xref: /linux-6.15/include/linux/blkdev.h (revision ab7b884a)
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 
215 /**
216  * disk_openers - returns how many openers are there for a disk
217  * @disk: disk to check
218  *
219  * This returns the number of openers for a disk.  Note that this value is only
220  * stable if disk->open_mutex is held.
221  *
222  * Note: Due to a quirk in the block layer open code, each open partition is
223  * only counted once even if there are multiple openers.
224  */
225 static inline unsigned int disk_openers(struct gendisk *disk)
226 {
227 	return atomic_read(&disk->part0->bd_openers);
228 }
229 
230 /**
231  * disk_has_partscan - return %true if partition scanning is enabled on a disk
232  * @disk: disk to check
233  *
234  * Returns %true if partitions scanning is enabled for @disk, or %false if
235  * partition scanning is disabled either permanently or temporarily.
236  */
237 static inline bool disk_has_partscan(struct gendisk *disk)
238 {
239 	return !(disk->flags & (GENHD_FL_NO_PART | GENHD_FL_HIDDEN)) &&
240 		!test_bit(GD_SUPPRESS_PART_SCAN, &disk->state);
241 }
242 
243 /*
244  * The gendisk is refcounted by the part0 block_device, and the bd_device
245  * therein is also used for device model presentation in sysfs.
246  */
247 #define dev_to_disk(device) \
248 	(dev_to_bdev(device)->bd_disk)
249 #define disk_to_dev(disk) \
250 	(&((disk)->part0->bd_device))
251 
252 #if IS_REACHABLE(CONFIG_CDROM)
253 #define disk_to_cdi(disk)	((disk)->cdi)
254 #else
255 #define disk_to_cdi(disk)	NULL
256 #endif
257 
258 static inline dev_t disk_devt(struct gendisk *disk)
259 {
260 	return MKDEV(disk->major, disk->first_minor);
261 }
262 
263 static inline int blk_validate_block_size(unsigned long bsize)
264 {
265 	if (bsize < 512 || bsize > PAGE_SIZE || !is_power_of_2(bsize))
266 		return -EINVAL;
267 
268 	return 0;
269 }
270 
271 static inline bool blk_op_is_passthrough(blk_opf_t op)
272 {
273 	op &= REQ_OP_MASK;
274 	return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
275 }
276 
277 /*
278  * BLK_BOUNCE_NONE:	never bounce (default)
279  * BLK_BOUNCE_HIGH:	bounce all highmem pages
280  */
281 enum blk_bounce {
282 	BLK_BOUNCE_NONE,
283 	BLK_BOUNCE_HIGH,
284 };
285 
286 struct queue_limits {
287 	enum blk_bounce		bounce;
288 	unsigned long		seg_boundary_mask;
289 	unsigned long		virt_boundary_mask;
290 
291 	unsigned int		max_hw_sectors;
292 	unsigned int		max_dev_sectors;
293 	unsigned int		chunk_sectors;
294 	unsigned int		max_sectors;
295 	unsigned int		max_user_sectors;
296 	unsigned int		max_segment_size;
297 	unsigned int		physical_block_size;
298 	unsigned int		logical_block_size;
299 	unsigned int		alignment_offset;
300 	unsigned int		io_min;
301 	unsigned int		io_opt;
302 	unsigned int		max_discard_sectors;
303 	unsigned int		max_hw_discard_sectors;
304 	unsigned int		max_user_discard_sectors;
305 	unsigned int		max_secure_erase_sectors;
306 	unsigned int		max_write_zeroes_sectors;
307 	unsigned int		max_zone_append_sectors;
308 	unsigned int		discard_granularity;
309 	unsigned int		discard_alignment;
310 	unsigned int		zone_write_granularity;
311 
312 	unsigned short		max_segments;
313 	unsigned short		max_integrity_segments;
314 	unsigned short		max_discard_segments;
315 
316 	unsigned char		misaligned;
317 	unsigned char		discard_misaligned;
318 	unsigned char		raid_partial_stripes_expensive;
319 	bool			zoned;
320 	unsigned int		max_open_zones;
321 	unsigned int		max_active_zones;
322 
323 	/*
324 	 * Drivers that set dma_alignment to less than 511 must be prepared to
325 	 * handle individual bvec's that are not a multiple of a SECTOR_SIZE
326 	 * due to possible offsets.
327 	 */
328 	unsigned int		dma_alignment;
329 };
330 
331 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
332 			       void *data);
333 
334 void disk_set_zoned(struct gendisk *disk);
335 
336 #define BLK_ALL_ZONES  ((unsigned int)-1)
337 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
338 		unsigned int nr_zones, report_zones_cb cb, void *data);
339 int blkdev_zone_mgmt(struct block_device *bdev, enum req_op op,
340 		sector_t sectors, sector_t nr_sectors);
341 int blk_revalidate_disk_zones(struct gendisk *disk);
342 
343 /*
344  * Independent access ranges: struct blk_independent_access_range describes
345  * a range of contiguous sectors that can be accessed using device command
346  * execution resources that are independent from the resources used for
347  * other access ranges. This is typically found with single-LUN multi-actuator
348  * HDDs where each access range is served by a different set of heads.
349  * The set of independent ranges supported by the device is defined using
350  * struct blk_independent_access_ranges. The independent ranges must not overlap
351  * and must include all sectors within the disk capacity (no sector holes
352  * allowed).
353  * For a device with multiple ranges, requests targeting sectors in different
354  * ranges can be executed in parallel. A request can straddle an access range
355  * boundary.
356  */
357 struct blk_independent_access_range {
358 	struct kobject		kobj;
359 	sector_t		sector;
360 	sector_t		nr_sectors;
361 };
362 
363 struct blk_independent_access_ranges {
364 	struct kobject				kobj;
365 	bool					sysfs_registered;
366 	unsigned int				nr_ia_ranges;
367 	struct blk_independent_access_range	ia_range[];
368 };
369 
370 struct request_queue {
371 	/*
372 	 * The queue owner gets to use this for whatever they like.
373 	 * ll_rw_blk doesn't touch it.
