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 #include <linux/file.h>
28
29 struct module;
30 struct request_queue;
31 struct elevator_queue;
32 struct blk_trace;
33 struct request;
34 struct sg_io_hdr;
35 struct blkcg_gq;
36 struct blk_flush_queue;
37 struct kiocb;
38 struct pr_ops;
39 struct rq_qos;
40 struct blk_queue_stats;
41 struct blk_stat_callback;
42 struct blk_crypto_profile;
43
44 extern const struct device_type disk_type;
45 extern const struct device_type part_type;
46 extern const struct class block_class;
47
48 /*
49 * Maximum number of blkcg policies allowed to be registered concurrently.
50 * Defined here to simplify include dependency.
51 */
52 #define BLKCG_MAX_POLS 6
53
54 #define DISK_MAX_PARTS 256
55 #define DISK_NAME_LEN 32
56
57 #define PARTITION_META_INFO_VOLNAMELTH 64
58 /*
59 * Enough for the string representation of any kind of UUID plus NULL.
60 * EFI UUID is 36 characters. MSDOS UUID is 11 characters.
61 */
62 #define PARTITION_META_INFO_UUIDLTH (UUID_STRING_LEN + 1)
63
64 struct partition_meta_info {
65 char uuid[PARTITION_META_INFO_UUIDLTH];
66 u8 volname[PARTITION_META_INFO_VOLNAMELTH];
67 };
68
69 /**
70 * DOC: genhd capability flags
71 *
72 * ``GENHD_FL_REMOVABLE``: indicates that the block device gives access to
73 * removable media. When set, the device remains present even when media is not
74 * inserted. Shall not be set for devices which are removed entirely when the
75 * media is removed.
76 *
77 * ``GENHD_FL_HIDDEN``: the block device is hidden; it doesn't produce events,
78 * doesn't appear in sysfs, and can't be opened from userspace or using
79 * blkdev_get*. Used for the underlying components of multipath devices.
80 *
81 * ``GENHD_FL_NO_PART``: partition support is disabled. The kernel will not
82 * scan for partitions from add_disk, and users can't add partitions manually.
83 *
84 */
85 enum {
86 GENHD_FL_REMOVABLE = 1 << 0,
87 GENHD_FL_HIDDEN = 1 << 1,
88 GENHD_FL_NO_PART = 1 << 2,
89 };
90
91 enum {
92 DISK_EVENT_MEDIA_CHANGE = 1 << 0, /* media changed */
93 DISK_EVENT_EJECT_REQUEST = 1 << 1, /* eject requested */
94 };
95
96 enum {
97 /* Poll even if events_poll_msecs is unset */
98 DISK_EVENT_FLAG_POLL = 1 << 0,
99 /* Forward events to udev */
100 DISK_EVENT_FLAG_UEVENT = 1 << 1,
101 /* Block event polling when open for exclusive write */
102 DISK_EVENT_FLAG_BLOCK_ON_EXCL_WRITE = 1 << 2,
103 };
104
105 struct disk_events;
106 struct badblocks;
107
108 struct blk_integrity {
109 const struct blk_integrity_profile *profile;
110 unsigned char flags;
111 unsigned char tuple_size;
112 unsigned char pi_offset;
113 unsigned char interval_exp;
114 unsigned char tag_size;
115 };
116
117 typedef unsigned int __bitwise blk_mode_t;
118
119 /* open for reading */
120 #define BLK_OPEN_READ ((__force blk_mode_t)(1 << 0))
121 /* open for writing */
122 #define BLK_OPEN_WRITE ((__force blk_mode_t)(1 << 1))
123 /* open exclusively (vs other exclusive openers */
124 #define BLK_OPEN_EXCL ((__force blk_mode_t)(1 << 2))
125 /* opened with O_NDELAY */
126 #define BLK_OPEN_NDELAY ((__force blk_mode_t)(1 << 3))
127 /* open for "writes" only for ioctls (specialy hack for floppy.c) */
128 #define BLK_OPEN_WRITE_IOCTL ((__force blk_mode_t)(1 << 4))
129 /* open is exclusive wrt all other BLK_OPEN_WRITE opens to the device */
130 #define BLK_OPEN_RESTRICT_WRITES ((__force blk_mode_t)(1 << 5))
131 /* return partition scanning errors */
132 #define BLK_OPEN_STRICT_SCAN ((__force blk_mode_t)(1 << 6))
133
134 struct gendisk {
135 /*
136 * major/first_minor/minors should not be set by any new driver, the
137 * block core will take care of allocating them automatically.
138 */
139 int major;
140 int first_minor;
141 int minors;
142
143 char disk_name[DISK_NAME_LEN]; /* name of major driver */
144
145 unsigned short events; /* supported events */
146 unsigned short event_flags; /* flags related to event processing */
147
148 struct xarray part_tbl;
149 struct block_device *part0;
150
151 const struct block_device_operations *fops;
152 struct request_queue *queue;
153 void *private_data;
154
155 struct bio_set bio_split;
156
157 int flags;
158 unsigned long state;
159 #define GD_NEED_PART_SCAN 0
160 #define GD_READ_ONLY 1
161 #define GD_DEAD 2
162 #define GD_NATIVE_CAPACITY 3
163 #define GD_ADDED 4
164 #define GD_SUPPRESS_PART_SCAN 5
165 #define GD_OWNS_QUEUE 6
166
167 struct mutex open_mutex; /* open/close mutex */
168 unsigned open_partitions; /* number of open partitions */
169
170 struct backing_dev_info *bdi;
171 struct kobject queue_kobj; /* the queue/ directory */
172 struct kobject *slave_dir;
173 #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED
174 struct list_head slave_bdevs;
175 #endif
176 struct timer_rand_state *random;
177 atomic_t sync_io; /* RAID */
178 struct disk_events *ev;
179
180 #ifdef CONFIG_BLK_DEV_ZONED
181 /*
182 * Zoned block device information. Reads of this information must be
183 * protected with blk_queue_enter() / blk_queue_exit(). Modifying this
184 * information is only allowed while no requests are being processed.
185 * See also blk_mq_freeze_queue() and blk_mq_unfreeze_queue().
186 */
187 unsigned int nr_zones;
188 unsigned int zone_capacity;
189 unsigned long *conv_zones_bitmap;
190 unsigned int zone_wplugs_hash_bits;
191 spinlock_t zone_wplugs_lock;
192 struct mempool_s *zone_wplugs_pool;
193 struct hlist_head *zone_wplugs_hash;
194 struct list_head zone_wplugs_err_list;
195 struct work_struct zone_wplugs_work;
196 struct workqueue_struct *zone_wplugs_wq;
197 #endif /* CONFIG_BLK_DEV_ZONED */
198
199 #if IS_ENABLED(CONFIG_CDROM)
200 struct cdrom_device_info *cdi;
201 #endif
202 int node_id;
203 struct badblocks *bb;
204 struct lockdep_map lockdep_map;
205 u64 diskseq;
206 blk_mode_t open_mode;
207
208 /*
209 * Independent sector access ranges. This is always NULL for
210 * devices that do not have multiple independent access ranges.
