projects/linux·blob v6.4-rc7 · refs

1527 lines · 44.2 KiB

1/* SPDX-License-Identifier: GPL-2.0 */
2/*
3 * Portions Copyright (C) 1992 Drew Eckhardt
4 */
5#ifndef _LINUX_BLKDEV_H
6#define _LINUX_BLKDEV_H
7
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
28struct module;
29struct request_queue;
30struct elevator_queue;
31struct blk_trace;
32struct request;
33struct sg_io_hdr;
34struct blkcg_gq;
35struct blk_flush_queue;
36struct kiocb;
37struct pr_ops;
38struct rq_qos;
39struct blk_queue_stats;
40struct blk_stat_callback;
41struct blk_crypto_profile;
42
43extern const struct device_type disk_type;
44extern struct device_type part_type;
45extern struct class block_class;
46
47/*
48 * Maximum number of blkcg policies allowed to be registered concurrently.
49 * Defined here to simplify include dependency.
50 */
51#define BLKCG_MAX_POLS 6
52
53#define DISK_MAX_PARTS 256
54#define DISK_NAME_LEN 32
55
56#define PARTITION_META_INFO_VOLNAMELTH 64
57/*
58 * Enough for the string representation of any kind of UUID plus NULL.
59 * EFI UUID is 36 characters. MSDOS UUID is 11 characters.
60 */
61#define PARTITION_META_INFO_UUIDLTH (UUID_STRING_LEN + 1)
62
63struct partition_meta_info {
64 char uuid[PARTITION_META_INFO_UUIDLTH];
65 u8 volname[PARTITION_META_INFO_VOLNAMELTH];
66};
67
68/**
69 * DOC: genhd capability flags
70 *
71 * ``GENHD_FL_REMOVABLE``: indicates that the block device gives access to
72 * removable media. When set, the device remains present even when media is not
73 * inserted. Shall not be set for devices which are removed entirely when the
74 * media is removed.
75 *
76 * ``GENHD_FL_HIDDEN``: the block device is hidden; it doesn't produce events,
77 * doesn't appear in sysfs, and can't be opened from userspace or using
78 * blkdev_get*. Used for the underlying components of multipath devices.
79 *
80 * ``GENHD_FL_NO_PART``: partition support is disabled. The kernel will not
81 * scan for partitions from add_disk, and users can't add partitions manually.
82 *
83 */
84enum {
85 GENHD_FL_REMOVABLE = 1 << 0,
86 GENHD_FL_HIDDEN = 1 << 1,
87 GENHD_FL_NO_PART = 1 << 2,
88};
89
90enum {
91 DISK_EVENT_MEDIA_CHANGE = 1 << 0, /* media changed */
92 DISK_EVENT_EJECT_REQUEST = 1 << 1, /* eject requested */
93};
94
95enum {
96 /* Poll even if events_poll_msecs is unset */
97 DISK_EVENT_FLAG_POLL = 1 << 0,
98 /* Forward events to udev */
99 DISK_EVENT_FLAG_UEVENT = 1 << 1,
100 /* Block event polling when open for exclusive write */
101 DISK_EVENT_FLAG_BLOCK_ON_EXCL_WRITE = 1 << 2,
102};
103
104struct disk_events;
105struct badblocks;
106
107struct blk_integrity {
108 const struct blk_integrity_profile *profile;
109 unsigned char flags;
110 unsigned char tuple_size;
111 unsigned char interval_exp;
112 unsigned char tag_size;
113};
114
115struct gendisk {
116 /*
117 * major/first_minor/minors should not be set by any new driver, the
118 * block core will take care of allocating them automatically.
119 */
120 int major;
121 int first_minor;
122 int minors;
123
124 char disk_name[DISK_NAME_LEN]; /* name of major driver */
125
126 unsigned short events; /* supported events */
127 unsigned short event_flags; /* flags related to event processing */
128
129 struct xarray part_tbl;
130 struct block_device *part0;
131
132 const struct block_device_operations *fops;
133 struct request_queue *queue;
134 void *private_data;
135
136 struct bio_set bio_split;
137
138 int flags;
139 unsigned long state;
140#define GD_NEED_PART_SCAN 0
141#define GD_READ_ONLY 1
142#define GD_DEAD 2
143#define GD_NATIVE_CAPACITY 3
144#define GD_ADDED 4
145#define GD_SUPPRESS_PART_SCAN 5
146#define GD_OWNS_QUEUE 6
147
148 struct mutex open_mutex; /* open/close mutex */
149 unsigned open_partitions; /* number of open partitions */
150
151 struct backing_dev_info *bdi;
152 struct kobject queue_kobj; /* the queue/ directory */
153 struct kobject *slave_dir;
154#ifdef CONFIG_BLOCK_HOLDER_DEPRECATED
155 struct list_head slave_bdevs;
156#endif
157 struct timer_rand_state *random;
158 atomic_t sync_io; /* RAID */
159 struct disk_events *ev;
160
161#ifdef CONFIG_BLK_DEV_ZONED
162 /*
163 * Zoned block device information for request dispatch control.
164 * nr_zones is the total number of zones of the device. This is always
165 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
166 * bits which indicates if a zone is conventional (bit set) or
167 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
168 * bits which indicates if a zone is write locked, that is, if a write
169 * request targeting the zone was dispatched.
170 *
171 * Reads of this information must be protected with blk_queue_enter() /
172 * blk_queue_exit(). Modifying this information is only allowed while
173 * no requests are being processed. See also blk_mq_freeze_queue() and
174 * blk_mq_unfreeze_queue().
175 */
176 unsigned int nr_zones;
177 unsigned int max_open_zones;
178 unsigned int max_active_zones;
179 unsigned long *conv_zones_bitmap;
180 unsigned long *seq_zones_wlock;
181#endif /* CONFIG_BLK_DEV_ZONED */
182
183#if IS_ENABLED(CONFIG_CDROM)
184 struct cdrom_device_info *cdi;
185#endif
186 int node_id;
187 struct badblocks *bb;
188 struct lockdep_map lockdep_map;
189 u64 diskseq;
190
191 /*
192 * Independent sector access ranges. This is always NULL for
193 * devices that do not have multiple independent access ranges.
