIAL_STRIPES_EXPENSIVE \
332 ((__force blk_features_t)(1u << 15))
334 /* stacked device can/does support atomic writes */
335 #define BLK_FEAT_ATOMIC_WRITES_STACKED \
336 ((__force blk_features_t)(1u << 16))
339 * Flags automatically inherited when stacking limits.
341 #define BLK_FEAT_INHERIT_MASK \
342 (BLK_FEAT_WRITE_CACHE | BLK_FEAT_FUA | BLK_FEAT_ROTATIONAL | \
343 BLK_FEAT_STABLE_WRITES | BLK_FEAT_ZONED | BLK_FEAT_BOUNCE_HIGH | \
344 BLK_FEAT_RAID_PARTIAL_STRIPES_EXPENSIVE)
346 /* internal flags in queue_limits.flags */
347 typedef unsigned int __bitwise blk_flags_t;
349 /* do not send FLUSH/FUA commands despite advertising a write cache */
350 #define BLK_FLAG_WRITE_CACHE_DISABLED ((__force blk_flags_t)(1u << 0))
352 /* I/O topology is misaligned */
353 #define BLK_FLAG_MISALIGNED ((__force blk_flags_t)(1u << 1))
355 /* passthrough command IO accounting */
356 #define BLK_FLAG_IOSTATS_PASSTHROUGH ((__force blk_flags_t)(1u << 2))
358 struct queue_limits {
359 blk_features_t features;
361 unsigned long seg_boundary_mask;
362 unsigned long virt_boundary_mask;
364 unsigned int max_hw_sectors;
365 unsigned int max_dev_sectors;
366 unsigned int chunk_sectors;
367 unsigned int max_sectors;
368 unsigned int max_user_sectors;
369 unsigned int max_segment_size;
370 unsigned int physical_block_size;
371 unsigned int logical_block_size;
372 unsigned int alignment_offset;
375 unsigned int max_discard_sectors;
376 unsigned int max_hw_discard_sectors;
377 unsigned int max_user_discard_sectors;
378 unsigned int max_secure_erase_sectors;
379 unsigned int max_write_zeroes_sectors;
380 unsigned int max_hw_zone_append_sectors;
381 unsigned int max_zone_append_sectors;
382 unsigned int discard_granularity;
383 unsigned int discard_alignment;
384 unsigned int zone_write_granularity;
386 /* atomic write limits */
387 unsigned int atomic_write_hw_max;
388 unsigned int atomic_write_max_sectors;
389 unsigned int atomic_write_hw_boundary;
390 unsigned int atomic_write_boundary_sectors;
391 unsigned int atomic_write_hw_unit_min;
392 unsigned int atomic_write_unit_min;
393 unsigned int atomic_write_hw_unit_max;
394 unsigned int atomic_write_unit_max;
396 unsigned short max_segments;
397 unsigned short max_integrity_segments;
398 unsigned short max_discard_segments;
400 unsigned int max_open_zones;
401 unsigned int max_active_zones;
404 * Drivers that set dma_alignment to less than 511 must be prepared to
405 * handle individual bvec's that are not a multiple of a SECTOR_SIZE
406 * due to possible offsets.
408 unsigned int dma_alignment;
409 unsigned int dma_pad_mask;
411 struct blk_integrity integrity;
414 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
417 #define BLK_ALL_ZONES ((unsigned int)-1)
418 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
419 unsigned int nr_zones, report_zones_cb cb, void *data);
420 int blkdev_zone_mgmt(struct block_device *bdev, enum req_op op,
421 sector_t sectors, sector_t nr_sectors);
422 int blk_revalidate_disk_zones(struct gendisk *disk);
425 * Independent access ranges: struct blk_independent_access_range describes
426 * a range of contiguous sectors that can be accessed using device command
427 * execution resources that are independent from the resources used for
428 * other access ranges. This is typically found with single-LUN multi-actuator
429 * HDDs where each access range is served by a different set of heads.
430 * The set of independent ranges supported by the device is defined using
431 * struct blk_independent_access_ranges. The independent ranges must not overlap
432 * and must include all sectors within the disk capacity (no sector holes
434 * For a device with multiple ranges, requests targeting sectors in different
435 * ranges can be executed in parallel. A request can straddle an access range
438 struct blk_independent_access_range {
444 struct blk_independent_access_ranges {
446 bool sysfs_registered;
447 unsigned int nr_ia_ranges;
448 struct blk_independent_access_range ia_range[];
451 struct request_queue {
453 * The queue owner gets to use this for whatever they like.
454 * ll_rw_blk doesn't touch it.
458 struct elevator_queue *elevator;
460 const struct blk_mq_ops *mq_ops;
463 struct blk_mq_ctx __percpu *queue_ctx;
466 * various queue flags, see QUEUE_* below
468 unsigned long queue_flags;
470 unsigned int rq_timeout;
472 unsigned int queue_depth;
476 /* hw dispatch queues */
477 unsigned int nr_hw_queues;
478 struct xarray hctx_table;
480 struct percpu_ref q_usage_counter;
481 struct lock_class_key io_lock_cls_key;
482 struct lockdep_map io_lockdep_map;
484 struct lock_class_key q_lock_cls_key;
485 struct lockdep_map q_lockdep_map;
487 struct request *last_merge;
489 spinlock_t queue_lock;
493 struct gendisk *disk;
498 struct kobject *mq_kobj;
500 struct queue_limits limits;
504 enum rpm_status rpm_status;
508 * Number of contexts that have called blk_set_pm_only(). If this
509 * counter is above zero then only RQF_PM requests are processed.