211 */
212 struct blk_independent_access_ranges *ia_ranges;
213 };
214
disk_live(struct gendisk * disk)215 static inline bool disk_live(struct gendisk *disk)
216 {
217 return !inode_unhashed(disk->part0->bd_inode);
218 }
219
220 /**
221 * disk_openers - returns how many openers are there for a disk
222 * @disk: disk to check
223 *
224 * This returns the number of openers for a disk. Note that this value is only
225 * stable if disk->open_mutex is held.
226 *
227 * Note: Due to a quirk in the block layer open code, each open partition is
228 * only counted once even if there are multiple openers.
229 */
disk_openers(struct gendisk * disk)230 static inline unsigned int disk_openers(struct gendisk *disk)
231 {
232 return atomic_read(&disk->part0->bd_openers);
233 }
234
235 /**
236 * disk_has_partscan - return %true if partition scanning is enabled on a disk
237 * @disk: disk to check
238 *
239 * Returns %true if partitions scanning is enabled for @disk, or %false if
240 * partition scanning is disabled either permanently or temporarily.
241 */
disk_has_partscan(struct gendisk * disk)242 static inline bool disk_has_partscan(struct gendisk *disk)
243 {
244 return !(disk->flags & (GENHD_FL_NO_PART | GENHD_FL_HIDDEN)) &&
245 !test_bit(GD_SUPPRESS_PART_SCAN, &disk->state);
246 }
247
248 /*
249 * The gendisk is refcounted by the part0 block_device, and the bd_device
250 * therein is also used for device model presentation in sysfs.
251 */
252 #define dev_to_disk(device) \
253 (dev_to_bdev(device)->bd_disk)
254 #define disk_to_dev(disk) \
255 (&((disk)->part0->bd_device))
256
257 #if IS_REACHABLE(CONFIG_CDROM)
258 #define disk_to_cdi(disk) ((disk)->cdi)
259 #else
260 #define disk_to_cdi(disk) NULL
261 #endif
262
disk_devt(struct gendisk * disk)263 static inline dev_t disk_devt(struct gendisk *disk)
264 {
265 return MKDEV(disk->major, disk->first_minor);
266 }
267
blk_validate_block_size(unsigned long bsize)268 static inline int blk_validate_block_size(unsigned long bsize)
269 {
270 if (bsize < 512 || bsize > PAGE_SIZE || !is_power_of_2(bsize))
271 return -EINVAL;
272
273 return 0;
274 }
275
blk_op_is_passthrough(blk_opf_t op)276 static inline bool blk_op_is_passthrough(blk_opf_t op)
277 {
278 op &= REQ_OP_MASK;
279 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
280 }
281
282 /*
283 * BLK_BOUNCE_NONE: never bounce (default)
284 * BLK_BOUNCE_HIGH: bounce all highmem pages
285 */
286 enum blk_bounce {
287 BLK_BOUNCE_NONE,
288 BLK_BOUNCE_HIGH,
289 };
290
291 struct queue_limits {
292 enum blk_bounce bounce;
293 unsigned long seg_boundary_mask;
294 unsigned long virt_boundary_mask;
295
296 unsigned int max_hw_sectors;
297 unsigned int max_dev_sectors;
298 unsigned int chunk_sectors;
299 unsigned int max_sectors;
300 unsigned int max_user_sectors;
301 unsigned int max_segment_size;
302 unsigned int physical_block_size;
303 unsigned int logical_block_size;
304 unsigned int alignment_offset;
305 unsigned int io_min;
306 unsigned int io_opt;
307 unsigned int max_discard_sectors;
308 unsigned int max_hw_discard_sectors;
309 unsigned int max_user_discard_sectors;
310 unsigned int max_secure_erase_sectors;
311 unsigned int max_write_zeroes_sectors;
312 unsigned int max_zone_append_sectors;
313 unsigned int discard_granularity;
314 unsigned int discard_alignment;
315 unsigned int zone_write_granularity;
316
317 unsigned short max_segments;
318 unsigned short max_integrity_segments;
319 unsigned short max_discard_segments;
320
321 unsigned char misaligned;
322 unsigned char discard_misaligned;
323 unsigned char raid_partial_stripes_expensive;
324 bool zoned;
325 unsigned int max_open_zones;
326 unsigned int max_active_zones;
327
328 /*
329 * Drivers that set dma_alignment to less than 511 must be prepared to
330 * handle individual bvec's that are not a multiple of a SECTOR_SIZE
331 * due to possible offsets.
332 */
333 unsigned int dma_alignment;
334 };
335
336 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
337 void *data);
338
339 void disk_set_zoned(struct gendisk *disk);
340
341 #define BLK_ALL_ZONES ((unsigned int)-1)
342 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
343 unsigned int nr_zones, report_zones_cb cb, void *data);
344 int blkdev_zone_mgmt(struct block_device *bdev, enum req_op op,
345 sector_t sectors, sector_t nr_sectors);
346 int blk_revalidate_disk_zones(struct gendisk *disk);
347
348 /*
349 * Independent access ranges: struct blk_independent_access_range describes
350 * a range of contiguous sectors that can be accessed using device command
351 * execution resources that are independent from the resources used for
352 * other access ranges. This is typically found with single-LUN multi-actuator
353 * HDDs where each access range is served by a different set of heads.
354 * The set of independent ranges supported by the device is defined using
355 * struct blk_independent_access_ranges. The independent ranges must not overlap
356 * and must include all sectors within the disk capacity (no sector holes
357 * allowed).
358 * For a device with multiple ranges, requests targeting sectors in different
359 * ranges can be executed in parallel. A request can straddle an access range
360 * boundary.
361 */
362 struct blk_independent_access_range {
363 struct kobject kobj;
364 sector_t sector;
365 sector_t nr_sectors;
366 };
367
368 struct blk_independent_access_ranges {
369 struct kobject kobj;
370 bool sysfs_registered;
371 unsigned int nr_ia_ranges;
372 struct blk_independent_access_range ia_range[];
373 };
374
375 struct request_queue {
376 /*
377 * The queue owner gets to use this for whatever they like.
378 * ll_rw_blk doesn't touch it.
379 */
380 void *queuedata;
381
382 struct elevator_queue *elevator;
383
384 const struct blk_mq_ops *mq_ops;
385
386 /* sw queues */
387 struct blk_mq_ctx __percpu *queue_ctx;
388
389 /*
390 * various queue flags, see QUEUE_* below
391 */
392 unsigned long queue_flags;
393
394 unsigned int rq_timeout;
395
396 unsigned int queue_depth;
397
398 refcount_t refs;
399
400 /* hw dispatch queues */
401 unsigned int nr_hw_queues;
402 struct xarray hctx_table;
403
404 struct percpu_ref q_usage_counter;
405
406 struct request *last_merge;
407
408 spinlock_t queue_lock;
409
410 int quiesce_depth;
411
412 struct gendisk *disk;
413
414 /*
415 * mq queue kobject
416 */
417 struct kobject *mq_kobj;
418
419 struct queue_limits limits;
420
421 #ifdef CONFIG_BLK_DEV_INTEGRITY
422 struct blk_integrity integrity;
423 #endif /* CONFIG_BLK_DEV_INTEGRITY */
424
425 #ifdef CONFIG_PM
426 struct device *dev;
427 enum rpm_status rpm_status;
428 #endif
429
430 /*
431 * Number of contexts that have called blk_set_pm_only(). If this
432 * counter is above zero then only RQF_PM requests are processed.