374 	 */
375 	void			*queuedata;
376 
377 	struct elevator_queue	*elevator;
378 
379 	const struct blk_mq_ops	*mq_ops;
380 
381 	/* sw queues */
382 	struct blk_mq_ctx __percpu	*queue_ctx;
383 
384 	/*
385 	 * various queue flags, see QUEUE_* below
386 	 */
387 	unsigned long		queue_flags;
388 
389 	unsigned int		rq_timeout;
390 
391 	unsigned int		queue_depth;
392 
393 	refcount_t		refs;
394 
395 	/* hw dispatch queues */
396 	unsigned int		nr_hw_queues;
397 	struct xarray		hctx_table;
398 
399 	struct percpu_ref	q_usage_counter;
400 
401 	struct request		*last_merge;
402 
403 	spinlock_t		queue_lock;
404 
405 	int			quiesce_depth;
406 
407 	struct gendisk		*disk;
408 
409 	/*
410 	 * mq queue kobject
411 	 */
412 	struct kobject *mq_kobj;
413 
414 	struct queue_limits	limits;
415 
416 #ifdef  CONFIG_BLK_DEV_INTEGRITY
417 	struct blk_integrity integrity;
418 #endif	/* CONFIG_BLK_DEV_INTEGRITY */
419 
420 #ifdef CONFIG_PM
421 	struct device		*dev;
422 	enum rpm_status		rpm_status;
423 #endif
424 
425 	/*
426 	 * Number of contexts that have called blk_set_pm_only(). If this
427 	 * counter is above zero then only RQF_PM requests are processed.
428 	 */
429 	atomic_t		pm_only;
430 
431 	struct blk_queue_stats	*stats;
432 	struct rq_qos		*rq_qos;
433 	struct mutex		rq_qos_mutex;
434 
435 	/*
436 	 * ida allocated id for this queue.  Used to index queues from
437 	 * ioctx.
438 	 */
439 	int			id;
440 
441 	unsigned int		dma_pad_mask;
442 
443 	/*
444 	 * queue settings
445 	 */
446 	unsigned long		nr_requests;	/* Max # of requests */
447 
448 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
449 	struct blk_crypto_profile *crypto_profile;
450 	struct kobject *crypto_kobject;
451 #endif
452 
453 	struct timer_list	timeout;
454 	struct work_struct	timeout_work;
455 
456 	atomic_t		nr_active_requests_shared_tags;
457 
458 	struct blk_mq_tags	*sched_shared_tags;
459 
460 	struct list_head	icq_list;
461 #ifdef CONFIG_BLK_CGROUP
462 	DECLARE_BITMAP		(blkcg_pols, BLKCG_MAX_POLS);
463 	struct blkcg_gq		*root_blkg;
464 	struct list_head	blkg_list;
465 	struct mutex		blkcg_mutex;
466 #endif
467 
468 	int			node;
469 
470 	spinlock_t		requeue_lock;
471 	struct list_head	requeue_list;
472 	struct delayed_work	requeue_work;
473 
474 #ifdef CONFIG_BLK_DEV_IO_TRACE
475 	struct blk_trace __rcu	*blk_trace;
476 #endif
477 	/*
478 	 * for flush operations
479 	 */
480 	struct blk_flush_queue	*fq;
481 	struct list_head	flush_list;
482 
483 	struct mutex		sysfs_lock;
484 	struct mutex		sysfs_dir_lock;
485 	struct mutex		limits_lock;
486 
487 	/*
488 	 * for reusing dead hctx instance in case of updating
489 	 * nr_hw_queues
490 	 */
491 	struct list_head	unused_hctx_list;
492 	spinlock_t		unused_hctx_lock;
493 
494 	int			mq_freeze_depth;
495 
496 #ifdef CONFIG_BLK_DEV_THROTTLING
497 	/* Throttle data */
498 	struct throtl_data *td;
499 #endif
500 	struct rcu_head		rcu_head;
501 	wait_queue_head_t	mq_freeze_wq;
502 	/*
503 	 * Protect concurrent access to q_usage_counter by
504 	 * percpu_ref_kill() and percpu_ref_reinit().
505 	 */
506 	struct mutex		mq_freeze_lock;
507 
508 	struct blk_mq_tag_set	*tag_set;
509 	struct list_head	tag_set_list;
510 
511 	struct dentry		*debugfs_dir;
512 	struct dentry		*sched_debugfs_dir;
513 	struct dentry		*rqos_debugfs_dir;
514 	/*
515 	 * Serializes all debugfs metadata operations using the above dentries.
516 	 */
517 	struct mutex		debugfs_mutex;
518 
519 	bool			mq_sysfs_init_done;
520 };
521 
522 /* Keep blk_queue_flag_name[] in sync with the definitions below */
523 #define QUEUE_FLAG_STOPPED	0	/* queue is stopped */
524 #define QUEUE_FLAG_DYING	1	/* queue being torn down */
525 #define QUEUE_FLAG_NOMERGES     3	/* disable merge attempts */
526 #define QUEUE_FLAG_SAME_COMP	4	/* complete on same CPU-group */
527 #define QUEUE_FLAG_FAIL_IO	5	/* fake timeout */
528 #define QUEUE_FLAG_NONROT	6	/* non-rotational device (SSD) */
529 #define QUEUE_FLAG_VIRT		QUEUE_FLAG_NONROT /* paravirt device */
530 #define QUEUE_FLAG_IO_STAT	7	/* do disk/partitions IO accounting */
531 #define QUEUE_FLAG_NOXMERGES	9	/* No extended merges */
532 #define QUEUE_FLAG_ADD_RANDOM	10	/* Contributes to random pool */
533 #define QUEUE_FLAG_SYNCHRONOUS	11	/* always completes in submit context */
534 #define QUEUE_FLAG_SAME_FORCE	12	/* force complete on same CPU */
535 #define QUEUE_FLAG_HW_WC	13	/* Write back caching supported */
536 #define QUEUE_FLAG_INIT_DONE	14	/* queue is initialized */
537 #define QUEUE_FLAG_STABLE_WRITES 15	/* don't modify blks until WB is done */
538 #define QUEUE_FLAG_POLL		16	/* IO polling enabled if set */
539 #define QUEUE_FLAG_WC		17	/* Write back caching */
540 #define QUEUE_FLAG_FUA		18	/* device supports FUA writes */
541 #define QUEUE_FLAG_DAX		19	/* device supports DAX */
542 #define QUEUE_FLAG_STATS	20	/* track IO start and completion times */
543 #define QUEUE_FLAG_REGISTERED	22	/* queue has been registered to a disk */
544 #define QUEUE_FLAG_QUIESCED	24	/* queue has been quiesced */
545 #define QUEUE_FLAG_PCI_P2PDMA	25	/* device supports PCI p2p requests */
546 #define QUEUE_FLAG_ZONE_RESETALL 26	/* supports Zone Reset All */
547 #define QUEUE_FLAG_RQ_ALLOC_TIME 27	/* record rq->alloc_time_ns */
548 #define QUEUE_FLAG_HCTX_ACTIVE	28	/* at least one blk-mq hctx is active */
549 #define QUEUE_FLAG_NOWAIT       29	/* device supports NOWAIT */
550 #define QUEUE_FLAG_SQ_SCHED     30	/* single queue style io dispatch */
551 #define QUEUE_FLAG_SKIP_TAGSET_QUIESCE	31 /* quiesce_tagset skip the queue*/
552 
553 #define QUEUE_FLAG_MQ_DEFAULT	((1UL << QUEUE_FLAG_IO_STAT) |		\
554 				 (1UL << QUEUE_FLAG_SAME_COMP) |	\