513 struct blk_queue_stats *stats;
514 struct rq_qos *rq_qos;
515 struct mutex rq_qos_mutex;
518 * ida allocated id for this queue. Used to index queues from
526 unsigned long nr_requests; /* Max # of requests */
528 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
529 struct blk_crypto_profile *crypto_profile;
530 struct kobject *crypto_kobject;
533 struct timer_list timeout;
534 struct work_struct timeout_work;
536 atomic_t nr_active_requests_shared_tags;
538 struct blk_mq_tags *sched_shared_tags;
540 struct list_head icq_list;
541 #ifdef CONFIG_BLK_CGROUP
542 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
543 struct blkcg_gq *root_blkg;
544 struct list_head blkg_list;
545 struct mutex blkcg_mutex;
550 spinlock_t requeue_lock;
551 struct list_head requeue_list;
552 struct delayed_work requeue_work;
554 #ifdef CONFIG_BLK_DEV_IO_TRACE
555 struct blk_trace __rcu *blk_trace;
558 * for flush operations
560 struct blk_flush_queue *fq;
561 struct list_head flush_list;
563 struct mutex sysfs_lock;
564 struct mutex sysfs_dir_lock;
565 struct mutex limits_lock;
568 * for reusing dead hctx instance in case of updating
571 struct list_head unused_hctx_list;
572 spinlock_t unused_hctx_lock;
576 #ifdef CONFIG_BLK_DEV_THROTTLING
578 struct throtl_data *td;
580 struct rcu_head rcu_head;
581 #ifdef CONFIG_LOCKDEP
582 struct task_struct *mq_freeze_owner;
583 int mq_freeze_owner_depth;
585 wait_queue_head_t mq_freeze_wq;
587 * Protect concurrent access to q_usage_counter by
588 * percpu_ref_kill() and percpu_ref_reinit().
590 struct mutex mq_freeze_lock;
592 struct blk_mq_tag_set *tag_set;
593 struct list_head tag_set_list;
595 struct dentry *debugfs_dir;
596 struct dentry *sched_debugfs_dir;
597 struct dentry *rqos_debugfs_dir;
599 * Serializes all debugfs metadata operations using the above dentries.
601 struct mutex debugfs_mutex;
603 bool mq_sysfs_init_done;
606 /* Keep blk_queue_flag_name[] in sync with the definitions below */
608 QUEUE_FLAG_DYING, /* queue being torn down */
609 QUEUE_FLAG_NOMERGES, /* disable merge attempts */
610 QUEUE_FLAG_SAME_COMP, /* complete on same CPU-group */
611 QUEUE_FLAG_FAIL_IO, /* fake timeout */
612 QUEUE_FLAG_NOXMERGES, /* No extended merges */
613 QUEUE_FLAG_SAME_FORCE, /* force complete on same CPU */
614 QUEUE_FLAG_INIT_DONE, /* queue is initialized */
615 QUEUE_FLAG_STATS, /* track IO start and completion times */
616 QUEUE_FLAG_REGISTERED, /* queue has been registered to a disk */
617 QUEUE_FLAG_QUIESCED, /* queue has been quiesced */
618 QUEUE_FLAG_RQ_ALLOC_TIME, /* record rq->alloc_time_ns */
619 QUEUE_FLAG_HCTX_ACTIVE, /* at least one blk-mq hctx is active */
620 QUEUE_FLAG_SQ_SCHED, /* single queue style io dispatch */
624 #define QUEUE_FLAG_MQ_DEFAULT (1UL << QUEUE_FLAG_SAME_COMP)
626 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
627 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
629 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
630 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
631 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
632 #define blk_queue_noxmerges(q) \
633 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
634 #define blk_queue_nonrot(q) (!((q)->limits.features & BLK_FEAT_ROTATIONAL))
635 #define blk_queue_io_stat(q) ((q)->limits.features & BLK_FEAT_IO_STAT)
636 #define blk_queue_passthrough_stat(q) \
637 ((q)->limits.flags & BLK_FLAG_IOSTATS_PASSTHROUGH)
638 #define blk_queue_dax(q) ((q)->limits.features & BLK_FEAT_DAX)
639 #define blk_queue_pci_p2pdma(q) ((q)->limits.features & BLK_FEAT_PCI_P2PDMA)
640 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
641 #define blk_queue_rq_alloc_time(q) \
642 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
644 #define blk_queue_rq_alloc_time(q) false
647 #define blk_noretry_request(rq) \
648 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
649 REQ_FAILFAST_DRIVER))
650 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
651 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
652 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
653 #define blk_queue_sq_sched(q) test_bit(QUEUE_FLAG_SQ_SCHED, &(q)->queue_flags)
654 #define blk_queue_skip_tagset_quiesce(q) \
655 ((q)->limits.features & BLK_FEAT_SKIP_TAGSET_QUIESCE)
657 extern void blk_set_pm_only(struct request_queue *q);
658 extern void blk_clear_pm_only(struct request_queue *q);
660 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
662 #define dma_map_bvec(dev, bv, dir, attrs) \
663 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