194 */
195 struct blk_independent_access_ranges *ia_ranges;
196};
197
198static inline bool disk_live(struct gendisk *disk)
199{
200 return !inode_unhashed(disk->part0->bd_inode);
201}
202
203/**
204 * disk_openers - returns how many openers are there for a disk
205 * @disk: disk to check
206 *
207 * This returns the number of openers for a disk. Note that this value is only
208 * stable if disk->open_mutex is held.
209 *
210 * Note: Due to a quirk in the block layer open code, each open partition is
211 * only counted once even if there are multiple openers.
212 */
213static inline unsigned int disk_openers(struct gendisk *disk)
214{
215 return atomic_read(&disk->part0->bd_openers);
216}
217
218/*
219 * The gendisk is refcounted by the part0 block_device, and the bd_device
220 * therein is also used for device model presentation in sysfs.
221 */
222#define dev_to_disk(device) \
223 (dev_to_bdev(device)->bd_disk)
224#define disk_to_dev(disk) \
225 (&((disk)->part0->bd_device))
226
227#if IS_REACHABLE(CONFIG_CDROM)
228#define disk_to_cdi(disk) ((disk)->cdi)
229#else
230#define disk_to_cdi(disk) NULL
231#endif
232
233static inline dev_t disk_devt(struct gendisk *disk)
234{
235 return MKDEV(disk->major, disk->first_minor);
236}
237
238static inline int blk_validate_block_size(unsigned long bsize)
239{
240 if (bsize < 512 || bsize > PAGE_SIZE || !is_power_of_2(bsize))
241 return -EINVAL;
242
243 return 0;
244}
245
246static inline bool blk_op_is_passthrough(blk_opf_t op)
247{
248 op &= REQ_OP_MASK;
249 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
250}
251
252/*
253 * Zoned block device models (zoned limit).
254 *
255 * Note: This needs to be ordered from the least to the most severe
256 * restrictions for the inheritance in blk_stack_limits() to work.
257 */
258enum blk_zoned_model {
259 BLK_ZONED_NONE = 0, /* Regular block device */
260 BLK_ZONED_HA, /* Host-aware zoned block device */
261 BLK_ZONED_HM, /* Host-managed zoned block device */
262};
263
264/*
265 * BLK_BOUNCE_NONE: never bounce (default)
266 * BLK_BOUNCE_HIGH: bounce all highmem pages
267 */
268enum blk_bounce {
269 BLK_BOUNCE_NONE,
270 BLK_BOUNCE_HIGH,
271};
272
273struct queue_limits {
274 enum blk_bounce bounce;
275 unsigned long seg_boundary_mask;
276 unsigned long virt_boundary_mask;
277
278 unsigned int max_hw_sectors;
279 unsigned int max_dev_sectors;
280 unsigned int chunk_sectors;
281 unsigned int max_sectors;
282 unsigned int max_user_sectors;
283 unsigned int max_segment_size;
284 unsigned int physical_block_size;
285 unsigned int logical_block_size;
286 unsigned int alignment_offset;
287 unsigned int io_min;
288 unsigned int io_opt;
289 unsigned int max_discard_sectors;
290 unsigned int max_hw_discard_sectors;
291 unsigned int max_secure_erase_sectors;
292 unsigned int max_write_zeroes_sectors;
293 unsigned int max_zone_append_sectors;
294 unsigned int discard_granularity;
295 unsigned int discard_alignment;
296 unsigned int zone_write_granularity;
297
298 unsigned short max_segments;
299 unsigned short max_integrity_segments;
300 unsigned short max_discard_segments;
301
302 unsigned char misaligned;
303 unsigned char discard_misaligned;
304 unsigned char raid_partial_stripes_expensive;
305 enum blk_zoned_model zoned;
306
307 /*
308 * Drivers that set dma_alignment to less than 511 must be prepared to
309 * handle individual bvec's that are not a multiple of a SECTOR_SIZE
310 * due to possible offsets.
311 */
312 unsigned int dma_alignment;
313};
314
315typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
316 void *data);
317
318void disk_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
319
320#ifdef CONFIG_BLK_DEV_ZONED
321
322#define BLK_ALL_ZONES ((unsigned int)-1)
323int blkdev_report_zones(struct block_device *bdev, sector_t sector,
324 unsigned int nr_zones, report_zones_cb cb, void *data);
325unsigned int bdev_nr_zones(struct block_device *bdev);
326extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_op op,
327 sector_t sectors, sector_t nr_sectors,
328 gfp_t gfp_mask);
329int blk_revalidate_disk_zones(struct gendisk *disk,
330 void (*update_driver_data)(struct gendisk *disk));
331
332extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
333 unsigned int cmd, unsigned long arg);
334extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
335 unsigned int cmd, unsigned long arg);
336
337#else /* CONFIG_BLK_DEV_ZONED */
338
339static inline unsigned int bdev_nr_zones(struct block_device *bdev)
340{
341 return 0;
342}
343
344static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
345 fmode_t mode, unsigned int cmd,
346 unsigned long arg)
347{
348 return -ENOTTY;
349}
350
351static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
352 fmode_t mode, unsigned int cmd,
353 unsigned long arg)
354{
355 return -ENOTTY;
356}
357
358#endif /* CONFIG_BLK_DEV_ZONED */
359
360/*
361 * Independent access ranges: struct blk_independent_access_range describes
362 * a range of contiguous sectors that can be accessed using device command
363 * execution resources that are independent from the resources used for
364 * other access ranges. This is typically found with single-LUN multi-actuator
365 * HDDs where each access range is served by a different set of heads.
366 * The set of independent ranges supported by the device is defined using
367 * struct blk_independent_access_ranges. The independent ranges must not overlap
368 * and must include all sectors within the disk capacity (no sector holes
369 * allowed).
370 * For a device with multiple ranges, requests targeting sectors in different
371 * ranges can be executed in parallel. A request can straddle an access range
372 * boundary.
