xref: /linux-6.15/include/linux/blkdev.h (revision 162bd18e)
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 
28 struct module;
29 struct request_queue;
30 struct elevator_queue;
31 struct blk_trace;
32 struct request;
33 struct sg_io_hdr;
34 struct blkcg_gq;
35 struct blk_flush_queue;
36 struct kiocb;
37 struct pr_ops;
38 struct rq_qos;
39 struct blk_queue_stats;
40 struct blk_stat_callback;
41 struct blk_crypto_profile;
42 
43 extern const struct device_type disk_type;
44 extern struct device_type part_type;
45 extern struct class block_class;
46 
47 /*
48  * Maximum number of blkcg policies allowed to be registered concurrently.
49  * Defined here to simplify include dependency.
50  */
51 #define BLKCG_MAX_POLS		6
52 
53 #define DISK_MAX_PARTS			256
54 #define DISK_NAME_LEN			32
55 
56 #define PARTITION_META_INFO_VOLNAMELTH	64
57 /*
58  * Enough for the string representation of any kind of UUID plus NULL.
59  * EFI UUID is 36 characters. MSDOS UUID is 11 characters.
60  */
61 #define PARTITION_META_INFO_UUIDLTH	(UUID_STRING_LEN + 1)
62 
63 struct partition_meta_info {
64 	char uuid[PARTITION_META_INFO_UUIDLTH];
65 	u8 volname[PARTITION_META_INFO_VOLNAMELTH];
66 };
67 
68 /**
69  * DOC: genhd capability flags
70  *
71  * ``GENHD_FL_REMOVABLE``: indicates that the block device gives access to
72  * removable media.  When set, the device remains present even when media is not
73  * inserted.  Shall not be set for devices which are removed entirely when the
74  * media is removed.
75  *
76  * ``GENHD_FL_HIDDEN``: the block device is hidden; it doesn't produce events,
77  * doesn't appear in sysfs, and can't be opened from userspace or using
78  * blkdev_get*. Used for the underlying components of multipath devices.
79  *
80  * ``GENHD_FL_NO_PART``: partition support is disabled.  The kernel will not
81  * scan for partitions from add_disk, and users can't add partitions manually.
82  *
83  */
84 enum {
85 	GENHD_FL_REMOVABLE			= 1 << 0,
86 	GENHD_FL_HIDDEN				= 1 << 1,
87 	GENHD_FL_NO_PART			= 1 << 2,
88 };
89 
90 enum {
91 	DISK_EVENT_MEDIA_CHANGE			= 1 << 0, /* media changed */
92 	DISK_EVENT_EJECT_REQUEST		= 1 << 1, /* eject requested */
93 };
94 
95 enum {
96 	/* Poll even if events_poll_msecs is unset */
97 	DISK_EVENT_FLAG_POLL			= 1 << 0,
98 	/* Forward events to udev */
99 	DISK_EVENT_FLAG_UEVENT			= 1 << 1,
100 	/* Block event polling when open for exclusive write */
101 	DISK_EVENT_FLAG_BLOCK_ON_EXCL_WRITE	= 1 << 2,
102 };
103 
104 struct disk_events;
105 struct badblocks;
106 
107 struct blk_integrity {
108 	const struct blk_integrity_profile	*profile;
109 	unsigned char				flags;
110 	unsigned char				tuple_size;
111 	unsigned char				interval_exp;
112 	unsigned char				tag_size;
113 };
114 
115 struct gendisk {
116 	/*
117 	 * major/first_minor/minors should not be set by any new driver, the
118 	 * block core will take care of allocating them automatically.
119 	 */
120 	int major;
121 	int first_minor;
122 	int minors;
123 
124 	char disk_name[DISK_NAME_LEN];	/* name of major driver */
125 
126 	unsigned short events;		/* supported events */
127 	unsigned short event_flags;	/* flags related to event processing */
128 
129 	struct xarray part_tbl;
130 	struct block_device *part0;
131 
132 	const struct block_device_operations *fops;
133 	struct request_queue *queue;
134 	void *private_data;
135 
136 	struct bio_set bio_split;
137 
138 	int flags;
139 	unsigned long state;
140 #define GD_NEED_PART_SCAN		0
141 #define GD_READ_ONLY			1
142 #define GD_DEAD				2
143 #define GD_NATIVE_CAPACITY		3
144 #define GD_ADDED			4
145 #define GD_SUPPRESS_PART_SCAN		5
146 #define GD_OWNS_QUEUE			6
147 
148 	struct mutex open_mutex;	/* open/close mutex */
149 	unsigned open_partitions;	/* number of open partitions */
150 
151 	struct backing_dev_info	*bdi;
152 	struct kobject queue_kobj;	/* the queue/ directory */
153 	struct kobject *slave_dir;
154 #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED
155 	struct list_head slave_bdevs;
156 #endif
157 	struct timer_rand_state *random;
158 	atomic_t sync_io;		/* RAID */
159 	struct disk_events *ev;
160 #ifdef  CONFIG_BLK_DEV_INTEGRITY
161 	struct kobject integrity_kobj;
162 #endif	/* CONFIG_BLK_DEV_INTEGRITY */
163 
164 #ifdef CONFIG_BLK_DEV_ZONED
165 	/*
166 	 * Zoned block device information for request dispatch control.