433 */
434 atomic_t pm_only;
435
436 struct blk_queue_stats *stats;
437 struct rq_qos *rq_qos;
438 struct mutex rq_qos_mutex;
439
440 /*
441 * ida allocated id for this queue. Used to index queues from
442 * ioctx.
443 */
444 int id;
445
446 unsigned int dma_pad_mask;
447
448 /*
449 * queue settings
450 */
451 unsigned long nr_requests; /* Max # of requests */
452
453 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
454 struct blk_crypto_profile *crypto_profile;
455 struct kobject *crypto_kobject;
456 #endif
457
458 struct timer_list timeout;
459 struct work_struct timeout_work;
460
461 atomic_t nr_active_requests_shared_tags;
462
463 struct blk_mq_tags *sched_shared_tags;
464
465 struct list_head icq_list;
466 #ifdef CONFIG_BLK_CGROUP
467 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
468 struct blkcg_gq *root_blkg;
469 struct list_head blkg_list;
470 struct mutex blkcg_mutex;
471 #endif
472
473 int node;
474
475 spinlock_t requeue_lock;
476 struct list_head requeue_list;
477 struct delayed_work requeue_work;
478
479 #ifdef CONFIG_BLK_DEV_IO_TRACE
480 struct blk_trace __rcu *blk_trace;
481 #endif
482 /*
483 * for flush operations
484 */
485 struct blk_flush_queue *fq;
486 struct list_head flush_list;
487
488 struct mutex sysfs_lock;
489 struct mutex sysfs_dir_lock;
490 struct mutex limits_lock;
491
492 /*
493 * for reusing dead hctx instance in case of updating
494 * nr_hw_queues
495 */
496 struct list_head unused_hctx_list;
497 spinlock_t unused_hctx_lock;
498
499 int mq_freeze_depth;
500
501 #ifdef CONFIG_BLK_DEV_THROTTLING
502 /* Throttle data */
503 struct throtl_data *td;
504 #endif
505 struct rcu_head rcu_head;
506 wait_queue_head_t mq_freeze_wq;
507 /*
508 * Protect concurrent access to q_usage_counter by
509 * percpu_ref_kill() and percpu_ref_reinit().
510 */
511 struct mutex mq_freeze_lock;
512
513 struct blk_mq_tag_set *tag_set;
514 struct list_head tag_set_list;
515
516 struct dentry *debugfs_dir;
517 struct dentry *sched_debugfs_dir;
518 struct dentry *rqos_debugfs_dir;
519 /*
520 * Serializes all debugfs metadata operations using the above dentries.
521 */
522 struct mutex debugfs_mutex;
523
524 bool mq_sysfs_init_done;
525 };
526
527 /* Keep blk_queue_flag_name[] in sync with the definitions below */
528 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
529 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
530 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
531 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
532 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
533 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
534 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
535 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
536 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
537 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
538 #define QUEUE_FLAG_SYNCHRONOUS 11 /* always completes in submit context */
539 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
540 #define QUEUE_FLAG_HW_WC 13 /* Write back caching supported */
541 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
542 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
543 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
544 #define QUEUE_FLAG_WC 17 /* Write back caching */
545 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
546 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
547 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
548 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
549 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
550 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
551 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
552 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
553 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
554 #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */
555 #define QUEUE_FLAG_SQ_SCHED 30 /* single queue style io dispatch */
556 #define QUEUE_FLAG_SKIP_TAGSET_QUIESCE 31 /* quiesce_tagset skip the queue*/
557
558 #define QUEUE_FLAG_MQ_DEFAULT ((1UL << QUEUE_FLAG_IO_STAT) | \
559 (1UL << QUEUE_FLAG_SAME_COMP) | \
560 (1UL << QUEUE_FLAG_NOWAIT))
561
562 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
563 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
564 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
565
566 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
567 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
568 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
569 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
570 #define blk_queue_noxmerges(q) \
571 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
572 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
573 #define blk_queue_stable_writes(q) \
574 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
575 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
576 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
577 #define blk_queue_zone_resetall(q) \
578 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
579 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
580 #define blk_queue_pci_p2pdma(q) \
581 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
582 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
583 #define blk_queue_rq_alloc_time(q) \
584 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
585 #else
586 #define blk_queue_rq_alloc_time(q) false
587 #endif
588
589 #define blk_noretry_request(rq) \
590 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
591 REQ_FAILFAST_DRIVER))
592 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
593 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
594 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
595 #define blk_queue_sq_sched(q) test_bit(QUEUE_FLAG_SQ_SCHED, &(q)->queue_flags)
596 #define blk_queue_skip_tagset_quiesce(q) \
597 test_bit(QUEUE_FLAG_SKIP_TAGSET_QUIESCE, &(q)->queue_flags)
598
599 extern void blk_set_pm_only(struct request_queue *q);
600 extern void blk_clear_pm_only(struct request_queue *q);
601
602 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
603
604 #define dma_map_bvec(dev, bv, dir, attrs) \
605 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
606 (dir), (attrs))
607
queue_is_mq(struct request_queue * q)608 static inline bool queue_is_mq(struct request_queue *q)
609 {
610 return q->mq_ops;
611 }
612
613 #ifdef CONFIG_PM
queue_rpm_status(struct request_queue * q)614 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
615 {
616 return q->rpm_status;
617 }
618 #else
queue_rpm_status(struct request_queue * q)619 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
620 {
621 return RPM_ACTIVE;
622 }
623 #endif
624
blk_queue_is_zoned(struct request_queue * q)625 static inline bool blk_queue_is_zoned(struct request_queue *q)
626 {
627 return IS_ENABLED(CONFIG_BLK_DEV_ZONED) && q->limits.zoned;
628 }
629
630 #ifdef CONFIG_BLK_DEV_ZONED
631 unsigned int bdev_nr_zones(struct block_device *bdev);
632
disk_nr_zones(struct gendisk * disk)633 static inline unsigned int disk_nr_zones(struct gendisk *disk)
634 {
635 return blk_queue_is_zoned(disk->queue) ? disk->nr_zones : 0;
636 }
637
disk_zone_no(struct gendisk * disk,sector_t sector)638 static inline unsigned int disk_zone_no(struct gendisk *disk, sector_t sector)
639 {
640 if (!blk_queue_is_zoned(disk->queue))
641 return 0;
642 return sector >> ilog2(disk->queue->limits.chunk_sectors);
643 }
644
disk_set_max_open_zones(struct gendisk * disk,unsigned int max_open_zones)645 static inline void disk_set_max_open_zones(struct gendisk *disk,
646 unsigned int max_open_zones)
647 {
648 disk->queue->limits.max_open_zones = max_open_zones;
649 }
650
disk_set_max_active_zones(struct gendisk * disk,unsigned int max_active_zones)651 static inline void disk_set_max_active_zones(struct gendisk *disk,