555 				 (1UL << QUEUE_FLAG_NOWAIT))
556 
557 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
558 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
559 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
560 
561 #define blk_queue_stopped(q)	test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
562 #define blk_queue_dying(q)	test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
563 #define blk_queue_init_done(q)	test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
564 #define blk_queue_nomerges(q)	test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
565 #define blk_queue_noxmerges(q)	\
566 	test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
567 #define blk_queue_nonrot(q)	test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
568 #define blk_queue_stable_writes(q) \
569 	test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
570 #define blk_queue_io_stat(q)	test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
571 #define blk_queue_add_random(q)	test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
572 #define blk_queue_zone_resetall(q)	\
573 	test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
574 #define blk_queue_dax(q)	test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
575 #define blk_queue_pci_p2pdma(q)	\
576 	test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
577 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
578 #define blk_queue_rq_alloc_time(q)	\
579 	test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
580 #else
581 #define blk_queue_rq_alloc_time(q)	false
582 #endif
583 
584 #define blk_noretry_request(rq) \
585 	((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
586 			     REQ_FAILFAST_DRIVER))
587 #define blk_queue_quiesced(q)	test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
588 #define blk_queue_pm_only(q)	atomic_read(&(q)->pm_only)
589 #define blk_queue_registered(q)	test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
590 #define blk_queue_sq_sched(q)	test_bit(QUEUE_FLAG_SQ_SCHED, &(q)->queue_flags)
591 #define blk_queue_skip_tagset_quiesce(q) \
592 	test_bit(QUEUE_FLAG_SKIP_TAGSET_QUIESCE, &(q)->queue_flags)
593 
594 extern void blk_set_pm_only(struct request_queue *q);
595 extern void blk_clear_pm_only(struct request_queue *q);
596 
597 #define list_entry_rq(ptr)	list_entry((ptr), struct request, queuelist)
598 
599 #define dma_map_bvec(dev, bv, dir, attrs) \
600 	dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
601 	(dir), (attrs))
602 
603 static inline bool queue_is_mq(struct request_queue *q)
604 {
605 	return q->mq_ops;
606 }
607 
608 #ifdef CONFIG_PM
609 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
610 {
611 	return q->rpm_status;
612 }
613 #else
614 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
615 {
616 	return RPM_ACTIVE;
617 }
618 #endif
619 
620 static inline bool blk_queue_is_zoned(struct request_queue *q)
621 {
622 	return IS_ENABLED(CONFIG_BLK_DEV_ZONED) && q->limits.zoned;
623 }
624 
625 #ifdef CONFIG_BLK_DEV_ZONED
626 unsigned int bdev_nr_zones(struct block_device *bdev);
627 
628 static inline unsigned int disk_nr_zones(struct gendisk *disk)
629 {
630 	return blk_queue_is_zoned(disk->queue) ? disk->nr_zones : 0;
631 }
632 
633 static inline unsigned int disk_zone_no(struct gendisk *disk, sector_t sector)
634 {
635 	if (!blk_queue_is_zoned(disk->queue))
636 		return 0;
637 	return sector >> ilog2(disk->queue->limits.chunk_sectors);
638 }
639 
640 static inline void disk_set_max_open_zones(struct gendisk *disk,
641 		unsigned int max_open_zones)
642 {
643 	disk->queue->limits.max_open_zones = max_open_zones;
644 }
645 
646 static inline void disk_set_max_active_zones(struct gendisk *disk,
647 		unsigned int max_active_zones)
648 {
649 	disk->queue->limits.max_active_zones = max_active_zones;
650 }
651 
652 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
653 {
654 	return bdev->bd_disk->queue->limits.max_open_zones;
655 }
656 
657 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
658 {
659 	return bdev->bd_disk->queue->limits.max_active_zones;
660 }
661 
662 bool blk_zone_plug_bio(struct bio *bio, unsigned int nr_segs);
663 #else /* CONFIG_BLK_DEV_ZONED */
664 static inline unsigned int bdev_nr_zones(struct block_device *bdev)
665 {
666 	return 0;
667 }
668 
669 static inline unsigned int disk_nr_zones(struct gendisk *disk)
670 {
671 	return 0;
672 }
673 static inline unsigned int disk_zone_no(struct gendisk *disk, sector_t sector)
674 {
675 	return 0;
676 }
677 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
678 {
679 	return 0;
680 }
681 
682 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
683 {
684 	return 0;
685 }
686 static inline bool blk_zone_plug_bio(struct bio *bio, unsigned int nr_segs)
687 {
688 	return false;
689 }
690 #endif /* CONFIG_BLK_DEV_ZONED */
691 
692 static inline unsigned int blk_queue_depth(struct request_queue *q)
693 {
694 	if (q->queue_depth)
695 		return q->queue_depth;
696 
697 	return q->nr_requests;
698 }
699 
700 /*
701  * default timeout for SG_IO if none specified
702  */
703 #define BLK_DEFAULT_SG_TIMEOUT	(60 * HZ)
704 #define BLK_MIN_SG_TIMEOUT	(7 * HZ)
705 
706 /* This should not be used directly - use rq_for_each_segment */
707 #define for_each_bio(_bio)		\
708 	for (; _bio; _bio = _bio->bi_next)
709 
710 int __must_check device_add_disk(struct device *parent, struct gendisk *disk,
711 				 const struct attribute_group **groups);
712 static inline int __must_check add_disk(struct gendisk *disk)
713 {
714 	return device_add_disk(NULL, disk, NULL);
715 }
716 void del_gendisk(struct gendisk *gp);
717 void invalidate_disk(struct gendisk *disk);