373 */
374struct blk_independent_access_range {
375 struct kobject kobj;
376 sector_t sector;
377 sector_t nr_sectors;
378};
379
380struct blk_independent_access_ranges {
381 struct kobject kobj;
382 bool sysfs_registered;
383 unsigned int nr_ia_ranges;
384 struct blk_independent_access_range ia_range[];
385};
386
387struct request_queue {
388 struct request *last_merge;
389 struct elevator_queue *elevator;
390
391 struct percpu_ref q_usage_counter;
392
393 struct blk_queue_stats *stats;
394 struct rq_qos *rq_qos;
395
396 const struct blk_mq_ops *mq_ops;
397
398 /* sw queues */
399 struct blk_mq_ctx __percpu *queue_ctx;
400
401 unsigned int queue_depth;
402
403 /* hw dispatch queues */
404 struct xarray hctx_table;
405 unsigned int nr_hw_queues;
406
407 /*
408 * The queue owner gets to use this for whatever they like.
409 * ll_rw_blk doesn't touch it.
410 */
411 void *queuedata;
412
413 /*
414 * various queue flags, see QUEUE_* below
415 */
416 unsigned long queue_flags;
417 /*
418 * Number of contexts that have called blk_set_pm_only(). If this
419 * counter is above zero then only RQF_PM requests are processed.
420 */
421 atomic_t pm_only;
422
423 /*
424 * ida allocated id for this queue. Used to index queues from
425 * ioctx.
426 */
427 int id;
428
429 spinlock_t queue_lock;
430
431 struct gendisk *disk;
432
433 refcount_t refs;
434
435 /*
436 * mq queue kobject
437 */
438 struct kobject *mq_kobj;
439
440#ifdef CONFIG_BLK_DEV_INTEGRITY
441 struct blk_integrity integrity;
442#endif /* CONFIG_BLK_DEV_INTEGRITY */
443
444#ifdef CONFIG_PM
445 struct device *dev;
446 enum rpm_status rpm_status;
447#endif
448
449 /*
450 * queue settings
451 */
452 unsigned long nr_requests; /* Max # of requests */
453
454 unsigned int dma_pad_mask;
455
456#ifdef CONFIG_BLK_INLINE_ENCRYPTION
457 struct blk_crypto_profile *crypto_profile;
458 struct kobject *crypto_kobject;
459#endif
460
461 unsigned int rq_timeout;
462
463 struct timer_list timeout;
464 struct work_struct timeout_work;
465
466 atomic_t nr_active_requests_shared_tags;
467
468 struct blk_mq_tags *sched_shared_tags;
469
470 struct list_head icq_list;
471#ifdef CONFIG_BLK_CGROUP
472 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
473 struct blkcg_gq *root_blkg;
474 struct list_head blkg_list;
475 struct mutex blkcg_mutex;
476#endif
477
478 struct queue_limits limits;
479
480 unsigned int required_elevator_features;
481
482 int node;
483#ifdef CONFIG_BLK_DEV_IO_TRACE
484 struct blk_trace __rcu *blk_trace;
485#endif
486 /*
487 * for flush operations
488 */
489 struct blk_flush_queue *fq;
490
491 struct list_head requeue_list;
492 spinlock_t requeue_lock;
493 struct delayed_work requeue_work;
494
495 struct mutex sysfs_lock;
496 struct mutex sysfs_dir_lock;
497
498 /*
499 * for reusing dead hctx instance in case of updating
500 * nr_hw_queues
501 */
502 struct list_head unused_hctx_list;
503 spinlock_t unused_hctx_lock;
504
505 int mq_freeze_depth;
506
507#ifdef CONFIG_BLK_DEV_THROTTLING
508 /* Throttle data */
509 struct throtl_data *td;
510#endif
511 struct rcu_head rcu_head;
512 wait_queue_head_t mq_freeze_wq;
513 /*
514 * Protect concurrent access to q_usage_counter by
515 * percpu_ref_kill() and percpu_ref_reinit().
516 */
517 struct mutex mq_freeze_lock;
518
519 int quiesce_depth;
520
521 struct blk_mq_tag_set *tag_set;
522 struct list_head tag_set_list;
523
524 struct dentry *debugfs_dir;
525 struct dentry *sched_debugfs_dir;
526 struct dentry *rqos_debugfs_dir;
527 /*
528 * Serializes all debugfs metadata operations using the above dentries.