652 unsigned int max_active_zones)
653 {
654 disk->queue->limits.max_active_zones = max_active_zones;
655 }
656
bdev_max_open_zones(struct block_device * bdev)657 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
658 {
659 return bdev->bd_disk->queue->limits.max_open_zones;
660 }
661
bdev_max_active_zones(struct block_device * bdev)662 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
663 {
664 return bdev->bd_disk->queue->limits.max_active_zones;
665 }
666
667 bool blk_zone_plug_bio(struct bio *bio, unsigned int nr_segs);
668 #else /* CONFIG_BLK_DEV_ZONED */
bdev_nr_zones(struct block_device * bdev)669 static inline unsigned int bdev_nr_zones(struct block_device *bdev)
670 {
671 return 0;
672 }
673
disk_nr_zones(struct gendisk * disk)674 static inline unsigned int disk_nr_zones(struct gendisk *disk)
675 {
676 return 0;
677 }
disk_zone_no(struct gendisk * disk,sector_t sector)678 static inline unsigned int disk_zone_no(struct gendisk *disk, sector_t sector)
679 {
680 return 0;
681 }
bdev_max_open_zones(struct block_device * bdev)682 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
683 {
684 return 0;
685 }
686
bdev_max_active_zones(struct block_device * bdev)687 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
688 {
689 return 0;
690 }
blk_zone_plug_bio(struct bio * bio,unsigned int nr_segs)691 static inline bool blk_zone_plug_bio(struct bio *bio, unsigned int nr_segs)
692 {
693 return false;
694 }
695 #endif /* CONFIG_BLK_DEV_ZONED */
696
blk_queue_depth(struct request_queue * q)697 static inline unsigned int blk_queue_depth(struct request_queue *q)
698 {
699 if (q->queue_depth)
700 return q->queue_depth;
701
702 return q->nr_requests;
703 }
704
705 /*
706 * default timeout for SG_IO if none specified
707 */
708 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
709 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
710
711 /* This should not be used directly - use rq_for_each_segment */
712 #define for_each_bio(_bio) \
713 for (; _bio; _bio = _bio->bi_next)
714
715 int __must_check device_add_disk(struct device *parent, struct gendisk *disk,
716 const struct attribute_group **groups);
add_disk(struct gendisk * disk)717 static inline int __must_check add_disk(struct gendisk *disk)
718 {
719 return device_add_disk(NULL, disk, NULL);
720 }
721 void del_gendisk(struct gendisk *gp);
722 void invalidate_disk(struct gendisk *disk);
723 void set_disk_ro(struct gendisk *disk, bool read_only);
724 void disk_uevent(struct gendisk *disk, enum kobject_action action);
725
get_disk_ro(struct gendisk * disk)726 static inline int get_disk_ro(struct gendisk *disk)
727 {
728 return disk->part0->bd_read_only ||
729 test_bit(GD_READ_ONLY, &disk->state);
730 }
731
bdev_read_only(struct block_device * bdev)732 static inline int bdev_read_only(struct block_device *bdev)
733 {
734 return bdev->bd_read_only || get_disk_ro(bdev->bd_disk);
735 }
736
737 bool set_capacity_and_notify(struct gendisk *disk, sector_t size);
738 void disk_force_media_change(struct gendisk *disk);
739 void bdev_mark_dead(struct block_device *bdev, bool surprise);
740
741 void add_disk_randomness(struct gendisk *disk) __latent_entropy;
742 void rand_initialize_disk(struct gendisk *disk);
743
get_start_sect(struct block_device * bdev)744 static inline sector_t get_start_sect(struct block_device *bdev)
745 {
746 return bdev->bd_start_sect;
747 }
748
bdev_nr_sectors(struct block_device * bdev)749 static inline sector_t bdev_nr_sectors(struct block_device *bdev)
750 {
751 return bdev->bd_nr_sectors;
752 }
753
bdev_nr_bytes(struct block_device * bdev)754 static inline loff_t bdev_nr_bytes(struct block_device *bdev)
755 {
756 return (loff_t)bdev_nr_sectors(bdev) << SECTOR_SHIFT;
757 }
758
get_capacity(struct gendisk * disk)759 static inline sector_t get_capacity(struct gendisk *disk)
760 {
761 return bdev_nr_sectors(disk->part0);
762 }
763
sb_bdev_nr_blocks(struct super_block * sb)764 static inline u64 sb_bdev_nr_blocks(struct super_block *sb)
765 {
766 return bdev_nr_sectors(sb->s_bdev) >>
767 (sb->s_blocksize_bits - SECTOR_SHIFT);
768 }
769
770 int bdev_disk_changed(struct gendisk *disk, bool invalidate);
771
772 void put_disk(struct gendisk *disk);
773 struct gendisk *__blk_alloc_disk(struct queue_limits *lim, int node,
774 struct lock_class_key *lkclass);
775
776 /**
777 * blk_alloc_disk - allocate a gendisk structure
778 * @lim: queue limits to be used for this disk.
779 * @node_id: numa node to allocate on
780 *
781 * Allocate and pre-initialize a gendisk structure for use with BIO based
782 * drivers.
783 *
784 * Returns an ERR_PTR on error, else the allocated disk.
785 *
786 * Context: can sleep
787 */
788 #define blk_alloc_disk(lim, node_id) \
789 ({ \
790 static struct lock_class_key __key; \
791 \
792 __blk_alloc_disk(lim, node_id, &__key); \
793 })
794
795 int __register_blkdev(unsigned int major, const char *name,
796 void (*probe)(dev_t devt));
797 #define register_blkdev(major, name) \
798 __register_blkdev(major, name, NULL)
799 void unregister_blkdev(unsigned int major, const char *name);
800
801 bool disk_check_media_change(struct gendisk *disk);
802 void set_capacity(struct gendisk *disk, sector_t size);
803
804 #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED
805 int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
806 void bd_unlink_disk_holder(struct block_device *bdev, struct gendisk *disk);
807 #else
bd_link_disk_holder(struct block_device * bdev,struct gendisk * disk)808 static inline int bd_link_disk_holder(struct block_device *bdev,
809 struct gendisk *disk)
810 {
811 return 0;
812 }
bd_unlink_disk_holder(struct block_device * bdev,struct gendisk * disk)813 static inline void bd_unlink_disk_holder(struct block_device *bdev,
814 struct gendisk *disk)
815 {
816 }
817 #endif /* CONFIG_BLOCK_HOLDER_DEPRECATED */
818
819 dev_t part_devt(struct gendisk *disk, u8 partno);
820 void inc_diskseq(struct gendisk *disk);
821 void blk_request_module(dev_t devt);
822
823 extern int blk_register_queue(struct gendisk *disk);
824 extern void blk_unregister_queue(struct gendisk *disk);
825 void submit_bio_noacct(struct bio *bio);
826 struct bio *bio_split_to_limits(struct bio *bio);
827
828 extern int blk_lld_busy(struct request_queue *q);
829 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
830 extern void blk_queue_exit(struct request_queue *q);
831 extern void blk_sync_queue(struct request_queue *q);
832
833 /* Helper to convert REQ_OP_XXX to its string format XXX */
834 extern const char *blk_op_str(enum req_op op);
835
836 int blk_status_to_errno(blk_status_t status);
837 blk_status_t errno_to_blk_status(int errno);
838 const char *blk_status_to_str(blk_status_t status);
839
840 /* only poll the hardware once, don't continue until a completion was found */
841 #define BLK_POLL_ONESHOT (1 << 0)
842 int bio_poll(struct bio *bio, struct io_comp_batch *iob, unsigned int flags);
843 int iocb_bio_iopoll(struct kiocb *kiocb, struct io_comp_batch *iob,
844 unsigned int flags);
845
bdev_get_queue(struct block_device * bdev)846 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
847 {
848 return bdev->bd_queue; /* this is never NULL */
849 }
850
851 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
852 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
853
bio_zone_no(struct bio * bio)854 static inline unsigned int bio_zone_no(struct bio *bio)
855 {
856 return disk_zone_no(bio->bi_bdev->bd_disk, bio->bi_iter.bi_sector);
857 }
858
bio_straddles_zones(struct bio * bio)859 static inline bool bio_straddles_zones(struct bio *bio)
860 {
861 return bio_sectors(bio) &&
862 bio_zone_no(bio) !=
863 disk_zone_no(bio->bi_bdev->bd_disk, bio_end_sector(bio) - 1);
864 }
865
866 /*
867 * Return how much of the chunk is left to be used for I/O at a given offset.