718 void set_disk_ro(struct gendisk *disk, bool read_only);
719 void disk_uevent(struct gendisk *disk, enum kobject_action action);
720 
721 static inline int get_disk_ro(struct gendisk *disk)
722 {
723 	return disk->part0->bd_read_only ||
724 		test_bit(GD_READ_ONLY, &disk->state);
725 }
726 
727 static inline int bdev_read_only(struct block_device *bdev)
728 {
729 	return bdev->bd_read_only || get_disk_ro(bdev->bd_disk);
730 }
731 
732 bool set_capacity_and_notify(struct gendisk *disk, sector_t size);
733 void disk_force_media_change(struct gendisk *disk);
734 void bdev_mark_dead(struct block_device *bdev, bool surprise);
735 
736 void add_disk_randomness(struct gendisk *disk) __latent_entropy;
737 void rand_initialize_disk(struct gendisk *disk);
738 
739 static inline sector_t get_start_sect(struct block_device *bdev)
740 {
741 	return bdev->bd_start_sect;
742 }
743 
744 static inline sector_t bdev_nr_sectors(struct block_device *bdev)
745 {
746 	return bdev->bd_nr_sectors;
747 }
748 
749 static inline loff_t bdev_nr_bytes(struct block_device *bdev)
750 {
751 	return (loff_t)bdev_nr_sectors(bdev) << SECTOR_SHIFT;
752 }
753 
754 static inline sector_t get_capacity(struct gendisk *disk)
755 {
756 	return bdev_nr_sectors(disk->part0);
757 }
758 
759 static inline u64 sb_bdev_nr_blocks(struct super_block *sb)
760 {
761 	return bdev_nr_sectors(sb->s_bdev) >>
762 		(sb->s_blocksize_bits - SECTOR_SHIFT);
763 }
764 
765 int bdev_disk_changed(struct gendisk *disk, bool invalidate);
766 
767 void put_disk(struct gendisk *disk);
768 struct gendisk *__blk_alloc_disk(struct queue_limits *lim, int node,
769 		struct lock_class_key *lkclass);
770 
771 /**
772  * blk_alloc_disk - allocate a gendisk structure
773  * @lim: queue limits to be used for this disk.
774  * @node_id: numa node to allocate on
775  *
776  * Allocate and pre-initialize a gendisk structure for use with BIO based
777  * drivers.
778  *
779  * Returns an ERR_PTR on error, else the allocated disk.
780  *
781  * Context: can sleep
782  */
783 #define blk_alloc_disk(lim, node_id)					\
784 ({									\
785 	static struct lock_class_key __key;				\
786 									\
787 	__blk_alloc_disk(lim, node_id, &__key);				\
788 })
789 
790 int __register_blkdev(unsigned int major, const char *name,
791 		void (*probe)(dev_t devt));
792 #define register_blkdev(major, name) \
793 	__register_blkdev(major, name, NULL)
794 void unregister_blkdev(unsigned int major, const char *name);
795 
796 bool disk_check_media_change(struct gendisk *disk);
797 void set_capacity(struct gendisk *disk, sector_t size);
798 
799 #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED
800 int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
801 void bd_unlink_disk_holder(struct block_device *bdev, struct gendisk *disk);
802 #else
803 static inline int bd_link_disk_holder(struct block_device *bdev,
804 				      struct gendisk *disk)
805 {
806 	return 0;
807 }
808 static inline void bd_unlink_disk_holder(struct block_device *bdev,
809 					 struct gendisk *disk)
810 {
811 }
812 #endif /* CONFIG_BLOCK_HOLDER_DEPRECATED */
813 
814 dev_t part_devt(struct gendisk *disk, u8 partno);
815 void inc_diskseq(struct gendisk *disk);
816 void blk_request_module(dev_t devt);
817 
818 extern int blk_register_queue(struct gendisk *disk);
819 extern void blk_unregister_queue(struct gendisk *disk);
820 void submit_bio_noacct(struct bio *bio);
821 struct bio *bio_split_to_limits(struct bio *bio);
822 
823 extern int blk_lld_busy(struct request_queue *q);
824 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
825 extern void blk_queue_exit(struct request_queue *q);
826 extern void blk_sync_queue(struct request_queue *q);
827 
828 /* Helper to convert REQ_OP_XXX to its string format XXX */
829 extern const char *blk_op_str(enum req_op op);
830 
831 int blk_status_to_errno(blk_status_t status);
832 blk_status_t errno_to_blk_status(int errno);
833 const char *blk_status_to_str(blk_status_t status);
834 
835 /* only poll the hardware once, don't continue until a completion was found */
836 #define BLK_POLL_ONESHOT		(1 << 0)
837 int bio_poll(struct bio *bio, struct io_comp_batch *iob, unsigned int flags);
838 int iocb_bio_iopoll(struct kiocb *kiocb, struct io_comp_batch *iob,
839 			unsigned int flags);
840 
841 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
842 {
843 	return bdev->bd_queue;	/* this is never NULL */
844 }
845 
846 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
847 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
848 
849 static inline unsigned int bio_zone_no(struct bio *bio)
850 {
851 	return disk_zone_no(bio->bi_bdev->bd_disk, bio->bi_iter.bi_sector);
852 }
853 
854 static inline bool bio_straddles_zones(struct bio *bio)
855 {
856 	return bio_sectors(bio) &&
857 		bio_zone_no(bio) !=
858 		disk_zone_no(bio->bi_bdev->bd_disk, bio_end_sector(bio) - 1);
859 }
860 
861 /*
862  * Return how much of the chunk is left to be used for I/O at a given offset.
863  */
864 static inline unsigned int blk_chunk_sectors_left(sector_t offset,
865 		unsigned int chunk_sectors)
866 {
867 	if (unlikely(!is_power_of_2(chunk_sectors)))
868 		return chunk_sectors - sector_div(offset, chunk_sectors);
869 	return chunk_sectors - (offset & (chunk_sectors - 1));
870 }
871 
872 /**
873  * queue_limits_start_update - start an atomic update of queue limits
874  * @q:		queue to update
875  *
876  * This functions starts an atomic update of the queue limits.  It takes a lock
877  * to prevent other updates and returns a snapshot of the current limits that
878  * the caller can modify.  The caller must call queue_limits_commit_update()
879  * to finish the update.