529 */
530 struct mutex debugfs_mutex;
531
532 bool mq_sysfs_init_done;
533};
534
535/* Keep blk_queue_flag_name[] in sync with the definitions below */
536#define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
537#define QUEUE_FLAG_DYING 1 /* queue being torn down */
538#define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
539#define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
540#define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
541#define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
542#define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
543#define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
544#define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
545#define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
546#define QUEUE_FLAG_SYNCHRONOUS 11 /* always completes in submit context */
547#define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
548#define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
549#define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
550#define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
551#define QUEUE_FLAG_WC 17 /* Write back caching */
552#define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
553#define QUEUE_FLAG_DAX 19 /* device supports DAX */
554#define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
555#define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
556#define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
557#define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
558#define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
559#define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
560#define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
561#define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */
562#define QUEUE_FLAG_SQ_SCHED 30 /* single queue style io dispatch */
563#define QUEUE_FLAG_SKIP_TAGSET_QUIESCE 31 /* quiesce_tagset skip the queue*/
564
565#define QUEUE_FLAG_MQ_DEFAULT ((1UL << QUEUE_FLAG_IO_STAT) | \
566 (1UL << QUEUE_FLAG_SAME_COMP) | \
567 (1UL << QUEUE_FLAG_NOWAIT))
568
569void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
570void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
571bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
572
573#define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
574#define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
575#define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
576#define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
577#define blk_queue_noxmerges(q) \
578 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
579#define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
580#define blk_queue_stable_writes(q) \
581 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
582#define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
583#define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
584#define blk_queue_zone_resetall(q) \
585 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
586#define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
587#define blk_queue_pci_p2pdma(q) \
588 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
589#ifdef CONFIG_BLK_RQ_ALLOC_TIME
590#define blk_queue_rq_alloc_time(q) \
591 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
592#else
593#define blk_queue_rq_alloc_time(q) false
594#endif
595
596#define blk_noretry_request(rq) \
597 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
598 REQ_FAILFAST_DRIVER))
599#define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
600#define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
601#define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
602#define blk_queue_sq_sched(q) test_bit(QUEUE_FLAG_SQ_SCHED, &(q)->queue_flags)
603#define blk_queue_skip_tagset_quiesce(q) \
604 test_bit(QUEUE_FLAG_SKIP_TAGSET_QUIESCE, &(q)->queue_flags)
605
606extern void blk_set_pm_only(struct request_queue *q);
607extern void blk_clear_pm_only(struct request_queue *q);
608
609#define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
610
611#define dma_map_bvec(dev, bv, dir, attrs) \
612 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
613 (dir), (attrs))
614
615static inline bool queue_is_mq(struct request_queue *q)
616{
617 return q->mq_ops;
618}
619
620#ifdef CONFIG_PM
621static inline enum rpm_status queue_rpm_status(struct request_queue *q)
622{
623 return q->rpm_status;
624}
625#else
626static inline enum rpm_status queue_rpm_status(struct request_queue *q)
627{
628 return RPM_ACTIVE;
629}
630#endif
631
632static inline enum blk_zoned_model
633blk_queue_zoned_model(struct request_queue *q)
634{
635 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
636 return q->limits.zoned;
637 return BLK_ZONED_NONE;
638}
639
640static inline bool blk_queue_is_zoned(struct request_queue *q)
641{
642 switch (blk_queue_zoned_model(q)) {
643 case BLK_ZONED_HA:
644 case BLK_ZONED_HM:
645 return true;
646 default:
647 return false;
648 }
649}
650
651#ifdef CONFIG_BLK_DEV_ZONED
652static inline unsigned int disk_nr_zones(struct gendisk *disk)
653{
654 return blk_queue_is_zoned(disk->queue) ? disk->nr_zones : 0;
655}
656
657static inline unsigned int disk_zone_no(struct gendisk *disk, sector_t sector)
658{
659 if (!blk_queue_is_zoned(disk->queue))
660 return 0;
661 return sector >> ilog2(disk->queue->limits.chunk_sectors);
662}
663
664static inline bool disk_zone_is_seq(struct gendisk *disk, sector_t sector)
665{
666 if (!blk_queue_is_zoned(disk->queue))
667 return false;
668 if (!disk->conv_zones_bitmap)
669 return true;
670 return !test_bit(disk_zone_no(disk, sector), disk->conv_zones_bitmap);
671}
672
673static inline void disk_set_max_open_zones(struct gendisk *disk,
674 unsigned int max_open_zones)
675{
676 disk->max_open_zones = max_open_zones;
677}
678
679static inline void disk_set_max_active_zones(struct gendisk *disk,
680 unsigned int max_active_zones)
681{
682 disk->max_active_zones = max_active_zones;
683}
684
685static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
686{
687 return bdev->bd_disk->max_open_zones;
688}
689
690static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
691{
692 return bdev->bd_disk->max_active_zones;
693}
694
695#else /* CONFIG_BLK_DEV_ZONED */
696static inline unsigned int disk_nr_zones(struct gendisk *disk)
697{
698 return 0;
699}
700static inline bool disk_zone_is_seq(struct gendisk *disk, sector_t sector)
701{
702 return false;
703}
704static inline unsigned int disk_zone_no(struct gendisk *disk, sector_t sector)
705{
706 return 0;
707}
708static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
709{
710 return 0;
711}
712
713static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
714{
715 return 0;
716}
717#endif /* CONFIG_BLK_DEV_ZONED */
718
719static inline unsigned int blk_queue_depth(struct request_queue *q)
720{
721 if (q->queue_depth)
722 return q->queue_depth;
723
724 return q->nr_requests;
725}
726
727/*
728 * default timeout for SG_IO if none specified
729 */
730#define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
731#define BLK_MIN_SG_TIMEOUT (7 * HZ)
732
733/* This should not be used directly - use rq_for_each_segment */
734#define for_each_bio(_bio) \
735 for (; _bio; _bio = _bio->bi_next)
736
737int __must_check device_add_disk(struct device *parent, struct gendisk *disk,
738 const struct attribute_group **groups);
739static inline int __must_check add_disk(struct gendisk *disk)
740{
741 return device_add_disk(NULL, disk, NULL);
742}
743void del_gendisk(struct gendisk *gp);
744void invalidate_disk(struct gendisk *disk);
745void set_disk_ro(struct gendisk *disk, bool read_only);
746void disk_uevent(struct gendisk *disk, enum kobject_action action);
747
748static inline int get_disk_ro(struct gendisk *disk)
749{
750 return disk->part0->bd_read_only ||
751 test_bit(GD_READ_ONLY, &disk->state);
752}
753
754static inline int bdev_read_only(struct block_device *bdev)
755{
756 return bdev->bd_read_only || get_disk_ro(bdev->bd_disk);
757}
758
759bool set_capacity_and_notify(struct gendisk *disk, sector_t size);
760bool disk_force_media_change(struct gendisk *disk, unsigned int events);
761
762void add_disk_randomness(struct gendisk *disk) __latent_entropy;
763void rand_initialize_disk(struct gendisk *disk);
764
765static inline sector_t get_start_sect(struct block_device *bdev)
766{
767 return bdev->bd_start_sect;
768}
769
770static inline sector_t bdev_nr_sectors(struct block_device *bdev)
771{
772 return bdev->bd_nr_sectors;
773}
774
775static inline loff_t bdev_nr_bytes(struct block_device *bdev)
776{
777 return (loff_t)bdev_nr_sectors(bdev) << SECTOR_SHIFT;
778}
779
780static inline sector_t get_capacity(struct gendisk *disk)
781{
782 return bdev_nr_sectors(disk->part0);
783}
784
785static inline u64 sb_bdev_nr_blocks(struct super_block *sb)
786{
787 return bdev_nr_sectors(sb->s_bdev) >>
788 (sb->s_blocksize_bits - SECTOR_SHIFT);
789}
790
791int bdev_disk_changed(struct gendisk *disk, bool invalidate);
792
793void put_disk(struct gendisk *disk);
794struct gendisk *__blk_alloc_disk(int node, struct lock_class_key *lkclass);
795
796/**
797 * blk_alloc_disk - allocate a gendisk structure
798 * @node_id: numa node to allocate on
799 *
800 * Allocate and pre-initialize a gendisk structure for use with BIO based
801 * drivers.