666 static inline bool queue_is_mq(struct request_queue *q)
672 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
674 return q->rpm_status;
677 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
683 static inline bool blk_queue_is_zoned(struct request_queue *q)
685 return IS_ENABLED(CONFIG_BLK_DEV_ZONED) &&
686 (q->limits.features & BLK_FEAT_ZONED);
689 #ifdef CONFIG_BLK_DEV_ZONED
690 static inline unsigned int disk_nr_zones(struct gendisk *disk)
692 return disk->nr_zones;
694 bool blk_zone_plug_bio(struct bio *bio, unsigned int nr_segs);
695 #else /* CONFIG_BLK_DEV_ZONED */
696 static inline unsigned int disk_nr_zones(struct gendisk *disk)
700 static inline bool blk_zone_plug_bio(struct bio *bio, unsigned int nr_segs)
704 #endif /* CONFIG_BLK_DEV_ZONED */
706 static inline unsigned int disk_zone_no(struct gendisk *disk, sector_t sector)
708 if (!blk_queue_is_zoned(disk->queue))
710 return sector >> ilog2(disk->queue->limits.chunk_sectors);
713 static inline unsigned int bdev_nr_zones(struct block_device *bdev)
715 return disk_nr_zones(bdev->bd_disk);
718 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
720 return bdev->bd_disk->queue->limits.max_open_zones;
723 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
725 return bdev->bd_disk->queue->limits.max_active_zones;
728 static inline unsigned int blk_queue_depth(struct request_queue *q)
731 return q->queue_depth;
733 return q->nr_requests;
737 * default timeout for SG_IO if none specified
739 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
740 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
742 /* This should not be used directly - use rq_for_each_segment */
743 #define for_each_bio(_bio) \
744 for (; _bio; _bio = _bio->bi_next)
746 int __must_check add_disk_fwnode(struct device *parent, struct gendisk *disk,
747 const struct attribute_group **groups,
748 struct fwnode_handle *fwnode);
749 int __must_check device_add_disk(struct device *parent, struct gendisk *disk,
750 const struct attribute_group **groups);
751 static inline int __must_check add_disk(struct gendisk *disk)
753 return device_add_disk(NULL, disk, NULL);
755 void del_gendisk(struct gendisk *gp);
756 void invalidate_disk(struct gendisk *disk);
757 void set_disk_ro(struct gendisk *disk, bool read_only);
758 void disk_uevent(struct gendisk *disk, enum kobject_action action);
760 static inline u8 bdev_partno(const struct block_device *bdev)
762 return atomic_read(&bdev->__bd_flags) & BD_PARTNO;
765 static inline bool bdev_test_flag(const struct block_device *bdev, unsigned flag)
767 return atomic_read(&bdev->__bd_flags) & flag;
770 static inline void bdev_set_flag(struct block_device *bdev, unsigned flag)
772 atomic_or(flag, &bdev->__bd_flags);
775 static inline void bdev_clear_flag(struct block_device *bdev, unsigned flag)
777 atomic_andnot(flag, &bdev->__bd_flags);
780 static inline bool get_disk_ro(struct gendisk *disk)
782 return bdev_test_flag(disk->part0, BD_READ_ONLY) ||
783 test_bit(GD_READ_ONLY, &disk->state);
786 static inline bool bdev_read_only(struct block_device *bdev)
788 return bdev_test_flag(bdev, BD_READ_ONLY) || get_disk_ro(bdev->bd_disk);
791 bool set_capacity_and_notify(struct gendisk *disk, sector_t size);
792 void disk_force_media_change(struct gendisk *disk);
793 void bdev_mark_dead(struct block_device *bdev, bool surprise);
795 void add_disk_randomness(struct gendisk *disk) __latent_entropy;
796 void rand_initialize_disk(struct gendisk *disk);
798 static inline sector_t get_start_sect(struct block_device *bdev)
800 return bdev->bd_start_sect;
803 static inline sector_t bdev_nr_sectors(struct block_device *bdev)
805 return bdev->bd_nr_sectors;
808 static inline loff_t bdev_nr_bytes(struct block_device *bdev)
810 return (loff_t)bdev_nr_sectors(bdev) << SECTOR_SHIFT;
813 static inline sector_t get_capacity(struct gendisk *disk)
815 return bdev_nr_sectors(disk->part0);
818 static inline u64 sb_bdev_nr_blocks(struct super_block *sb)
820 return bdev_nr_sectors(sb->s_bdev) >>
821 (sb->s_blocksize_bits - SECTOR_SHIFT);
824 int bdev_disk_changed(struct gendisk *disk, bool invalidate);
826 void put_disk(struct gendisk *disk);
827 struct gendisk *__blk_alloc_disk(struct queue_limits *lim, int node,
828 struct lock_class_key *lkclass);
831 * blk_alloc_disk - allocate a gendisk structure
832 * @lim: queue limits to be used for this disk.
833 * @node_id: numa node to allocate on
835 * Allocate and pre-initialize a gendisk structure for use with BIO based
838 * Returns an ERR_PTR on error, else the allocated disk.