880  *
881  * Context: process context.  The caller must have frozen the queue or ensured
882  * that there is outstanding I/O by other means.
883  */
884 static inline struct queue_limits
885 queue_limits_start_update(struct request_queue *q)
886 	__acquires(q->limits_lock)
887 {
888 	mutex_lock(&q->limits_lock);
889 	return q->limits;
890 }
891 int queue_limits_commit_update(struct request_queue *q,
892 		struct queue_limits *lim);
893 int queue_limits_set(struct request_queue *q, struct queue_limits *lim);
894 
895 /**
896  * queue_limits_cancel_update - cancel an atomic update of queue limits
897  * @q:		queue to update
898  *
899  * This functions cancels an atomic update of the queue limits started by
900  * queue_limits_start_update() and should be used when an error occurs after
901  * starting update.
902  */
903 static inline void queue_limits_cancel_update(struct request_queue *q)
904 {
905 	mutex_unlock(&q->limits_lock);
906 }
907 
908 /*
909  * Access functions for manipulating queue properties
910  */
911 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
912 void blk_queue_max_secure_erase_sectors(struct request_queue *q,
913 		unsigned int max_sectors);
914 extern void blk_queue_max_discard_sectors(struct request_queue *q,
915 		unsigned int max_discard_sectors);
916 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
917 		unsigned int max_write_same_sectors);
918 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
919 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
920 		unsigned int max_zone_append_sectors);
921 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
922 void blk_queue_zone_write_granularity(struct request_queue *q,
923 				      unsigned int size);
924 extern void blk_queue_alignment_offset(struct request_queue *q,
925 				       unsigned int alignment);
926 void disk_update_readahead(struct gendisk *disk);
927 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
928 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
929 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
930 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
931 extern void blk_set_stacking_limits(struct queue_limits *lim);
932 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
933 			    sector_t offset);
934 void queue_limits_stack_bdev(struct queue_limits *t, struct block_device *bdev,
935 		sector_t offset, const char *pfx);
936 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
937 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
938 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
939 
940 struct blk_independent_access_ranges *
941 disk_alloc_independent_access_ranges(struct gendisk *disk, int nr_ia_ranges);
942 void disk_set_independent_access_ranges(struct gendisk *disk,
943 				struct blk_independent_access_ranges *iars);
944 
945 bool __must_check blk_get_queue(struct request_queue *);
946 extern void blk_put_queue(struct request_queue *);
947 
948 void blk_mark_disk_dead(struct gendisk *disk);
949 
950 #ifdef CONFIG_BLOCK
951 /*
952  * blk_plug permits building a queue of related requests by holding the I/O
953  * fragments for a short period. This allows merging of sequential requests
954  * into single larger request. As the requests are moved from a per-task list to
955  * the device's request_queue in a batch, this results in improved scalability
956  * as the lock contention for request_queue lock is reduced.
957  *
958  * It is ok not to disable preemption when adding the request to the plug list
959  * or when attempting a merge. For details, please see schedule() where
960  * blk_flush_plug() is called.
961  */
962 struct blk_plug {
963 	struct request *mq_list; /* blk-mq requests */
964 
965 	/* if ios_left is > 1, we can batch tag/rq allocations */
966 	struct request *cached_rq;
967 	u64 cur_ktime;
968 	unsigned short nr_ios;
969 
970 	unsigned short rq_count;
971 
972 	bool multiple_queues;
973 	bool has_elevator;
974 
975 	struct list_head cb_list; /* md requires an unplug callback */
976 };
977 
978 struct blk_plug_cb;
979 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
980 struct blk_plug_cb {
981 	struct list_head list;
982 	blk_plug_cb_fn callback;
983 	void *data;
984 };
985 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
986 					     void *data, int size);
987 extern void blk_start_plug(struct blk_plug *);
988 extern void blk_start_plug_nr_ios(struct blk_plug *, unsigned short);
989 extern void blk_finish_plug(struct blk_plug *);
990 
991 void __blk_flush_plug(struct blk_plug *plug, bool from_schedule);
992 static inline void blk_flush_plug(struct blk_plug *plug, bool async)
993 {
994 	if (plug)
995 		__blk_flush_plug(plug, async);
996 }
997 
998 /*
999  * tsk == current here
1000  */
1001 static inline void blk_plug_invalidate_ts(struct task_struct *tsk)
1002 {
1003 	struct blk_plug *plug = tsk->plug;
1004 
1005 	if (plug)
1006 		plug->cur_ktime = 0;
1007 	current->flags &= ~PF_BLOCK_TS;
1008 }
1009 
1010 int blkdev_issue_flush(struct block_device *bdev);
1011 long nr_blockdev_pages(void);
1012 #else /* CONFIG_BLOCK */
1013 struct blk_plug {
1014 };
1015 
1016 static inline void blk_start_plug_nr_ios(struct blk_plug *plug,
1017 					 unsigned short nr_ios)
1018 {
1019 }
1020 
1021 static inline void blk_start_plug(struct blk_plug *plug)
1022 {
1023 }
1024 
1025 static inline void blk_finish_plug(struct blk_plug *plug)
1026 {
1027 }
1028 
1029 static inline void blk_flush_plug(struct blk_plug *plug, bool async)
1030 {
1031 }
1032 
1033 static inline void blk_plug_invalidate_ts(struct task_struct *tsk)
1034 {
1035 }
1036 
1037 static inline int blkdev_issue_flush(struct block_device *bdev)
1038 {
1039 	return 0;
1040 }
1041 
1042 static inline long nr_blockdev_pages(void)
1043 {
1044 	return 0;
1045 }
1046 #endif /* CONFIG_BLOCK */
1047 
1048 extern void blk_io_schedule(void);
1049 
1050 int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1051 		sector_t nr_sects, gfp_t gfp_mask);
1052 int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1053 		sector_t nr_sects, gfp_t gfp_mask, struct bio **biop);
1054 int blkdev_issue_secure_erase(struct block_device *bdev, sector_t sector,
1055 		sector_t nr_sects, gfp_t gfp);
1056 
1057 #define BLKDEV_ZERO_NOUNMAP	(1 << 0)  /* do not free blocks */
1058 #define BLKDEV_ZERO_NOFALLBACK	(1 << 1)  /* don't write explicit zeroes */
1059 
1060 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1061 		sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1062 		unsigned flags);
1063 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1064 		sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1065 
1066 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1067 		sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1068 {
1069 	return blkdev_issue_discard(sb->s_bdev,
1070 				    block << (sb->s_blocksize_bits -
1071 					      SECTOR_SHIFT),
1072 				    nr_blocks << (sb->s_blocksize_bits -
1073 						  SECTOR_SHIFT),
1074 				    gfp_mask);
1075 }
1076 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1077 		sector_t nr_blocks, gfp_t gfp_mask)
1078 {
1079 	return blkdev_issue_zeroout(sb->s_bdev,
1080 				    block << (sb->s_blocksize_bits -
1081 					      SECTOR_SHIFT),
1082 				    nr_blocks << (sb->s_blocksize_bits -
1083 						  SECTOR_SHIFT),
1084 				    gfp_mask, 0);
1085 }
1086 
1087 static inline bool bdev_is_partition(struct block_device *bdev)
1088 {
1089 	return bdev->bd_partno;
1090 }
1091 
1092 enum blk_default_limits {
1093 	BLK_MAX_SEGMENTS	= 128,
1094 	BLK_SAFE_MAX_SECTORS	= 255,
1095 	BLK_MAX_SEGMENT_SIZE	= 65536,
1096 	BLK_SEG_BOUNDARY_MASK	= 0xFFFFFFFFUL,
1097 };
1098 
1099 /*
1100  * Default upper limit for the software max_sectors limit used for
1101  * regular file system I/O.  This can be increased through sysfs.