802 *
803 * Context: can sleep
804 */
805#define blk_alloc_disk(node_id) \
806({ \
807 static struct lock_class_key __key; \
808 \
809 __blk_alloc_disk(node_id, &__key); \
810})
811
812int __register_blkdev(unsigned int major, const char *name,
813 void (*probe)(dev_t devt));
814#define register_blkdev(major, name) \
815 __register_blkdev(major, name, NULL)
816void unregister_blkdev(unsigned int major, const char *name);
817
818bool bdev_check_media_change(struct block_device *bdev);
819int __invalidate_device(struct block_device *bdev, bool kill_dirty);
820void set_capacity(struct gendisk *disk, sector_t size);
821
822#ifdef CONFIG_BLOCK_HOLDER_DEPRECATED
823int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
824void bd_unlink_disk_holder(struct block_device *bdev, struct gendisk *disk);
825#else
826static inline int bd_link_disk_holder(struct block_device *bdev,
827 struct gendisk *disk)
828{
829 return 0;
830}
831static inline void bd_unlink_disk_holder(struct block_device *bdev,
832 struct gendisk *disk)
833{
834}
835#endif /* CONFIG_BLOCK_HOLDER_DEPRECATED */
836
837dev_t part_devt(struct gendisk *disk, u8 partno);
838void inc_diskseq(struct gendisk *disk);
839dev_t blk_lookup_devt(const char *name, int partno);
840void blk_request_module(dev_t devt);
841
842extern int blk_register_queue(struct gendisk *disk);
843extern void blk_unregister_queue(struct gendisk *disk);
844void submit_bio_noacct(struct bio *bio);
845struct bio *bio_split_to_limits(struct bio *bio);
846
847extern int blk_lld_busy(struct request_queue *q);
848extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
849extern void blk_queue_exit(struct request_queue *q);
850extern void blk_sync_queue(struct request_queue *q);
851
852/* Helper to convert REQ_OP_XXX to its string format XXX */
853extern const char *blk_op_str(enum req_op op);
854
855int blk_status_to_errno(blk_status_t status);
856blk_status_t errno_to_blk_status(int errno);
857
858/* only poll the hardware once, don't continue until a completion was found */
859#define BLK_POLL_ONESHOT (1 << 0)
860int bio_poll(struct bio *bio, struct io_comp_batch *iob, unsigned int flags);
861int iocb_bio_iopoll(struct kiocb *kiocb, struct io_comp_batch *iob,
862 unsigned int flags);
863
864static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
865{
866 return bdev->bd_queue; /* this is never NULL */
867}
868
869/* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
870const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
871
872static inline unsigned int bio_zone_no(struct bio *bio)
873{
874 return disk_zone_no(bio->bi_bdev->bd_disk, bio->bi_iter.bi_sector);
875}
876
877static inline unsigned int bio_zone_is_seq(struct bio *bio)
878{
879 return disk_zone_is_seq(bio->bi_bdev->bd_disk, bio->bi_iter.bi_sector);
880}
881
882/*
883 * Return how much of the chunk is left to be used for I/O at a given offset.
884 */
885static inline unsigned int blk_chunk_sectors_left(sector_t offset,
886 unsigned int chunk_sectors)
887{
888 if (unlikely(!is_power_of_2(chunk_sectors)))
889 return chunk_sectors - sector_div(offset, chunk_sectors);
890 return chunk_sectors - (offset & (chunk_sectors - 1));
891}
892
893/*
894 * Access functions for manipulating queue properties
895 */
896void blk_queue_bounce_limit(struct request_queue *q, enum blk_bounce limit);
897extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
898extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
899extern void blk_queue_max_segments(struct request_queue *, unsigned short);
900extern void blk_queue_max_discard_segments(struct request_queue *,
901 unsigned short);
902void blk_queue_max_secure_erase_sectors(struct request_queue *q,
903 unsigned int max_sectors);
904extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
905extern void blk_queue_max_discard_sectors(struct request_queue *q,
906 unsigned int max_discard_sectors);
907extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
908 unsigned int max_write_same_sectors);
909extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
910extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
911 unsigned int max_zone_append_sectors);
912extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
913void blk_queue_zone_write_granularity(struct request_queue *q,
914 unsigned int size);
915extern void blk_queue_alignment_offset(struct request_queue *q,
916 unsigned int alignment);
917void disk_update_readahead(struct gendisk *disk);
918extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
919extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
920extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
921extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
922extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
923extern void blk_set_stacking_limits(struct queue_limits *lim);
924extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
925 sector_t offset);
926extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
927 sector_t offset);
928extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
929extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
930extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
931extern void blk_queue_dma_alignment(struct request_queue *, int);
932extern void blk_queue_update_dma_alignment(struct request_queue *, int);
933extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
934extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
935
936struct blk_independent_access_ranges *
937disk_alloc_independent_access_ranges(struct gendisk *disk, int nr_ia_ranges);
938void disk_set_independent_access_ranges(struct gendisk *disk,
939 struct blk_independent_access_ranges *iars);
940
941/*
942 * Elevator features for blk_queue_required_elevator_features:
943 */
944/* Supports zoned block devices sequential write constraint */
945#define ELEVATOR_F_ZBD_SEQ_WRITE (1U << 0)
946
947extern void blk_queue_required_elevator_features(struct request_queue *q,
948 unsigned int features);
949extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
950 struct device *dev);
951
952bool __must_check blk_get_queue(struct request_queue *);
953extern void blk_put_queue(struct request_queue *);
954
955void blk_mark_disk_dead(struct gendisk *disk);
956
957#ifdef CONFIG_BLOCK
958/*
959 * blk_plug permits building a queue of related requests by holding the I/O
960 * fragments for a short period. This allows merging of sequential requests
961 * into single larger request. As the requests are moved from a per-task list to
962 * the device's request_queue in a batch, this results in improved scalability
963 * as the lock contention for request_queue lock is reduced.