842 #define blk_alloc_disk(lim, node_id) \
844 static struct lock_class_key __key; \
846 __blk_alloc_disk(lim, node_id, &__key); \
849 int __register_blkdev(unsigned int major, const char *name,
850 void (*probe)(dev_t devt));
851 #define register_blkdev(major, name) \
852 __register_blkdev(major, name, NULL)
853 void unregister_blkdev(unsigned int major, const char *name);
855 bool disk_check_media_change(struct gendisk *disk);
856 void set_capacity(struct gendisk *disk, sector_t size);
858 #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED
859 int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
860 void bd_unlink_disk_holder(struct block_device *bdev, struct gendisk *disk);
862 static inline int bd_link_disk_holder(struct block_device *bdev,
863 struct gendisk *disk)
867 static inline void bd_unlink_disk_holder(struct block_device *bdev,
868 struct gendisk *disk)
871 #endif /* CONFIG_BLOCK_HOLDER_DEPRECATED */
873 dev_t part_devt(struct gendisk *disk, u8 partno);
874 void inc_diskseq(struct gendisk *disk);
875 void blk_request_module(dev_t devt);
877 extern int blk_register_queue(struct gendisk *disk);
878 extern void blk_unregister_queue(struct gendisk *disk);
879 void submit_bio_noacct(struct bio *bio);
880 struct bio *bio_split_to_limits(struct bio *bio);
882 extern int blk_lld_busy(struct request_queue *q);
883 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
884 extern void blk_queue_exit(struct request_queue *q);
885 extern void blk_sync_queue(struct request_queue *q);
887 /* Helper to convert REQ_OP_XXX to its string format XXX */
888 extern const char *blk_op_str(enum req_op op);
890 int blk_status_to_errno(blk_status_t status);
891 blk_status_t errno_to_blk_status(int errno);
892 const char *blk_status_to_str(blk_status_t status);
894 /* only poll the hardware once, don't continue until a completion was found */
895 #define BLK_POLL_ONESHOT (1 << 0)
896 int bio_poll(struct bio *bio, struct io_comp_batch *iob, unsigned int flags);
897 int iocb_bio_iopoll(struct kiocb *kiocb, struct io_comp_batch *iob,
900 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
902 return bdev->bd_queue; /* this is never NULL */
905 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
906 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
908 static inline unsigned int bio_zone_no(struct bio *bio)
910 return disk_zone_no(bio->bi_bdev->bd_disk, bio->bi_iter.bi_sector);
913 static inline bool bio_straddles_zones(struct bio *bio)
915 return bio_sectors(bio) &&
917 disk_zone_no(bio->bi_bdev->bd_disk, bio_end_sector(bio) - 1);
921 * Return how much within the boundary is left to be used for I/O at a given
924 static inline unsigned int blk_boundary_sectors_left(sector_t offset,
925 unsigned int boundary_sectors)
927 if (unlikely(!is_power_of_2(boundary_sectors)))
928 return boundary_sectors - sector_div(offset, boundary_sectors);
929 return boundary_sectors - (offset & (boundary_sectors - 1));
933 * queue_limits_start_update - start an atomic update of queue limits
934 * @q: queue to update
936 * This functions starts an atomic update of the queue limits. It takes a lock
937 * to prevent other updates and returns a snapshot of the current limits that
938 * the caller can modify. The caller must call queue_limits_commit_update()
939 * to finish the update.
941 * Context: process context. The caller must have frozen the queue or ensured
942 * that there is outstanding I/O by other means.
944 static inline struct queue_limits
945 queue_limits_start_update(struct request_queue *q)
947 mutex_lock(&q->limits_lock);
950 int queue_limits_commit_update(struct request_queue *q,
951 struct queue_limits *lim);
952 int queue_limits_set(struct request_queue *q, struct queue_limits *lim);
953 int blk_validate_limits(struct queue_limits *lim);
956 * queue_limits_cancel_update - cancel an atomic update of queue limits
957 * @q: queue to update
959 * This functions cancels an atomic update of the queue limits started by
960 * queue_limits_start_update() and should be used when an error occurs after
963 static inline void queue_limits_cancel_update(struct request_queue *q)
965 mutex_unlock(&q->limits_lock);
969 * These helpers are for drivers that have sloppy feature negotiation and might
970 * have to disable DISCARD, WRITE_ZEROES or SECURE_DISCARD from the I/O
971 * completion handler when the device returned an indicator that the respective
972 * feature is not actually supported. They are racy and the driver needs to
973 * cope with that. Try to avoid this scheme if you can.
975 static inline void blk_queue_disable_discard(struct request_queue *q)
977 q->limits.max_discard_sectors = 0;
980 static inline void blk_queue_disable_secure_erase(struct request_queue *q)
982 q->limits.max_secure_erase_sectors = 0;
985 static inline void blk_queue_disable_write_zeroes(struct request_queue *q)
987 q->limits.max_write_zeroes_sectors = 0;
991 * Access functions for manipulating queue properties
993 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
994 extern void blk_set_stacking_limits(struct queue_limits *lim);
995 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
997 void queue_limits_stack_bdev(struct queue_limits *t, struct block_device *bdev,
998 sector_t offset, const char *pfx);
999 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1001 struct blk_independent_access_ranges *
1002 disk_alloc_independent_access_ranges(struct gendisk *disk, int nr_ia_ranges);
1003 void disk_set_independent_access_ranges(struct gendisk *disk,
1004 struct blk_independent_access_ranges *iars);
1006 bool __must_check blk_get_queue(struct request_queue *);
1007 extern void blk_put_queue(struct request_queue *);
1009 void blk_mark_disk_dead(struct gendisk *disk);
1012 struct request *head;
1013 struct request *tail;
1018 * blk_plug permits building a queue of related requests by holding the I/O
1019 * fragments for a short period. This allows merging of sequential requests
1020 * into single larger request. As the requests are moved from a per-task list to
1021 * the device's request_queue in a batch, this results in improved scalability
1022 * as the lock contention for request_queue lock is reduced.