1102  *
1103  * Not to be confused with the max_hw_sector limit that is entirely
1104  * controlled by the driver, usually based on hardware limits.
1105  */
1106 #define BLK_DEF_MAX_SECTORS_CAP	2560u
1107 
1108 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1109 {
1110 	return q->limits.seg_boundary_mask;
1111 }
1112 
1113 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1114 {
1115 	return q->limits.virt_boundary_mask;
1116 }
1117 
1118 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1119 {
1120 	return q->limits.max_sectors;
1121 }
1122 
1123 static inline unsigned int queue_max_bytes(struct request_queue *q)
1124 {
1125 	return min_t(unsigned int, queue_max_sectors(q), INT_MAX >> 9) << 9;
1126 }
1127 
1128 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1129 {
1130 	return q->limits.max_hw_sectors;
1131 }
1132 
1133 static inline unsigned short queue_max_segments(const struct request_queue *q)
1134 {
1135 	return q->limits.max_segments;
1136 }
1137 
1138 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1139 {
1140 	return q->limits.max_discard_segments;
1141 }
1142 
1143 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1144 {
1145 	return q->limits.max_segment_size;
1146 }
1147 
1148 static inline unsigned int queue_limits_max_zone_append_sectors(struct queue_limits *l)
1149 {
1150 	unsigned int max_sectors = min(l->chunk_sectors, l->max_hw_sectors);
1151 
1152 	return min_not_zero(l->max_zone_append_sectors, max_sectors);
1153 }
1154 
1155 static inline unsigned int queue_max_zone_append_sectors(struct request_queue *q)
1156 {
1157 	if (!blk_queue_is_zoned(q))
1158 		return 0;
1159 
1160 	return queue_limits_max_zone_append_sectors(&q->limits);
1161 }
1162 
1163 static inline bool queue_emulates_zone_append(struct request_queue *q)
1164 {
1165 	return blk_queue_is_zoned(q) && !q->limits.max_zone_append_sectors;
1166 }
1167 
1168 static inline bool bdev_emulates_zone_append(struct block_device *bdev)
1169 {
1170 	return queue_emulates_zone_append(bdev_get_queue(bdev));
1171 }
1172 
1173 static inline unsigned int
1174 bdev_max_zone_append_sectors(struct block_device *bdev)
1175 {
1176 	return queue_max_zone_append_sectors(bdev_get_queue(bdev));
1177 }
1178 
1179 static inline unsigned int bdev_max_segments(struct block_device *bdev)
1180 {
1181 	return queue_max_segments(bdev_get_queue(bdev));
1182 }
1183 
1184 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1185 {
1186 	int retval = 512;
1187 
1188 	if (q && q->limits.logical_block_size)
1189 		retval = q->limits.logical_block_size;
1190 
1191 	return retval;
1192 }
1193 
1194 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1195 {
1196 	return queue_logical_block_size(bdev_get_queue(bdev));
1197 }
1198 
1199 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1200 {
1201 	return q->limits.physical_block_size;
1202 }
1203 
1204 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1205 {
1206 	return queue_physical_block_size(bdev_get_queue(bdev));
1207 }
1208 
1209 static inline unsigned int queue_io_min(const struct request_queue *q)
1210 {
1211 	return q->limits.io_min;
1212 }
1213 
1214 static inline int bdev_io_min(struct block_device *bdev)
1215 {
1216 	return queue_io_min(bdev_get_queue(bdev));
1217 }
1218 
1219 static inline unsigned int queue_io_opt(const struct request_queue *q)
1220 {
1221 	return q->limits.io_opt;
1222 }
1223 
1224 static inline int bdev_io_opt(struct block_device *bdev)
1225 {
1226 	return queue_io_opt(bdev_get_queue(bdev));
1227 }
1228 
1229 static inline unsigned int
1230 queue_zone_write_granularity(const struct request_queue *q)
1231 {
1232 	return q->limits.zone_write_granularity;
1233 }
1234 
1235 static inline unsigned int
1236 bdev_zone_write_granularity(struct block_device *bdev)
1237 {
1238 	return queue_zone_write_granularity(bdev_get_queue(bdev));
1239 }
1240 
1241 int bdev_alignment_offset(struct block_device *bdev);
1242 unsigned int bdev_discard_alignment(struct block_device *bdev);
1243 
1244 static inline unsigned int bdev_max_discard_sectors(struct block_device *bdev)
1245 {
1246 	return bdev_get_queue(bdev)->limits.max_discard_sectors;
1247 }
1248 
1249 static inline unsigned int bdev_discard_granularity(struct block_device *bdev)
1250 {
1251 	return bdev_get_queue(bdev)->limits.discard_granularity;
1252 }
1253 
1254 static inline unsigned int
1255 bdev_max_secure_erase_sectors(struct block_device *bdev)
1256 {
1257 	return bdev_get_queue(bdev)->limits.max_secure_erase_sectors;
1258 }
1259 
1260 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1261 {
1262 	struct request_queue *q = bdev_get_queue(bdev);
1263 
1264 	if (q)
1265 		return q->limits.max_write_zeroes_sectors;
1266 
1267 	return 0;
1268 }
1269 
1270 static inline bool bdev_nonrot(struct block_device *bdev)
1271 {
1272 	return blk_queue_nonrot(bdev_get_queue(bdev));
1273 }
1274 
1275 static inline bool bdev_synchronous(struct block_device *bdev)
1276 {
1277 	return test_bit(QUEUE_FLAG_SYNCHRONOUS,
1278 			&bdev_get_queue(bdev)->queue_flags);
1279 }
1280 
1281 static inline bool bdev_stable_writes(struct block_device *bdev)
1282 {