167 	 * nr_zones is the total number of zones of the device. This is always
168 	 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
169 	 * bits which indicates if a zone is conventional (bit set) or
170 	 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
171 	 * bits which indicates if a zone is write locked, that is, if a write
172 	 * request targeting the zone was dispatched.
173 	 *
174 	 * Reads of this information must be protected with blk_queue_enter() /
175 	 * blk_queue_exit(). Modifying this information is only allowed while
176 	 * no requests are being processed. See also blk_mq_freeze_queue() and
177 	 * blk_mq_unfreeze_queue().
178 	 */
179 	unsigned int		nr_zones;
180 	unsigned int		max_open_zones;
181 	unsigned int		max_active_zones;
182 	unsigned long		*conv_zones_bitmap;
183 	unsigned long		*seq_zones_wlock;
184 #endif /* CONFIG_BLK_DEV_ZONED */
185 
186 #if IS_ENABLED(CONFIG_CDROM)
187 	struct cdrom_device_info *cdi;
188 #endif
189 	int node_id;
190 	struct badblocks *bb;
191 	struct lockdep_map lockdep_map;
192 	u64 diskseq;
193 
194 	/*
195 	 * Independent sector access ranges. This is always NULL for
196 	 * devices that do not have multiple independent access ranges.
197 	 */
198 	struct blk_independent_access_ranges *ia_ranges;
199 };
200 
201 static inline bool disk_live(struct gendisk *disk)
202 {
203 	return !inode_unhashed(disk->part0->bd_inode);
204 }
205 
206 /**
207  * disk_openers - returns how many openers are there for a disk
208  * @disk: disk to check
209  *
210  * This returns the number of openers for a disk.  Note that this value is only
211  * stable if disk->open_mutex is held.
212  *
213  * Note: Due to a quirk in the block layer open code, each open partition is
214  * only counted once even if there are multiple openers.
215  */
216 static inline unsigned int disk_openers(struct gendisk *disk)
217 {
218 	return atomic_read(&disk->part0->bd_openers);
219 }
220 
221 /*
222  * The gendisk is refcounted by the part0 block_device, and the bd_device
223  * therein is also used for device model presentation in sysfs.
224  */
225 #define dev_to_disk(device) \
226 	(dev_to_bdev(device)->bd_disk)
227 #define disk_to_dev(disk) \
228 	(&((disk)->part0->bd_device))
229 
230 #if IS_REACHABLE(CONFIG_CDROM)
231 #define disk_to_cdi(disk)	((disk)->cdi)
232 #else
233 #define disk_to_cdi(disk)	NULL
234 #endif
235 
236 static inline dev_t disk_devt(struct gendisk *disk)
237 {
238 	return MKDEV(disk->major, disk->first_minor);
239 }
240 
241 static inline int blk_validate_block_size(unsigned long bsize)
242 {
243 	if (bsize < 512 || bsize > PAGE_SIZE || !is_power_of_2(bsize))
244 		return -EINVAL;
245 
246 	return 0;
247 }
248 
249 static inline bool blk_op_is_passthrough(blk_opf_t op)
250 {
251 	op &= REQ_OP_MASK;
252 	return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
253 }
254 
255 /*
256  * Zoned block device models (zoned limit).
257  *
258  * Note: This needs to be ordered from the least to the most severe
259  * restrictions for the inheritance in blk_stack_limits() to work.