868 */
blk_chunk_sectors_left(sector_t offset,unsigned int chunk_sectors)869 static inline unsigned int blk_chunk_sectors_left(sector_t offset,
870 unsigned int chunk_sectors)
871 {
872 if (unlikely(!is_power_of_2(chunk_sectors)))
873 return chunk_sectors - sector_div(offset, chunk_sectors);
874 return chunk_sectors - (offset & (chunk_sectors - 1));
875 }
876
877 /**
878 * queue_limits_start_update - start an atomic update of queue limits
879 * @q: queue to update
880 *
881 * This functions starts an atomic update of the queue limits. It takes a lock
882 * to prevent other updates and returns a snapshot of the current limits that
883 * the caller can modify. The caller must call queue_limits_commit_update()
884 * to finish the update.
885 *
886 * Context: process context. The caller must have frozen the queue or ensured
887 * that there is outstanding I/O by other means.
888 */
889 static inline struct queue_limits
queue_limits_start_update(struct request_queue * q)890 queue_limits_start_update(struct request_queue *q)
891 __acquires(q->limits_lock)
892 {
893 mutex_lock(&q->limits_lock);
894 return q->limits;
895 }
896 int queue_limits_commit_update(struct request_queue *q,
897 struct queue_limits *lim);
898 int queue_limits_set(struct request_queue *q, struct queue_limits *lim);
899
900 /**
901 * queue_limits_cancel_update - cancel an atomic update of queue limits
902 * @q: queue to update
903 *
904 * This functions cancels an atomic update of the queue limits started by
905 * queue_limits_start_update() and should be used when an error occurs after
906 * starting update.
907 */
queue_limits_cancel_update(struct request_queue * q)908 static inline void queue_limits_cancel_update(struct request_queue *q)
909 {
910 mutex_unlock(&q->limits_lock);
911 }
912
913 /*
914 * Access functions for manipulating queue properties
915 */
916 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
917 void blk_queue_max_secure_erase_sectors(struct request_queue *q,
918 unsigned int max_sectors);
919 extern void blk_queue_max_discard_sectors(struct request_queue *q,
920 unsigned int max_discard_sectors);
921 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
922 unsigned int max_write_same_sectors);
923 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
924 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
925 unsigned int max_zone_append_sectors);
926 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
927 void blk_queue_zone_write_granularity(struct request_queue *q,
928 unsigned int size);
929 extern void blk_queue_alignment_offset(struct request_queue *q,
930 unsigned int alignment);
931 void disk_update_readahead(struct gendisk *disk);
932 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
933 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
934 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
935 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
936 extern void blk_set_stacking_limits(struct queue_limits *lim);
937 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
938 sector_t offset);
939 void queue_limits_stack_bdev(struct queue_limits *t, struct block_device *bdev,
940 sector_t offset, const char *pfx);
941 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
942 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
943 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
944
945 struct blk_independent_access_ranges *
946 disk_alloc_independent_access_ranges(struct gendisk *disk, int nr_ia_ranges);
947 void disk_set_independent_access_ranges(struct gendisk *disk,
948 struct blk_independent_access_ranges *iars);
949
950 bool __must_check blk_get_queue(struct request_queue *);
951 extern void blk_put_queue(struct request_queue *);
952
953 void blk_mark_disk_dead(struct gendisk *disk);
954
955 #ifdef CONFIG_BLOCK
956 /*
957 * blk_plug permits building a queue of related requests by holding the I/O
958 * fragments for a short period. This allows merging of sequential requests
959 * into single larger request. As the requests are moved from a per-task list to
960 * the device's request_queue in a batch, this results in improved scalability
961 * as the lock contention for request_queue lock is reduced.
962 *
963 * It is ok not to disable preemption when adding the request to the plug list
964 * or when attempting a merge. For details, please see schedule() where
965 * blk_flush_plug() is called.
966 */
967 struct blk_plug {
968 struct request *mq_list; /* blk-mq requests */
969
970 /* if ios_left is > 1, we can batch tag/rq allocations */
971 struct request *cached_rq;
972 u64 cur_ktime;
973 unsigned short nr_ios;
974
975 unsigned short rq_count;
976
977 bool multiple_queues;
978 bool has_elevator;
979
980 struct list_head cb_list; /* md requires an unplug callback */
981 };
982
983 struct blk_plug_cb;
984 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
985 struct blk_plug_cb {
986 struct list_head list;
987 blk_plug_cb_fn callback;
988 void *data;
989 };
990 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
991 void *data, int size);
992 extern void blk_start_plug(struct blk_plug *);
993 extern void blk_start_plug_nr_ios(struct blk_plug *, unsigned short);
994 extern void blk_finish_plug(struct blk_plug *);
995
996 void __blk_flush_plug(struct blk_plug *plug, bool from_schedule);
blk_flush_plug(struct blk_plug * plug,bool async)997 static inline void blk_flush_plug(struct blk_plug *plug, bool async)
998 {
999 if (plug)
1000 __blk_flush_plug(plug, async);
1001 }
1002
1003 /*
1004 * tsk == current here
1005 */
blk_plug_invalidate_ts(struct task_struct * tsk)1006 static inline void blk_plug_invalidate_ts(struct task_struct *tsk)
1007 {
1008 struct blk_plug *plug = tsk->plug;
1009
1010 if (plug)
1011 plug->cur_ktime = 0;
1012 current->flags &= ~PF_BLOCK_TS;
1013 }
1014
1015 int blkdev_issue_flush(struct block_device *bdev);
1016 long nr_blockdev_pages(void);
1017 #else /* CONFIG_BLOCK */
1018 struct blk_plug {
1019 };
1020
blk_start_plug_nr_ios(struct blk_plug * plug,unsigned short nr_ios)1021 static inline void blk_start_plug_nr_ios(struct blk_plug *plug,
1022 unsigned short nr_ios)
1023 {
1024 }
1025
blk_start_plug(struct blk_plug * plug)1026 static inline void blk_start_plug(struct blk_plug *plug)
1027 {
1028 }
1029
blk_finish_plug(struct blk_plug * plug)1030 static inline void blk_finish_plug(struct blk_plug *plug)
1031 {
1032 }
1033
blk_flush_plug(struct blk_plug * plug,bool async)1034 static inline void blk_flush_plug(struct blk_plug *plug, bool async)
1035 {
1036 }
1037
blk_plug_invalidate_ts(struct task_struct * tsk)1038 static inline void blk_plug_invalidate_ts(struct task_struct *tsk)
1039 {
1040 }
1041
blkdev_issue_flush(struct block_device * bdev)1042 static inline int blkdev_issue_flush(struct block_device *bdev)
1043 {
1044 return 0;
1045 }
1046
nr_blockdev_pages(void)1047 static inline long nr_blockdev_pages(void)
1048 {
1049 return 0;
1050 }
1051 #endif /* CONFIG_BLOCK */
1052
1053 extern void blk_io_schedule(void);
1054
1055 int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1056 sector_t nr_sects, gfp_t gfp_mask);
1057 int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1058 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop);
1059 int blkdev_issue_secure_erase(struct block_device *bdev, sector_t sector,
1060 sector_t nr_sects, gfp_t gfp);
1061
1062 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1063 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1064
1065 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1066 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1067 unsigned flags);
1068 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1069 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1070
sb_issue_discard(struct super_block * sb,sector_t block,sector_t nr_blocks,gfp_t gfp_mask,unsigned long flags)1071 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1072 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1073 {
1074 return blkdev_issue_discard(sb->s_bdev,
1075 block << (sb->s_blocksize_bits -
1076 SECTOR_SHIFT),
1077 nr_blocks << (sb->s_blocksize_bits -
1078 SECTOR_SHIFT),
1079 gfp_mask);
1080 }
sb_issue_zeroout(struct super_block * sb,sector_t block,sector_t nr_blocks,gfp_t gfp_mask)1081 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1082 sector_t nr_blocks, gfp_t gfp_mask)
1083 {
1084 return blkdev_issue_zeroout(sb->s_bdev,
1085 block << (sb->s_blocksize_bits -
1086 SECTOR_SHIFT),
1087 nr_blocks << (sb->s_blocksize_bits -
1088 SECTOR_SHIFT),
1089 gfp_mask, 0);
1090 }
1091
bdev_is_partition(struct block_device * bdev)1092 static inline bool bdev_is_partition(struct block_device *bdev)
1093 {
1094 return bdev->bd_partno;
1095 }
1096
1097 enum blk_default_limits {
1098 BLK_MAX_SEGMENTS = 128,
1099 BLK_SAFE_MAX_SECTORS = 255,
1100 BLK_MAX_SEGMENT_SIZE = 65536,
1101 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1102 };
1103
1104 /*
1105 * Default upper limit for the software max_sectors limit used for
1106 * regular file system I/O. This can be increased through sysfs.