1283 	return test_bit(QUEUE_FLAG_STABLE_WRITES,
1284 			&bdev_get_queue(bdev)->queue_flags);
1285 }
1286 
1287 static inline bool bdev_write_cache(struct block_device *bdev)
1288 {
1289 	return test_bit(QUEUE_FLAG_WC, &bdev_get_queue(bdev)->queue_flags);
1290 }
1291 
1292 static inline bool bdev_fua(struct block_device *bdev)
1293 {
1294 	return test_bit(QUEUE_FLAG_FUA, &bdev_get_queue(bdev)->queue_flags);
1295 }
1296 
1297 static inline bool bdev_nowait(struct block_device *bdev)
1298 {
1299 	return test_bit(QUEUE_FLAG_NOWAIT, &bdev_get_queue(bdev)->queue_flags);
1300 }
1301 
1302 static inline bool bdev_is_zoned(struct block_device *bdev)
1303 {
1304 	return blk_queue_is_zoned(bdev_get_queue(bdev));
1305 }
1306 
1307 static inline unsigned int bdev_zone_no(struct block_device *bdev, sector_t sec)
1308 {
1309 	return disk_zone_no(bdev->bd_disk, sec);
1310 }
1311 
1312 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1313 {
1314 	struct request_queue *q = bdev_get_queue(bdev);
1315 
1316 	if (!blk_queue_is_zoned(q))
1317 		return 0;
1318 	return q->limits.chunk_sectors;
1319 }
1320 
1321 static inline sector_t bdev_offset_from_zone_start(struct block_device *bdev,
1322 						   sector_t sector)
1323 {
1324 	return sector & (bdev_zone_sectors(bdev) - 1);
1325 }
1326 
1327 static inline sector_t bio_offset_from_zone_start(struct bio *bio)
1328 {
1329 	return bdev_offset_from_zone_start(bio->bi_bdev,
1330 					   bio->bi_iter.bi_sector);
1331 }
1332 
1333 static inline bool bdev_is_zone_start(struct block_device *bdev,
1334 				      sector_t sector)
1335 {
1336 	return bdev_offset_from_zone_start(bdev, sector) == 0;
1337 }
1338 
1339 static inline int queue_dma_alignment(const struct request_queue *q)
1340 {
1341 	return q ? q->limits.dma_alignment : 511;
1342 }
1343 
1344 static inline unsigned int bdev_dma_alignment(struct block_device *bdev)
1345 {
1346 	return queue_dma_alignment(bdev_get_queue(bdev));
1347 }
1348 
1349 static inline bool bdev_iter_is_aligned(struct block_device *bdev,
1350 					struct iov_iter *iter)
1351 {
1352 	return iov_iter_is_aligned(iter, bdev_dma_alignment(bdev),
1353 				   bdev_logical_block_size(bdev) - 1);
1354 }
1355 
1356 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1357 				 unsigned int len)
1358 {
1359 	unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1360 	return !(addr & alignment) && !(len & alignment);
1361 }
1362 
1363 /* assumes size > 256 */
1364 static inline unsigned int blksize_bits(unsigned int size)
1365 {
1366 	return order_base_2(size >> SECTOR_SHIFT) + SECTOR_SHIFT;
1367 }
1368 
1369 int kblockd_schedule_work(struct work_struct *work);
1370 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1371 
1372 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1373 	MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1374 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1375 	MODULE_ALIAS("block-major-" __stringify(major) "-*")
1376 
1377 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1378 
1379 bool blk_crypto_register(struct blk_crypto_profile *profile,
1380 			 struct request_queue *q);
1381 
1382 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1383 
1384 static inline bool blk_crypto_register(struct blk_crypto_profile *profile,
1385 				       struct request_queue *q)
1386 {
1387 	return true;
1388 }
1389 
1390 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1391 
1392 enum blk_unique_id {
1393 	/* these match the Designator Types specified in SPC */
1394 	BLK_UID_T10	= 1,
1395 	BLK_UID_EUI64	= 2,
1396 	BLK_UID_NAA	= 3,
1397 };
1398 
1399 struct block_device_operations {
1400 	void (*submit_bio)(struct bio *bio);
1401 	int (*poll_bio)(struct bio *bio, struct io_comp_batch *iob,
1402 			unsigned int flags);
1403 	int (*open)(struct gendisk *disk, blk_mode_t mode);
1404 	void (*release)(struct gendisk *disk);
1405 	int (*ioctl)(struct block_device *bdev, blk_mode_t mode,
1406 			unsigned cmd, unsigned long arg);
1407 	int (*compat_ioctl)(struct block_device *bdev, blk_mode_t mode,
1408 			unsigned cmd, unsigned long arg);
1409 	unsigned int (*check_events) (struct gendisk *disk,
1410 				      unsigned int clearing);
1411 	void (*unlock_native_capacity) (struct gendisk *);
1412 	int (*getgeo)(struct block_device *, struct hd_geometry *);
1413 	int (*set_read_only)(struct block_device *bdev, bool ro);
1414 	void (*free_disk)(struct gendisk *disk);
1415 	/* this callback is with swap_lock and sometimes page table lock held */
1416 	void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1417 	int (*report_zones)(struct gendisk *, sector_t sector,
1418 			unsigned int nr_zones, report_zones_cb cb, void *data);
1419 	char *(*devnode)(struct gendisk *disk, umode_t *mode);
1420 	/* returns the length of the identifier or a negative errno: */
1421 	int (*get_unique_id)(struct gendisk *disk, u8 id[16],
1422 			enum blk_unique_id id_type);
1423 	struct module *owner;
1424 	const struct pr_ops *pr_ops;
1425 
1426 	/*
1427 	 * Special callback for probing GPT entry at a given sector.
1428 	 * Needed by Android devices, used by GPT scanner and MMC blk
1429 	 * driver.