1024 * It is ok not to disable preemption when adding the request to the plug list
1025 * or when attempting a merge. For details, please see schedule() where
1026 * blk_flush_plug() is called.
1029 struct rq_list mq_list; /* blk-mq requests */
1031 /* if ios_left is > 1, we can batch tag/rq allocations */
1032 struct rq_list cached_rqs;
1034 unsigned short nr_ios;
1036 unsigned short rq_count;
1038 bool multiple_queues;
1041 struct list_head cb_list; /* md requires an unplug callback */
1045 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1046 struct blk_plug_cb {
1047 struct list_head list;
1048 blk_plug_cb_fn callback;
1051 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1052 void *data, int size);
1053 extern void blk_start_plug(struct blk_plug *);
1054 extern void blk_start_plug_nr_ios(struct blk_plug *, unsigned short);
1055 extern void blk_finish_plug(struct blk_plug *);
1057 void __blk_flush_plug(struct blk_plug *plug, bool from_schedule);
1058 static inline void blk_flush_plug(struct blk_plug *plug, bool async)
1061 __blk_flush_plug(plug, async);
1065 * tsk == current here
1067 static inline void blk_plug_invalidate_ts(struct task_struct *tsk)
1069 struct blk_plug *plug = tsk->plug;
1072 plug->cur_ktime = 0;
1073 current->flags &= ~PF_BLOCK_TS;
1076 int blkdev_issue_flush(struct block_device *bdev);
1077 long nr_blockdev_pages(void);
1078 #else /* CONFIG_BLOCK */
1082 static inline void blk_start_plug_nr_ios(struct blk_plug *plug,
1083 unsigned short nr_ios)
1087 static inline void blk_start_plug(struct blk_plug *plug)
1091 static inline void blk_finish_plug(struct blk_plug *plug)
1095 static inline void blk_flush_plug(struct blk_plug *plug, bool async)
1099 static inline void blk_plug_invalidate_ts(struct task_struct *tsk)
1103 static inline int blkdev_issue_flush(struct block_device *bdev)
1108 static inline long nr_blockdev_pages(void)
1112 #endif /* CONFIG_BLOCK */
1114 extern void blk_io_schedule(void);
1116 int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1117 sector_t nr_sects, gfp_t gfp_mask);
1118 int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1119 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop);
1120 int blkdev_issue_secure_erase(struct block_device *bdev, sector_t sector,
1121 sector_t nr_sects, gfp_t gfp);
1123 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1124 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1125 #define BLKDEV_ZERO_KILLABLE (1 << 2) /* interruptible by fatal signals */
1127 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1128 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1130 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1131 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1133 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1134 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1136 return blkdev_issue_discard(sb->s_bdev,
1137 block << (sb->s_blocksize_bits -
1139 nr_blocks << (sb->s_blocksize_bits -
1143 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1144 sector_t nr_blocks, gfp_t gfp_mask)
1146 return blkdev_issue_zeroout(sb->s_bdev,
1147 block << (sb->s_blocksize_bits -
1149 nr_blocks << (sb->s_blocksize_bits -
1154 static inline bool bdev_is_partition(struct block_device *bdev)
1156 return bdev_partno(bdev) != 0;
1159 enum blk_default_limits {
1160 BLK_MAX_SEGMENTS = 128,
1161 BLK_SAFE_MAX_SECTORS = 255,
1162 BLK_MAX_SEGMENT_SIZE = 65536,
1163 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1167 * Default upper limit for the software max_sectors limit used for
1168 * regular file system I/O. This can be increased through sysfs.
1170 * Not to be confused with the max_hw_sector limit that is entirely
1171 * controlled by the driver, usually based on hardware limits.