964 *
965 * It is ok not to disable preemption when adding the request to the plug list
966 * or when attempting a merge. For details, please see schedule() where
967 * blk_flush_plug() is called.
968 */
969struct blk_plug {
970 struct request *mq_list; /* blk-mq requests */
971
972 /* if ios_left is > 1, we can batch tag/rq allocations */
973 struct request *cached_rq;
974 unsigned short nr_ios;
975
976 unsigned short rq_count;
977
978 bool multiple_queues;
979 bool has_elevator;
980 bool nowait;
981
982 struct list_head cb_list; /* md requires an unplug callback */
983};
984
985struct blk_plug_cb;
986typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
987struct blk_plug_cb {
988 struct list_head list;
989 blk_plug_cb_fn callback;
990 void *data;
991};
992extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
993 void *data, int size);
994extern void blk_start_plug(struct blk_plug *);
995extern void blk_start_plug_nr_ios(struct blk_plug *, unsigned short);
996extern void blk_finish_plug(struct blk_plug *);
997
998void __blk_flush_plug(struct blk_plug *plug, bool from_schedule);
999static inline void blk_flush_plug(struct blk_plug *plug, bool async)
1000{
1001 if (plug)
1002 __blk_flush_plug(plug, async);
1003}
1004
1005int blkdev_issue_flush(struct block_device *bdev);
1006long nr_blockdev_pages(void);
1007#else /* CONFIG_BLOCK */
1008struct blk_plug {
1009};
1010
1011static inline void blk_start_plug_nr_ios(struct blk_plug *plug,
1012 unsigned short nr_ios)
1013{
1014}
1015
1016static inline void blk_start_plug(struct blk_plug *plug)
1017{
1018}
1019
1020static inline void blk_finish_plug(struct blk_plug *plug)
1021{
1022}
1023
1024static inline void blk_flush_plug(struct blk_plug *plug, bool async)
1025{
1026}
1027
1028static inline int blkdev_issue_flush(struct block_device *bdev)
1029{
1030 return 0;
1031}
1032
1033static inline long nr_blockdev_pages(void)
1034{
1035 return 0;
1036}
1037#endif /* CONFIG_BLOCK */
1038
1039extern void blk_io_schedule(void);
1040
1041int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1042 sector_t nr_sects, gfp_t gfp_mask);
1043int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1044 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop);
1045int blkdev_issue_secure_erase(struct block_device *bdev, sector_t sector,
1046 sector_t nr_sects, gfp_t gfp);
1047
1048#define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1049#define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1050
1051extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1052 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1053 unsigned flags);
1054extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1055 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1056
1057static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1058 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1059{
1060 return blkdev_issue_discard(sb->s_bdev,
1061 block << (sb->s_blocksize_bits -
1062 SECTOR_SHIFT),
1063 nr_blocks << (sb->s_blocksize_bits -
1064 SECTOR_SHIFT),
1065 gfp_mask);
1066}
1067static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1068 sector_t nr_blocks, gfp_t gfp_mask)
1069{
1070 return blkdev_issue_zeroout(sb->s_bdev,
1071 block << (sb->s_blocksize_bits -
1072 SECTOR_SHIFT),
1073 nr_blocks << (sb->s_blocksize_bits -
1074 SECTOR_SHIFT),
1075 gfp_mask, 0);
1076}
1077
1078static inline bool bdev_is_partition(struct block_device *bdev)
1079{
1080 return bdev->bd_partno;
1081}
1082
1083enum blk_default_limits {
1084 BLK_MAX_SEGMENTS = 128,
1085 BLK_SAFE_MAX_SECTORS = 255,
1086 BLK_MAX_SEGMENT_SIZE = 65536,
1087 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1088};
1089
1090#define BLK_DEF_MAX_SECTORS 2560u
1091
1092static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1093{
1094 return q->limits.seg_boundary_mask;
1095}
1096
1097static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1098{
1099 return q->limits.virt_boundary_mask;
1100}
1101
1102static inline unsigned int queue_max_sectors(const struct request_queue *q)
1103{
1104 return q->limits.max_sectors;
1105}
1106
1107static inline unsigned int queue_max_bytes(struct request_queue *q)
1108{
1109 return min_t(unsigned int, queue_max_sectors(q), INT_MAX >> 9) << 9;
1110}
1111
1112static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1113{
1114 return q->limits.max_hw_sectors;
1115}
1116
1117static inline unsigned short queue_max_segments(const struct request_queue *q)
1118{
1119 return q->limits.max_segments;
1120}
1121
1122static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1123{
1124 return q->limits.max_discard_segments;
1125}
1126
1127static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1128{
1129 return q->limits.max_segment_size;
1130}
1131
1132static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1133{
1134
1135 const struct queue_limits *l = &q->limits;
1136
1137 return min(l->max_zone_append_sectors, l->max_sectors);
1138}
1139
1140static inline unsigned int
1141bdev_max_zone_append_sectors(struct block_device *bdev)
1142{
1143 return queue_max_zone_append_sectors(bdev_get_queue(bdev));
1144}
1145
1146static inline unsigned int bdev_max_segments(struct block_device *bdev)
1147{
1148 return queue_max_segments(bdev_get_queue(bdev));
1149}
1150
1151static inline unsigned queue_logical_block_size(const struct request_queue *q)
1152{
1153 int retval = 512;
1154
1155 if (q && q->limits.logical_block_size)
1156 retval = q->limits.logical_block_size;
1157
1158 return retval;
1159}
1160
1161static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1162{
1163 return queue_logical_block_size(bdev_get_queue(bdev));
1164}
1165
1166static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1167{
1168 return q->limits.physical_block_size;
1169}
1170
1171static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1172{
1173 return queue_physical_block_size(bdev_get_queue(bdev));
1174}
1175
1176static inline unsigned int queue_io_min(const struct request_queue *q)
1177{
1178 return q->limits.io_min;
1179}
1180
1181static inline int bdev_io_min(struct block_device *bdev)
1182{
1183 return queue_io_min(bdev_get_queue(bdev));
1184}
1185
1186static inline unsigned int queue_io_opt(const struct request_queue *q)
1187{
1188 return q->limits.io_opt;
1189}
1190
1191static inline int bdev_io_opt(struct block_device *bdev)
1192{
1193 return queue_io_opt(bdev_get_queue(bdev));
1194}
1195
1196static inline unsigned int
1197queue_zone_write_granularity(const struct request_queue *q)