1430 	 */
1431 	int (*alternative_gpt_sector)(struct gendisk *disk, sector_t *sector);
1432 };
1433 
1434 #ifdef CONFIG_COMPAT
1435 extern int blkdev_compat_ptr_ioctl(struct block_device *, blk_mode_t,
1436 				      unsigned int, unsigned long);
1437 #else
1438 #define blkdev_compat_ptr_ioctl NULL
1439 #endif
1440 
1441 static inline void blk_wake_io_task(struct task_struct *waiter)
1442 {
1443 	/*
1444 	 * If we're polling, the task itself is doing the completions. For
1445 	 * that case, we don't need to signal a wakeup, it's enough to just
1446 	 * mark us as RUNNING.
1447 	 */
1448 	if (waiter == current)
1449 		__set_current_state(TASK_RUNNING);
1450 	else
1451 		wake_up_process(waiter);
1452 }
1453 
1454 unsigned long bdev_start_io_acct(struct block_device *bdev, enum req_op op,
1455 				 unsigned long start_time);
1456 void bdev_end_io_acct(struct block_device *bdev, enum req_op op,
1457 		      unsigned int sectors, unsigned long start_time);
1458 
1459 unsigned long bio_start_io_acct(struct bio *bio);
1460 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1461 		struct block_device *orig_bdev);
1462 
1463 /**
1464  * bio_end_io_acct - end I/O accounting for bio based drivers
1465  * @bio:	bio to end account for
1466  * @start_time:	start time returned by bio_start_io_acct()
1467  */
1468 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1469 {
1470 	return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1471 }
1472 
1473 int bdev_read_only(struct block_device *bdev);
1474 int set_blocksize(struct file *file, int size);
1475 
1476 int lookup_bdev(const char *pathname, dev_t *dev);
1477 
1478 void blkdev_show(struct seq_file *seqf, off_t offset);
1479 
1480 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
1481 #define BDEVT_SIZE	10	/* Largest string for MAJ:MIN for blkdev */
1482 #ifdef CONFIG_BLOCK
1483 #define BLKDEV_MAJOR_MAX	512
1484 #else
1485 #define BLKDEV_MAJOR_MAX	0
1486 #endif
1487 
1488 struct blk_holder_ops {
1489 	void (*mark_dead)(struct block_device *bdev, bool surprise);
1490 
1491 	/*
1492 	 * Sync the file system mounted on the block device.
1493 	 */
1494 	void (*sync)(struct block_device *bdev);
1495 
1496 	/*
1497 	 * Freeze the file system mounted on the block device.
1498 	 */
1499 	int (*freeze)(struct block_device *bdev);
1500 
1501 	/*
1502 	 * Thaw the file system mounted on the block device.
1503 	 */
1504 	int (*thaw)(struct block_device *bdev);
1505 };
1506 
1507 /*
1508  * For filesystems using @fs_holder_ops, the @holder argument passed to
1509  * helpers used to open and claim block devices via
1510  * bd_prepare_to_claim() must point to a superblock.
1511  */
1512 extern const struct blk_holder_ops fs_holder_ops;
1513 
1514 /*
1515  * Return the correct open flags for blkdev_get_by_* for super block flags
1516  * as stored in sb->s_flags.
1517  */
1518 #define sb_open_mode(flags) \
1519 	(BLK_OPEN_READ | BLK_OPEN_RESTRICT_WRITES | \
1520 	 (((flags) & SB_RDONLY) ? 0 : BLK_OPEN_WRITE))
1521 
1522 struct file *bdev_file_open_by_dev(dev_t dev, blk_mode_t mode, void *holder,
1523 		const struct blk_holder_ops *hops);
1524 struct file *bdev_file_open_by_path(const char *path, blk_mode_t mode,
1525 		void *holder, const struct blk_holder_ops *hops);
1526 int bd_prepare_to_claim(struct block_device *bdev, void *holder,
1527 		const struct blk_holder_ops *hops);
1528 void bd_abort_claiming(struct block_device *bdev, void *holder);
1529 
1530 /* just for blk-cgroup, don't use elsewhere */
1531 struct block_device *blkdev_get_no_open(dev_t dev);
1532 void blkdev_put_no_open(struct block_device *bdev);
1533 
1534 struct block_device *I_BDEV(struct inode *inode);
1535 struct block_device *file_bdev(struct file *bdev_file);
1536 bool disk_live(struct gendisk *disk);
1537 unsigned int block_size(struct block_device *bdev);
1538 
1539 #ifdef CONFIG_BLOCK
1540 void invalidate_bdev(struct block_device *bdev);
1541 int sync_blockdev(struct block_device *bdev);
1542 int sync_blockdev_range(struct block_device *bdev, loff_t lstart, loff_t lend);
1543 int sync_blockdev_nowait(struct block_device *bdev);
1544 void sync_bdevs(bool wait);
1545 void bdev_statx_dioalign(struct inode *inode, struct kstat *stat);
1546 void printk_all_partitions(void);
1547 int __init early_lookup_bdev(const char *pathname, dev_t *dev);
1548 #else
1549 static inline void invalidate_bdev(struct block_device *bdev)
1550 {
1551 }
1552 static inline int sync_blockdev(struct block_device *bdev)
1553 {
1554 	return 0;
1555 }
1556 static inline int sync_blockdev_nowait(struct block_device *bdev)
1557 {
1558 	return 0;
1559 }
1560 static inline void sync_bdevs(bool wait)
1561 {
1562 }
1563 static inline void bdev_statx_dioalign(struct inode *inode, struct kstat *stat)
1564 {
1565 }
1566 static inline void printk_all_partitions(void)
1567 {
1568 }
1569 static inline int early_lookup_bdev(const char *pathname, dev_t *dev)
1570 {
1571 	return -EINVAL;
1572 }
1573 #endif /* CONFIG_BLOCK */
1574 
1575 int bdev_freeze(struct block_device *bdev);
1576 int bdev_thaw(struct block_device *bdev);
1577 void bdev_fput(struct file *bdev_file);
1578 
1579 struct io_comp_batch {
1580 	struct request *req_list;
1581 	bool need_ts;
1582 	void (*complete)(struct io_comp_batch *);
1583 };
1584 
1585 #define DEFINE_IO_COMP_BATCH(name)	struct io_comp_batch name = { }
1586 
1587 #endif /* _LINUX_BLKDEV_H */
1588