1173 #define BLK_DEF_MAX_SECTORS_CAP 2560u
1175 static inline struct queue_limits *bdev_limits(struct block_device *bdev)
1177 return &bdev_get_queue(bdev)->limits;
1180 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1182 return q->limits.seg_boundary_mask;
1185 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1187 return q->limits.virt_boundary_mask;
1190 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1192 return q->limits.max_sectors;
1195 static inline unsigned int queue_max_bytes(struct request_queue *q)
1197 return min_t(unsigned int, queue_max_sectors(q), INT_MAX >> 9) << 9;
1200 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1202 return q->limits.max_hw_sectors;
1205 static inline unsigned short queue_max_segments(const struct request_queue *q)
1207 return q->limits.max_segments;
1210 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1212 return q->limits.max_discard_segments;
1215 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1217 return q->limits.max_segment_size;
1220 static inline bool queue_emulates_zone_append(struct request_queue *q)
1222 return blk_queue_is_zoned(q) && !q->limits.max_hw_zone_append_sectors;
1225 static inline bool bdev_emulates_zone_append(struct block_device *bdev)
1227 return queue_emulates_zone_append(bdev_get_queue(bdev));
1230 static inline unsigned int
1231 bdev_max_zone_append_sectors(struct block_device *bdev)
1233 return bdev_limits(bdev)->max_zone_append_sectors;
1236 static inline unsigned int bdev_max_segments(struct block_device *bdev)
1238 return queue_max_segments(bdev_get_queue(bdev));
1241 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1243 return q->limits.logical_block_size;
1246 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1248 return queue_logical_block_size(bdev_get_queue(bdev));
1251 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1253 return q->limits.physical_block_size;
1256 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1258 return queue_physical_block_size(bdev_get_queue(bdev));
1261 static inline unsigned int queue_io_min(const struct request_queue *q)
1263 return q->limits.io_min;
1266 static inline unsigned int bdev_io_min(struct block_device *bdev)
1268 return queue_io_min(bdev_get_queue(bdev));
1271 static inline unsigned int queue_io_opt(const struct request_queue *q)
1273 return q->limits.io_opt;
1276 static inline unsigned int bdev_io_opt(struct block_device *bdev)
1278 return queue_io_opt(bdev_get_queue(bdev));
1281 static inline unsigned int
1282 queue_zone_write_granularity(const struct request_queue *q)
1284 return q->limits.zone_write_granularity;
1287 static inline unsigned int
1288 bdev_zone_write_granularity(struct block_device *bdev)
1290 return queue_zone_write_granularity(bdev_get_queue(bdev));
1293 int bdev_alignment_offset(struct block_device *bdev);
1294 unsigned int bdev_discard_alignment(struct block_device *bdev);
1296 static inline unsigned int bdev_max_discard_sectors(struct block_device *bdev)
1298 return bdev_limits(bdev)->max_discard_sectors;
1301 static inline unsigned int bdev_discard_granularity(struct block_device *bdev)
1303 return bdev_limits(bdev)->discard_granularity;
1306 static inline unsigned int
1307 bdev_max_secure_erase_sectors(struct block_device *bdev)
1309 return bdev_limits(bdev)->max_secure_erase_sectors;
1312 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1314 return bdev_limits(bdev)->max_write_zeroes_sectors;
1317 static inline bool bdev_nonrot(struct block_device *bdev)
1319 return blk_queue_nonrot(bdev_get_queue(bdev));
1322 static inline bool bdev_synchronous(struct block_device *bdev)
1324 return bdev->bd_disk->queue->limits.features & BLK_FEAT_SYNCHRONOUS;
1327 static inline bool bdev_stable_writes(struct block_device *bdev)
1329 struct request_queue *q = bdev_get_queue(bdev);
1331 if (IS_ENABLED(CONFIG_BLK_DEV_INTEGRITY) &&
1332 q->limits.integrity.csum_type != BLK_INTEGRITY_CSUM_NONE)
1334 return q->limits.features & BLK_FEAT_STABLE_WRITES;
1337 static inline bool blk_queue_write_cache(struct request_queue *q)
1339 return (q->limits.features & BLK_FEAT_WRITE_CACHE) &&
1340 !(q->limits.flags & BLK_FLAG_WRITE_CACHE_DISABLED);
1343 static inline bool bdev_write_cache(struct block_device *bdev)
1345 return blk_queue_write_cache(bdev_get_queue(bdev));
1348 static inline bool bdev_fua(struct block_device *bdev)
1350 return bdev_limits(bdev)->features & BLK_FEAT_FUA;
1353 static inline bool bdev_nowait(struct block_device *bdev)
1355 return bdev->bd_disk->queue->limits.features & BLK_FEAT_NOWAIT;
1358 static inline bool bdev_is_zoned(struct block_device *bdev)
1360 return blk_queue_is_zoned(bdev_get_queue(bdev));
1363 static inline unsigned int bdev_zone_no(struct block_device *bdev, sector_t sec)
1365 return disk_zone_no(bdev->bd_disk, sec);
1368 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1370 struct request_queue *q = bdev_get_queue(bdev);
1372 if (!blk_queue_is_zoned(q))
1374 return q->limits.chunk_sectors;
1377 static inline sector_t bdev_offset_from_zone_start(struct block_device *bdev,
1380 return sector & (bdev_zone_sectors(bdev) - 1);
1383 static inline sector_t bio_offset_from_zone_start(struct bio *bio)
1385 return bdev_offset_from_zone_start(bio->bi_bdev,
1386 bio->bi_iter.bi_sector);
1389 static inline bool bdev_is_zone_start(struct block_device *bdev,
1392 return bdev_offset_from_zone_start(bdev, sector) == 0;
1396 * bdev_zone_is_seq - check if a sector belongs to a sequential write zone
1397 * @bdev: block device to check
1398 * @sector: sector number
1400 * Check if @sector on @bdev is contained in a sequential write required zone.