1198{
1199 return q->limits.zone_write_granularity;
1200}
1201
1202static inline unsigned int
1203bdev_zone_write_granularity(struct block_device *bdev)
1204{
1205 return queue_zone_write_granularity(bdev_get_queue(bdev));
1206}
1207
1208int bdev_alignment_offset(struct block_device *bdev);
1209unsigned int bdev_discard_alignment(struct block_device *bdev);
1210
1211static inline unsigned int bdev_max_discard_sectors(struct block_device *bdev)
1212{
1213 return bdev_get_queue(bdev)->limits.max_discard_sectors;
1214}
1215
1216static inline unsigned int bdev_discard_granularity(struct block_device *bdev)
1217{
1218 return bdev_get_queue(bdev)->limits.discard_granularity;
1219}
1220
1221static inline unsigned int
1222bdev_max_secure_erase_sectors(struct block_device *bdev)
1223{
1224 return bdev_get_queue(bdev)->limits.max_secure_erase_sectors;
1225}
1226
1227static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1228{
1229 struct request_queue *q = bdev_get_queue(bdev);
1230
1231 if (q)
1232 return q->limits.max_write_zeroes_sectors;
1233
1234 return 0;
1235}
1236
1237static inline bool bdev_nonrot(struct block_device *bdev)
1238{
1239 return blk_queue_nonrot(bdev_get_queue(bdev));
1240}
1241
1242static inline bool bdev_synchronous(struct block_device *bdev)
1243{
1244 return test_bit(QUEUE_FLAG_SYNCHRONOUS,
1245 &bdev_get_queue(bdev)->queue_flags);
1246}
1247
1248static inline bool bdev_stable_writes(struct block_device *bdev)
1249{
1250 return test_bit(QUEUE_FLAG_STABLE_WRITES,
1251 &bdev_get_queue(bdev)->queue_flags);
1252}
1253
1254static inline bool bdev_write_cache(struct block_device *bdev)
1255{
1256 return test_bit(QUEUE_FLAG_WC, &bdev_get_queue(bdev)->queue_flags);
1257}
1258
1259static inline bool bdev_fua(struct block_device *bdev)
1260{
1261 return test_bit(QUEUE_FLAG_FUA, &bdev_get_queue(bdev)->queue_flags);
1262}
1263
1264static inline bool bdev_nowait(struct block_device *bdev)
1265{
1266 return test_bit(QUEUE_FLAG_NOWAIT, &bdev_get_queue(bdev)->queue_flags);
1267}
1268
1269static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1270{
1271 return blk_queue_zoned_model(bdev_get_queue(bdev));
1272}
1273
1274static inline bool bdev_is_zoned(struct block_device *bdev)
1275{
1276 return blk_queue_is_zoned(bdev_get_queue(bdev));
1277}
1278
1279static inline unsigned int bdev_zone_no(struct block_device *bdev, sector_t sec)
1280{
1281 return disk_zone_no(bdev->bd_disk, sec);
1282}
1283
1284static inline bool bdev_op_is_zoned_write(struct block_device *bdev,
1285 blk_opf_t op)
1286{
1287 if (!bdev_is_zoned(bdev))
1288 return false;
1289
1290 return op == REQ_OP_WRITE || op == REQ_OP_WRITE_ZEROES;
1291}
1292
1293static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1294{
1295 struct request_queue *q = bdev_get_queue(bdev);
1296
1297 if (!blk_queue_is_zoned(q))
1298 return 0;
1299 return q->limits.chunk_sectors;
1300}
1301
1302static inline sector_t bdev_offset_from_zone_start(struct block_device *bdev,
1303 sector_t sector)
1304{
1305 return sector & (bdev_zone_sectors(bdev) - 1);
1306}
1307
1308static inline bool bdev_is_zone_start(struct block_device *bdev,
1309 sector_t sector)
1310{
1311 return bdev_offset_from_zone_start(bdev, sector) == 0;
1312}
1313
1314static inline int queue_dma_alignment(const struct request_queue *q)
1315{
1316 return q ? q->limits.dma_alignment : 511;
1317}
1318
1319static inline unsigned int bdev_dma_alignment(struct block_device *bdev)
1320{
1321 return queue_dma_alignment(bdev_get_queue(bdev));
1322}
1323
1324static inline bool bdev_iter_is_aligned(struct block_device *bdev,
1325 struct iov_iter *iter)
1326{
1327 return iov_iter_is_aligned(iter, bdev_dma_alignment(bdev),
1328 bdev_logical_block_size(bdev) - 1);
1329}
1330
1331static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1332 unsigned int len)
1333{
1334 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1335 return !(addr & alignment) && !(len & alignment);
1336}
1337
1338/* assumes size > 256 */
1339static inline unsigned int blksize_bits(unsigned int size)
1340{
1341 return order_base_2(size >> SECTOR_SHIFT) + SECTOR_SHIFT;
1342}
1343
1344static inline unsigned int block_size(struct block_device *bdev)
1345{
1346 return 1 << bdev->bd_inode->i_blkbits;
1347}
1348
1349int kblockd_schedule_work(struct work_struct *work);
1350int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1351
1352#define MODULE_ALIAS_BLOCKDEV(major,minor) \
1353 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1354#define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1355 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1356
1357#ifdef CONFIG_BLK_INLINE_ENCRYPTION
1358
1359bool blk_crypto_register(struct blk_crypto_profile *profile,
1360 struct request_queue *q);
1361
1362#else /* CONFIG_BLK_INLINE_ENCRYPTION */
1363
1364static inline bool blk_crypto_register(struct blk_crypto_profile *profile,
1365 struct request_queue *q)
1366{
1367 return true;
1368}
1369
1370#endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1371
1372enum blk_unique_id {
1373 /* these match the Designator Types specified in SPC */
1374 BLK_UID_T10 = 1,
1375 BLK_UID_EUI64 = 2,
1376 BLK_UID_NAA = 3,
1377};
1378
1379struct block_device_operations {
1380 void (*submit_bio)(struct bio *bio);
1381 int (*poll_bio)(struct bio *bio, struct io_comp_batch *iob,
1382 unsigned int flags);
1383 int (*open) (struct block_device *, fmode_t);
1384 void (*release) (struct gendisk *, fmode_t);
1385 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1386 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1387 unsigned int (*check_events) (struct gendisk *disk,
1388 unsigned int clearing);
1389 void (*unlock_native_capacity) (struct gendisk *);
1390 int (*getgeo)(struct block_device *, struct hd_geometry *);
1391 int (*set_read_only)(struct block_device *bdev, bool ro);
1392 void (*free_disk)(struct gendisk *disk);
1393 /* this callback is with swap_lock and sometimes page table lock held */
1394 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1395 int (*report_zones)(struct gendisk *, sector_t sector,
1396 unsigned int nr_zones, report_zones_cb cb, void *data);
1397 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1398 /* returns the length of the identifier or a negative errno: */
1399 int (*get_unique_id)(struct gendisk *disk, u8 id[16],
1400 enum blk_unique_id id_type);
1401 struct module *owner;
1402 const struct pr_ops *pr_ops;
1403
1404 /*
1405 * Special callback for probing GPT entry at a given sector.