1107 *
1108 * Not to be confused with the max_hw_sector limit that is entirely
1109 * controlled by the driver, usually based on hardware limits.
1110 */
1111 #define BLK_DEF_MAX_SECTORS_CAP 2560u
1112
queue_segment_boundary(const struct request_queue * q)1113 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1114 {
1115 return q->limits.seg_boundary_mask;
1116 }
1117
queue_virt_boundary(const struct request_queue * q)1118 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1119 {
1120 return q->limits.virt_boundary_mask;
1121 }
1122
queue_max_sectors(const struct request_queue * q)1123 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1124 {
1125 return q->limits.max_sectors;
1126 }
1127
queue_max_bytes(struct request_queue * q)1128 static inline unsigned int queue_max_bytes(struct request_queue *q)
1129 {
1130 return min_t(unsigned int, queue_max_sectors(q), INT_MAX >> 9) << 9;
1131 }
1132
queue_max_hw_sectors(const struct request_queue * q)1133 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1134 {
1135 return q->limits.max_hw_sectors;
1136 }
1137
queue_max_segments(const struct request_queue * q)1138 static inline unsigned short queue_max_segments(const struct request_queue *q)
1139 {
1140 return q->limits.max_segments;
1141 }
1142
queue_max_discard_segments(const struct request_queue * q)1143 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1144 {
1145 return q->limits.max_discard_segments;
1146 }
1147
queue_max_segment_size(const struct request_queue * q)1148 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1149 {
1150 return q->limits.max_segment_size;
1151 }
1152
queue_limits_max_zone_append_sectors(struct queue_limits * l)1153 static inline unsigned int queue_limits_max_zone_append_sectors(struct queue_limits *l)
1154 {
1155 unsigned int max_sectors = min(l->chunk_sectors, l->max_hw_sectors);
1156
1157 return min_not_zero(l->max_zone_append_sectors, max_sectors);
1158 }
1159
queue_max_zone_append_sectors(struct request_queue * q)1160 static inline unsigned int queue_max_zone_append_sectors(struct request_queue *q)
1161 {
1162 if (!blk_queue_is_zoned(q))
1163 return 0;
1164
1165 return queue_limits_max_zone_append_sectors(&q->limits);
1166 }
1167
queue_emulates_zone_append(struct request_queue * q)1168 static inline bool queue_emulates_zone_append(struct request_queue *q)
1169 {
1170 return blk_queue_is_zoned(q) && !q->limits.max_zone_append_sectors;
1171 }
1172
bdev_emulates_zone_append(struct block_device * bdev)1173 static inline bool bdev_emulates_zone_append(struct block_device *bdev)
1174 {
1175 return queue_emulates_zone_append(bdev_get_queue(bdev));
1176 }
1177
1178 static inline unsigned int
bdev_max_zone_append_sectors(struct block_device * bdev)1179 bdev_max_zone_append_sectors(struct block_device *bdev)
1180 {
1181 return queue_max_zone_append_sectors(bdev_get_queue(bdev));
1182 }
1183
bdev_max_segments(struct block_device * bdev)1184 static inline unsigned int bdev_max_segments(struct block_device *bdev)
1185 {
1186 return queue_max_segments(bdev_get_queue(bdev));
1187 }
1188
queue_logical_block_size(const struct request_queue * q)1189 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1190 {
1191 int retval = 512;
1192
1193 if (q && q->limits.logical_block_size)
1194 retval = q->limits.logical_block_size;
1195
1196 return retval;
1197 }
1198
bdev_logical_block_size(struct block_device * bdev)1199 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1200 {
1201 return queue_logical_block_size(bdev_get_queue(bdev));
1202 }
1203
queue_physical_block_size(const struct request_queue * q)1204 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1205 {
1206 return q->limits.physical_block_size;
1207 }
1208
bdev_physical_block_size(struct block_device * bdev)1209 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1210 {
1211 return queue_physical_block_size(bdev_get_queue(bdev));
1212 }
1213
queue_io_min(const struct request_queue * q)1214 static inline unsigned int queue_io_min(const struct request_queue *q)
1215 {
1216 return q->limits.io_min;
1217 }
1218
bdev_io_min(struct block_device * bdev)1219 static inline int bdev_io_min(struct block_device *bdev)
1220 {
1221 return queue_io_min(bdev_get_queue(bdev));
1222 }
1223
queue_io_opt(const struct request_queue * q)1224 static inline unsigned int queue_io_opt(const struct request_queue *q)
1225 {
1226 return q->limits.io_opt;
1227 }
1228
bdev_io_opt(struct block_device * bdev)1229 static inline int bdev_io_opt(struct block_device *bdev)
1230 {
1231 return queue_io_opt(bdev_get_queue(bdev));
1232 }
1233
1234 static inline unsigned int
queue_zone_write_granularity(const struct request_queue * q)1235 queue_zone_write_granularity(const struct request_queue *q)
1236 {
1237 return q->limits.zone_write_granularity;
1238 }
1239
1240 static inline unsigned int
bdev_zone_write_granularity(struct block_device * bdev)1241 bdev_zone_write_granularity(struct block_device *bdev)
1242 {
1243 return queue_zone_write_granularity(bdev_get_queue(bdev));
1244 }
1245
1246 int bdev_alignment_offset(struct block_device *bdev);
1247 unsigned int bdev_discard_alignment(struct block_device *bdev);
1248
bdev_max_discard_sectors(struct block_device * bdev)1249 static inline unsigned int bdev_max_discard_sectors(struct block_device *bdev)
1250 {
1251 return bdev_get_queue(bdev)->limits.max_discard_sectors;
1252 }
1253
bdev_discard_granularity(struct block_device * bdev)1254 static inline unsigned int bdev_discard_granularity(struct block_device *bdev)
1255 {
1256 return bdev_get_queue(bdev)->limits.discard_granularity;
1257 }
1258
1259 static inline unsigned int
bdev_max_secure_erase_sectors(struct block_device * bdev)1260 bdev_max_secure_erase_sectors(struct block_device *bdev)
1261 {
1262 return bdev_get_queue(bdev)->limits.max_secure_erase_sectors;
1263 }
1264
bdev_write_zeroes_sectors(struct block_device * bdev)1265 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1266 {
1267 struct request_queue *q = bdev_get_queue(bdev);
1268
1269 if (q)
1270 return q->limits.max_write_zeroes_sectors;
1271
1272 return 0;
1273 }
1274
bdev_nonrot(struct block_device * bdev)1275 static inline bool bdev_nonrot(struct block_device *bdev)
1276 {