1402 static inline bool bdev_zone_is_seq(struct block_device *bdev, sector_t sector)
1404 bool is_seq = false;
1406 #if IS_ENABLED(CONFIG_BLK_DEV_ZONED)
1407 if (bdev_is_zoned(bdev)) {
1408 struct gendisk *disk = bdev->bd_disk;
1409 unsigned long *bitmap;
1412 bitmap = rcu_dereference(disk->conv_zones_bitmap);
1414 !test_bit(disk_zone_no(disk, sector), bitmap);
1422 int blk_zone_issue_zeroout(struct block_device *bdev, sector_t sector,
1423 sector_t nr_sects, gfp_t gfp_mask);
1425 static inline unsigned int queue_dma_alignment(const struct request_queue *q)
1427 return q->limits.dma_alignment;
1430 static inline unsigned int
1431 queue_atomic_write_unit_max_bytes(const struct request_queue *q)
1433 return q->limits.atomic_write_unit_max;
1436 static inline unsigned int
1437 queue_atomic_write_unit_min_bytes(const struct request_queue *q)
1439 return q->limits.atomic_write_unit_min;
1442 static inline unsigned int
1443 queue_atomic_write_boundary_bytes(const struct request_queue *q)
1445 return q->limits.atomic_write_boundary_sectors << SECTOR_SHIFT;
1448 static inline unsigned int
1449 queue_atomic_write_max_bytes(const struct request_queue *q)
1451 return q->limits.atomic_write_max_sectors << SECTOR_SHIFT;
1454 static inline unsigned int bdev_dma_alignment(struct block_device *bdev)
1456 return queue_dma_alignment(bdev_get_queue(bdev));
1459 static inline bool bdev_iter_is_aligned(struct block_device *bdev,
1460 struct iov_iter *iter)
1462 return iov_iter_is_aligned(iter, bdev_dma_alignment(bdev),
1463 bdev_logical_block_size(bdev) - 1);
1466 static inline unsigned int
1467 blk_lim_dma_alignment_and_pad(struct queue_limits *lim)
1469 return lim->dma_alignment | lim->dma_pad_mask;
1472 static inline bool blk_rq_aligned(struct request_queue *q, unsigned long addr,
1475 unsigned int alignment = blk_lim_dma_alignment_and_pad(&q->limits);
1477 return !(addr & alignment) && !(len & alignment);
1480 /* assumes size > 256 */
1481 static inline unsigned int blksize_bits(unsigned int size)
1483 return order_base_2(size >> SECTOR_SHIFT) + SECTOR_SHIFT;
1486 int kblockd_schedule_work(struct work_struct *work);
1487 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1489 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1490 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1491 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1492 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1494 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1496 bool blk_crypto_register(struct blk_crypto_profile *profile,
1497 struct request_queue *q);
1499 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1501 static inline bool blk_crypto_register(struct blk_crypto_profile *profile,
1502 struct request_queue *q)
1507 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1509 enum blk_unique_id {
1510 /* these match the Designator Types specified in SPC */
1516 struct block_device_operations {
1517 void (*submit_bio)(struct bio *bio);
1518 int (*poll_bio)(struct bio *bio, struct io_comp_batch *iob,
1519 unsigned int flags);
1520 int (*open)(struct gendisk *disk, blk_mode_t mode);
1521 void (*release)(struct gendisk *disk);
1522 int (*ioctl)(struct block_device *bdev, blk_mode_t mode,
1523 unsigned cmd, unsigned long arg);
1524 int (*compat_ioctl)(struct block_device *bdev, blk_mode_t mode,
1525 unsigned cmd, unsigned long arg);
1526 unsigned int (*check_events) (struct gendisk *disk,
1527 unsigned int clearing);
1528 void (*unlock_native_capacity) (struct gendisk *);
1529 int (*getgeo)(struct block_device *, struct hd_geometry *);
1530 int (*set_read_only)(struct block_device *bdev, bool ro);
1531 void (*free_disk)(struct gendisk *disk);
1532 /* this callback is with swap_lock and sometimes page table lock held */
1533 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1534 int (*report_zones)(struct gendisk *, sector_t sector,
1535 unsigned int nr_zones, report_zones_cb cb, void *data);
1536 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1537 /* returns the length of the identifier or a negative errno: */
1538 int (*get_unique_id)(struct gendisk *disk, u8 id[16],
1539 enum blk_unique_id id_type);
1540 struct module *owner;
1541 const struct pr_ops *pr_ops;
1544 * Special callback for probing GPT entry at a given sector.
1545 * Needed by Android devices, used by GPT scanner and MMC blk
1548 int (*alternative_gpt_sector)(struct gendisk *disk, sector_t *sector);
1551 #ifdef CONFIG_COMPAT
1552 extern int blkdev_compat_ptr_ioctl(struct block_device *, blk_mode_t,
1553 unsigned int, unsigned long);
1555 #define blkdev_compat_ptr_ioctl NULL
1558 static inline void blk_wake_io_task(struct task_struct *waiter)
1561 * If we're polling, the task itself is doing the completions. For
1562 * that case, we don't need to signal a wakeup, it's enough to just
1563 * mark us as RUNNING.