1406 * Needed by Android devices, used by GPT scanner and MMC blk
1407 * driver.
1408 */
1409 int (*alternative_gpt_sector)(struct gendisk *disk, sector_t *sector);
1410};
1411
1412#ifdef CONFIG_COMPAT
1413extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1414 unsigned int, unsigned long);
1415#else
1416#define blkdev_compat_ptr_ioctl NULL
1417#endif
1418
1419static inline void blk_wake_io_task(struct task_struct *waiter)
1420{
1421 /*
1422 * If we're polling, the task itself is doing the completions. For
1423 * that case, we don't need to signal a wakeup, it's enough to just
1424 * mark us as RUNNING.
1425 */
1426 if (waiter == current)
1427 __set_current_state(TASK_RUNNING);
1428 else
1429 wake_up_process(waiter);
1430}
1431
1432unsigned long bdev_start_io_acct(struct block_device *bdev, enum req_op op,
1433 unsigned long start_time);
1434void bdev_end_io_acct(struct block_device *bdev, enum req_op op,
1435 unsigned int sectors, unsigned long start_time);
1436
1437unsigned long bio_start_io_acct(struct bio *bio);
1438void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1439 struct block_device *orig_bdev);
1440
1441/**
1442 * bio_end_io_acct - end I/O accounting for bio based drivers
1443 * @bio: bio to end account for
1444 * @start_time: start time returned by bio_start_io_acct()
1445 */
1446static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1447{
1448 return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1449}
1450
1451int bdev_read_only(struct block_device *bdev);
1452int set_blocksize(struct block_device *bdev, int size);
1453
1454int lookup_bdev(const char *pathname, dev_t *dev);
1455
1456void blkdev_show(struct seq_file *seqf, off_t offset);
1457
1458#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1459#define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1460#ifdef CONFIG_BLOCK
1461#define BLKDEV_MAJOR_MAX 512
1462#else
1463#define BLKDEV_MAJOR_MAX 0
1464#endif
1465
1466struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1467 void *holder);
1468struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1469int bd_prepare_to_claim(struct block_device *bdev, void *holder);
1470void bd_abort_claiming(struct block_device *bdev, void *holder);
1471void blkdev_put(struct block_device *bdev, fmode_t mode);
1472
1473/* just for blk-cgroup, don't use elsewhere */
1474struct block_device *blkdev_get_no_open(dev_t dev);
1475void blkdev_put_no_open(struct block_device *bdev);
1476
1477struct block_device *bdev_alloc(struct gendisk *disk, u8 partno);
1478void bdev_add(struct block_device *bdev, dev_t dev);
1479struct block_device *I_BDEV(struct inode *inode);
1480int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
1481 loff_t lend);
1482
1483#ifdef CONFIG_BLOCK
1484void invalidate_bdev(struct block_device *bdev);
1485int sync_blockdev(struct block_device *bdev);
1486int sync_blockdev_range(struct block_device *bdev, loff_t lstart, loff_t lend);
1487int sync_blockdev_nowait(struct block_device *bdev);
1488void sync_bdevs(bool wait);
1489void bdev_statx_dioalign(struct inode *inode, struct kstat *stat);
1490void printk_all_partitions(void);
1491#else
1492static inline void invalidate_bdev(struct block_device *bdev)
1493{
1494}
1495static inline int sync_blockdev(struct block_device *bdev)
1496{
1497 return 0;
1498}
1499static inline int sync_blockdev_nowait(struct block_device *bdev)
1500{
1501 return 0;
1502}
1503static inline void sync_bdevs(bool wait)
1504{
1505}
1506static inline void bdev_statx_dioalign(struct inode *inode, struct kstat *stat)
1507{
1508}
1509static inline void printk_all_partitions(void)
1510{
1511}
1512#endif /* CONFIG_BLOCK */
1513
1514int fsync_bdev(struct block_device *bdev);
1515
1516int freeze_bdev(struct block_device *bdev);
1517int thaw_bdev(struct block_device *bdev);
1518
1519struct io_comp_batch {
1520 struct request *req_list;
1521 bool need_ts;
1522 void (*complete)(struct io_comp_batch *);
1523};
1524
1525#define DEFINE_IO_COMP_BATCH(name) struct io_comp_batch name = { }
1526
1527#endif /* _LINUX_BLKDEV_H */