1277 return blk_queue_nonrot(bdev_get_queue(bdev));
1278 }
1279
bdev_synchronous(struct block_device * bdev)1280 static inline bool bdev_synchronous(struct block_device *bdev)
1281 {
1282 return test_bit(QUEUE_FLAG_SYNCHRONOUS,
1283 &bdev_get_queue(bdev)->queue_flags);
1284 }
1285
bdev_stable_writes(struct block_device * bdev)1286 static inline bool bdev_stable_writes(struct block_device *bdev)
1287 {
1288 return test_bit(QUEUE_FLAG_STABLE_WRITES,
1289 &bdev_get_queue(bdev)->queue_flags);
1290 }
1291
bdev_write_cache(struct block_device * bdev)1292 static inline bool bdev_write_cache(struct block_device *bdev)
1293 {
1294 return test_bit(QUEUE_FLAG_WC, &bdev_get_queue(bdev)->queue_flags);
1295 }
1296
bdev_fua(struct block_device * bdev)1297 static inline bool bdev_fua(struct block_device *bdev)
1298 {
1299 return test_bit(QUEUE_FLAG_FUA, &bdev_get_queue(bdev)->queue_flags);
1300 }
1301
bdev_nowait(struct block_device * bdev)1302 static inline bool bdev_nowait(struct block_device *bdev)
1303 {
1304 return test_bit(QUEUE_FLAG_NOWAIT, &bdev_get_queue(bdev)->queue_flags);
1305 }
1306
bdev_is_zoned(struct block_device * bdev)1307 static inline bool bdev_is_zoned(struct block_device *bdev)
1308 {
1309 return blk_queue_is_zoned(bdev_get_queue(bdev));
1310 }
1311
bdev_zone_no(struct block_device * bdev,sector_t sec)1312 static inline unsigned int bdev_zone_no(struct block_device *bdev, sector_t sec)
1313 {
1314 return disk_zone_no(bdev->bd_disk, sec);
1315 }
1316
bdev_zone_sectors(struct block_device * bdev)1317 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1318 {
1319 struct request_queue *q = bdev_get_queue(bdev);
1320
1321 if (!blk_queue_is_zoned(q))
1322 return 0;
1323 return q->limits.chunk_sectors;
1324 }
1325
bdev_offset_from_zone_start(struct block_device * bdev,sector_t sector)1326 static inline sector_t bdev_offset_from_zone_start(struct block_device *bdev,
1327 sector_t sector)
1328 {
1329 return sector & (bdev_zone_sectors(bdev) - 1);
1330 }
1331
bio_offset_from_zone_start(struct bio * bio)1332 static inline sector_t bio_offset_from_zone_start(struct bio *bio)
1333 {
1334 return bdev_offset_from_zone_start(bio->bi_bdev,
1335 bio->bi_iter.bi_sector);
1336 }
1337
bdev_is_zone_start(struct block_device * bdev,sector_t sector)1338 static inline bool bdev_is_zone_start(struct block_device *bdev,
1339 sector_t sector)
1340 {
1341 return bdev_offset_from_zone_start(bdev, sector) == 0;
1342 }
1343
queue_dma_alignment(const struct request_queue * q)1344 static inline int queue_dma_alignment(const struct request_queue *q)
1345 {
1346 return q ? q->limits.dma_alignment : 511;
1347 }
1348
bdev_dma_alignment(struct block_device * bdev)1349 static inline unsigned int bdev_dma_alignment(struct block_device *bdev)
1350 {
1351 return queue_dma_alignment(bdev_get_queue(bdev));
1352 }
1353
bdev_iter_is_aligned(struct block_device * bdev,struct iov_iter * iter)1354 static inline bool bdev_iter_is_aligned(struct block_device *bdev,
1355 struct iov_iter *iter)
1356 {
1357 return iov_iter_is_aligned(iter, bdev_dma_alignment(bdev),
1358 bdev_logical_block_size(bdev) - 1);
1359 }
1360
blk_rq_aligned(struct request_queue * q,unsigned long addr,unsigned int len)1361 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1362 unsigned int len)
1363 {
1364 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1365 return !(addr & alignment) && !(len & alignment);
1366 }
1367
1368 /* assumes size > 256 */
blksize_bits(unsigned int size)1369 static inline unsigned int blksize_bits(unsigned int size)
1370 {
1371 return order_base_2(size >> SECTOR_SHIFT) + SECTOR_SHIFT;
1372 }
1373
block_size(struct block_device * bdev)1374 static inline unsigned int block_size(struct block_device *bdev)
1375 {
1376 return 1 << bdev->bd_inode->i_blkbits;
1377 }
1378
1379 int kblockd_schedule_work(struct work_struct *work);
1380 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1381
1382 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1383 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1384 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1385 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1386
1387 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1388
1389 bool blk_crypto_register(struct blk_crypto_profile *profile,
1390 struct request_queue *q);
1391
1392 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1393
blk_crypto_register(struct blk_crypto_profile * profile,struct request_queue * q)1394 static inline bool blk_crypto_register(struct blk_crypto_profile *profile,
1395 struct request_queue *q)
1396 {
1397 return true;
1398 }
1399
1400 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1401
1402 enum blk_unique_id {
1403 /* these match the Designator Types specified in SPC */
1404 BLK_UID_T10 = 1,
1405 BLK_UID_EUI64 = 2,
1406 BLK_UID_NAA = 3,
1407 };
1408
1409 struct block_device_operations {
1410 void (*submit_bio)(struct bio *bio);
1411 int (*poll_bio)(struct bio *bio, struct io_comp_batch *iob,
1412 unsigned int flags);
1413 int (*open)(struct gendisk *disk, blk_mode_t mode);
1414 void (*release)(struct gendisk *disk);
1415 int (*ioctl)(struct block_device *bdev, blk_mode_t mode,
1416 unsigned cmd, unsigned long arg);
1417 int (*compat_ioctl)(struct block_device *bdev, blk_mode_t mode,
1418 unsigned cmd, unsigned long arg);
1419 unsigned int (*check_events) (struct gendisk *disk,
1420 unsigned int clearing);
1421 void (*unlock_native_capacity) (struct gendisk *);
1422 int (*getgeo)(struct block_device *, struct hd_geometry *);
1423 int (*set_read_only)(struct block_device *bdev, bool ro);
1424 void (*free_disk)(struct gendisk *disk);
1425 /* this callback is with swap_lock and sometimes page table lock held */
1426 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1427 int (*report_zones)(struct gendisk *, sector_t sector,
1428 unsigned int nr_zones, report_zones_cb cb, void *data);
1429 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1430 /* returns the length of the identifier or a negative errno: */
1431 int (*get_unique_id)(struct gendisk *disk, u8 id[16],
1432 enum blk_unique_id id_type);
1433 struct module *owner;
1434 const struct pr_ops *pr_ops;
1435
1436 /*
1437 * Special callback for probing GPT entry at a given sector.