1565 if (waiter == current)
1566 __set_current_state(TASK_RUNNING);
1568 wake_up_process(waiter);
1571 unsigned long bdev_start_io_acct(struct block_device *bdev, enum req_op op,
1572 unsigned long start_time);
1573 void bdev_end_io_acct(struct block_device *bdev, enum req_op op,
1574 unsigned int sectors, unsigned long start_time);
1576 unsigned long bio_start_io_acct(struct bio *bio);
1577 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1578 struct block_device *orig_bdev);
1581 * bio_end_io_acct - end I/O accounting for bio based drivers
1582 * @bio: bio to end account for
1583 * @start_time: start time returned by bio_start_io_acct()
1585 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1587 return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1590 int set_blocksize(struct file *file, int size);
1592 int lookup_bdev(const char *pathname, dev_t *dev);
1594 void blkdev_show(struct seq_file *seqf, off_t offset);
1596 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1597 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1599 #define BLKDEV_MAJOR_MAX 512
1601 #define BLKDEV_MAJOR_MAX 0
1604 struct blk_holder_ops {
1605 void (*mark_dead)(struct block_device *bdev, bool surprise);
1608 * Sync the file system mounted on the block device.
1610 void (*sync)(struct block_device *bdev);
1613 * Freeze the file system mounted on the block device.
1615 int (*freeze)(struct block_device *bdev);
1618 * Thaw the file system mounted on the block device.
1620 int (*thaw)(struct block_device *bdev);
1624 * For filesystems using @fs_holder_ops, the @holder argument passed to
1625 * helpers used to open and claim block devices via
1626 * bd_prepare_to_claim() must point to a superblock.
1628 extern const struct blk_holder_ops fs_holder_ops;
1631 * Return the correct open flags for blkdev_get_by_* for super block flags
1632 * as stored in sb->s_flags.
1634 #define sb_open_mode(flags) \
1635 (BLK_OPEN_READ | BLK_OPEN_RESTRICT_WRITES | \
1636 (((flags) & SB_RDONLY) ? 0 : BLK_OPEN_WRITE))
1638 struct file *bdev_file_open_by_dev(dev_t dev, blk_mode_t mode, void *holder,
1639 const struct blk_holder_ops *hops);
1640 struct file *bdev_file_open_by_path(const char *path, blk_mode_t mode,
1641 void *holder, const struct blk_holder_ops *hops);
1642 int bd_prepare_to_claim(struct block_device *bdev, void *holder,
1643 const struct blk_holder_ops *hops);
1644 void bd_abort_claiming(struct block_device *bdev, void *holder);
1646 /* just for blk-cgroup, don't use elsewhere */
1647 struct block_device *blkdev_get_no_open(dev_t dev);
1648 void blkdev_put_no_open(struct block_device *bdev);
1650 struct block_device *I_BDEV(struct inode *inode);
1651 struct block_device *file_bdev(struct file *bdev_file);
1652 bool disk_live(struct gendisk *disk);
1653 unsigned int block_size(struct block_device *bdev);
1656 void invalidate_bdev(struct block_device *bdev);
1657 int sync_blockdev(struct block_device *bdev);
1658 int sync_blockdev_range(struct block_device *bdev, loff_t lstart, loff_t lend);
1659 int sync_blockdev_nowait(struct block_device *bdev);
1660 void sync_bdevs(bool wait);
1661 void bdev_statx(struct path *, struct kstat *, u32);
1662 void printk_all_partitions(void);
1663 int __init early_lookup_bdev(const char *pathname, dev_t *dev);
1665 static inline void invalidate_bdev(struct block_device *bdev)
1668 static inline int sync_blockdev(struct block_device *bdev)
1672 static inline int sync_blockdev_nowait(struct block_device *bdev)
1676 static inline void sync_bdevs(bool wait)
1679 static inline void bdev_statx(struct path *path, struct kstat *stat,
1683 static inline void printk_all_partitions(void)
1686 static inline int early_lookup_bdev(const char *pathname, dev_t *dev)
1690 #endif /* CONFIG_BLOCK */
1692 int bdev_freeze(struct block_device *bdev);
1693 int bdev_thaw(struct block_device *bdev);
1694 void bdev_fput(struct file *bdev_file);
1696 struct io_comp_batch {
1697 struct rq_list req_list;
1699 void (*complete)(struct io_comp_batch *);
1702 static inline bool bdev_can_atomic_write(struct block_device *bdev)
1704 struct request_queue *bd_queue = bdev->bd_queue;
1705 struct queue_limits *limits = &bd_queue->limits;
1707 if (!limits->atomic_write_unit_min)
1710 if (bdev_is_partition(bdev)) {
1711 sector_t bd_start_sect = bdev->bd_start_sect;
1712 unsigned int alignment =
1713 max(limits->atomic_write_unit_min,
1714 limits->atomic_write_hw_boundary);
1716 if (!IS_ALIGNED(bd_start_sect, alignment >> SECTOR_SHIFT))
1723 static inline unsigned int
1724 bdev_atomic_write_unit_min_bytes(struct block_device *bdev)
1726 if (!bdev_can_atomic_write(bdev))
1728 return queue_atomic_write_unit_min_bytes(bdev_get_queue(bdev));
1731 static inline unsigned int
1732 bdev_atomic_write_unit_max_bytes(struct block_device *bdev)
1734 if (!bdev_can_atomic_write(bdev))
1736 return queue_atomic_write_unit_max_bytes(bdev_get_queue(bdev));
1739 #define DEFINE_IO_COMP_BATCH(name) struct io_comp_batch name = { }
1741 #endif /* _LINUX_BLKDEV_H */