1438 * Needed by Android devices, used by GPT scanner and MMC blk
1439 * driver.
1440 */
1441 int (*alternative_gpt_sector)(struct gendisk *disk, sector_t *sector);
1442 };
1443
1444 #ifdef CONFIG_COMPAT
1445 extern int blkdev_compat_ptr_ioctl(struct block_device *, blk_mode_t,
1446 unsigned int, unsigned long);
1447 #else
1448 #define blkdev_compat_ptr_ioctl NULL
1449 #endif
1450
blk_wake_io_task(struct task_struct * waiter)1451 static inline void blk_wake_io_task(struct task_struct *waiter)
1452 {
1453 /*
1454 * If we're polling, the task itself is doing the completions. For
1455 * that case, we don't need to signal a wakeup, it's enough to just
1456 * mark us as RUNNING.
1457 */
1458 if (waiter == current)
1459 __set_current_state(TASK_RUNNING);
1460 else
1461 wake_up_process(waiter);
1462 }
1463
1464 unsigned long bdev_start_io_acct(struct block_device *bdev, enum req_op op,
1465 unsigned long start_time);
1466 void bdev_end_io_acct(struct block_device *bdev, enum req_op op,
1467 unsigned int sectors, unsigned long start_time);
1468
1469 unsigned long bio_start_io_acct(struct bio *bio);
1470 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1471 struct block_device *orig_bdev);
1472
1473 /**
1474 * bio_end_io_acct - end I/O accounting for bio based drivers
1475 * @bio: bio to end account for
1476 * @start_time: start time returned by bio_start_io_acct()
1477 */
bio_end_io_acct(struct bio * bio,unsigned long start_time)1478 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1479 {
1480 return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1481 }
1482
1483 int bdev_read_only(struct block_device *bdev);
1484 int set_blocksize(struct block_device *bdev, int size);
1485
1486 int lookup_bdev(const char *pathname, dev_t *dev);
1487
1488 void blkdev_show(struct seq_file *seqf, off_t offset);
1489
1490 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1491 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1492 #ifdef CONFIG_BLOCK
1493 #define BLKDEV_MAJOR_MAX 512
1494 #else
1495 #define BLKDEV_MAJOR_MAX 0
1496 #endif
1497
1498 struct blk_holder_ops {
1499 void (*mark_dead)(struct block_device *bdev, bool surprise);
1500
1501 /*
1502 * Sync the file system mounted on the block device.
1503 */
1504 void (*sync)(struct block_device *bdev);
1505
1506 /*
1507 * Freeze the file system mounted on the block device.
1508 */
1509 int (*freeze)(struct block_device *bdev);
1510
1511 /*
1512 * Thaw the file system mounted on the block device.
1513 */
1514 int (*thaw)(struct block_device *bdev);
1515 };
1516
1517 /*
1518 * For filesystems using @fs_holder_ops, the @holder argument passed to
1519 * helpers used to open and claim block devices via
1520 * bd_prepare_to_claim() must point to a superblock.
1521 */
1522 extern const struct blk_holder_ops fs_holder_ops;
1523
1524 /*
1525 * Return the correct open flags for blkdev_get_by_* for super block flags
1526 * as stored in sb->s_flags.
1527 */
1528 #define sb_open_mode(flags) \
1529 (BLK_OPEN_READ | BLK_OPEN_RESTRICT_WRITES | \
1530 (((flags) & SB_RDONLY) ? 0 : BLK_OPEN_WRITE))
1531
1532 struct file *bdev_file_open_by_dev(dev_t dev, blk_mode_t mode, void *holder,
1533 const struct blk_holder_ops *hops);
1534 struct file *bdev_file_open_by_path(const char *path, blk_mode_t mode,
1535 void *holder, const struct blk_holder_ops *hops);
1536 int bd_prepare_to_claim(struct block_device *bdev, void *holder,
1537 const struct blk_holder_ops *hops);
1538 void bd_abort_claiming(struct block_device *bdev, void *holder);
1539
1540 /* just for blk-cgroup, don't use elsewhere */
1541 struct block_device *blkdev_get_no_open(dev_t dev);
1542 void blkdev_put_no_open(struct block_device *bdev);
1543
1544 struct block_device *I_BDEV(struct inode *inode);
1545 struct block_device *file_bdev(struct file *bdev_file);
1546
1547 #ifdef CONFIG_BLOCK
1548 void invalidate_bdev(struct block_device *bdev);
1549 int sync_blockdev(struct block_device *bdev);
1550 int sync_blockdev_range(struct block_device *bdev, loff_t lstart, loff_t lend);
1551 int sync_blockdev_nowait(struct block_device *bdev);
1552 void sync_bdevs(bool wait);
1553 void bdev_statx_dioalign(struct inode *inode, struct kstat *stat);
1554 void printk_all_partitions(void);
1555 int __init early_lookup_bdev(const char *pathname, dev_t *dev);
1556 #else
invalidate_bdev(struct block_device * bdev)1557 static inline void invalidate_bdev(struct block_device *bdev)
1558 {
1559 }
sync_blockdev(struct block_device * bdev)1560 static inline int sync_blockdev(struct block_device *bdev)
1561 {
1562 return 0;
1563 }
sync_blockdev_nowait(struct block_device * bdev)1564 static inline int sync_blockdev_nowait(struct block_device *bdev)
1565 {
1566 return 0;
1567 }
sync_bdevs(bool wait)1568 static inline void sync_bdevs(bool wait)
1569 {
1570 }
bdev_statx_dioalign(struct inode * inode,struct kstat * stat)1571 static inline void bdev_statx_dioalign(struct inode *inode, struct kstat *stat)
1572 {
1573 }
printk_all_partitions(void)1574 static inline void printk_all_partitions(void)
1575 {
1576 }
early_lookup_bdev(const char * pathname,dev_t * dev)1577 static inline int early_lookup_bdev(const char *pathname, dev_t *dev)
1578 {
1579 return -EINVAL;
1580 }
1581 #endif /* CONFIG_BLOCK */
1582
1583 int bdev_freeze(struct block_device *bdev);
1584 int bdev_thaw(struct block_device *bdev);
1585 void bdev_fput(struct file *bdev_file);
1586
1587 struct io_comp_batch {
1588 struct request *req_list;
1589 bool need_ts;
1590 void (*complete)(struct io_comp_batch *);
1591 };
1592
1593 #define DEFINE_IO_COMP_BATCH(name) struct io_comp_batch name = { }
1594
1595 #endif /* _LINUX_BLKDEV_H */
1596