260  */
261 enum blk_zoned_model {
262 	BLK_ZONED_NONE = 0,	/* Regular block device */
263 	BLK_ZONED_HA,		/* Host-aware zoned block device */
264 	BLK_ZONED_HM,		/* Host-managed zoned block device */
265 };
266 
267 /*
268  * BLK_BOUNCE_NONE:	never bounce (default)
269  * BLK_BOUNCE_HIGH:	bounce all highmem pages
270  */
271 enum blk_bounce {
272 	BLK_BOUNCE_NONE,
273 	BLK_BOUNCE_HIGH,
274 };
275 
276 struct queue_limits {
277 	enum blk_bounce		bounce;
278 	unsigned long		seg_boundary_mask;
279 	unsigned long		virt_boundary_mask;
280 
281 	unsigned int		max_hw_sectors;
282 	unsigned int		max_dev_sectors;
283 	unsigned int		chunk_sectors;
284 	unsigned int		max_sectors;
285 	unsigned int		max_user_sectors;
286 	unsigned int		max_segment_size;
287 	unsigned int		physical_block_size;
288 	unsigned int		logical_block_size;
289 	unsigned int		alignment_offset;
290 	unsigned int		io_min;
291 	unsigned int		io_opt;
292 	unsigned int		max_discard_sectors;
293 	unsigned int		max_hw_discard_sectors;
294 	unsigned int		max_secure_erase_sectors;
295 	unsigned int		max_write_zeroes_sectors;
296 	unsigned int		max_zone_append_sectors;
297 	unsigned int		discard_granularity;
298 	unsigned int		discard_alignment;
299 	unsigned int		zone_write_granularity;
300 
301 	unsigned short		max_segments;
302 	unsigned short		max_integrity_segments;
303 	unsigned short		max_discard_segments;
304 
305 	unsigned char		misaligned;
306 	unsigned char		discard_misaligned;
307 	unsigned char		raid_partial_stripes_expensive;
308 	enum blk_zoned_model	zoned;
309 
310 	/*
311 	 * Drivers that set dma_alignment to less than 511 must be prepared to
312 	 * handle individual bvec's that are not a multiple of a SECTOR_SIZE
313 	 * due to possible offsets.
314 	 */
315 	unsigned int		dma_alignment;
316 };
317 
318 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
319 			       void *data);
320 
321 void disk_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
322 
323 #ifdef CONFIG_BLK_DEV_ZONED
324 
325 #define BLK_ALL_ZONES  ((unsigned int)-1)
326 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
327 			unsigned int nr_zones, report_zones_cb cb, void *data);
328 unsigned int bdev_nr_zones(struct block_device *bdev);
329 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_op op,
330 			    sector_t sectors, sector_t nr_sectors,
331 			    gfp_t gfp_mask);
332 int blk_revalidate_disk_zones(struct gendisk *disk,
333 			      void (*update_driver_data)(struct gendisk *disk));
334 
335 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
336 				     unsigned int cmd, unsigned long arg);
337 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
338 				  unsigned int cmd, unsigned long arg);
339 
340 #else /* CONFIG_BLK_DEV_ZONED */
341 
342 static inline unsigned int bdev_nr_zones(struct block_device *bdev)
343 {
344 	return 0;
345 }
346 
347 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
348 					    fmode_t mode, unsigned int cmd,
349 					    unsigned long arg)
350 {
351 	return -ENOTTY;
352 }
353 
354 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
355 					 fmode_t mode, unsigned int cmd,
356 					 unsigned long arg)
357 {
358 	return -ENOTTY;
359 }
360 
361 #endif /* CONFIG_BLK_DEV_ZONED */
362 
363 /*
364  * Independent access ranges: struct blk_independent_access_range describes
365  * a range of contiguous sectors that can be accessed using device command
366  * execution resources that are independent from the resources used for
367  * other access ranges. This is typically found with single-LUN multi-actuator
368  * HDDs where each access range is served by a different set of heads.
369  * The set of independent ranges supported by the device is defined using
370  * struct blk_independent_access_ranges. The independent ranges must not overlap
371  * and must include all sectors within the disk capacity (no sector holes
372  * allowed).
373  * For a device with multiple ranges, requests targeting sectors in different
374  * ranges can be executed in parallel. A request can straddle an access range
375  * boundary.
376  */
377 struct blk_independent_access_range {
378 	struct kobject		kobj;
379 	sector_t		sector;
380 	sector_t		nr_sectors;
381 };
382 
383 struct blk_independent_access_ranges {
384 	struct kobject				kobj;
385 	bool					sysfs_registered;
386 	unsigned int				nr_ia_ranges;
387 	struct blk_independent_access_range	ia_range[];
388 };
389 
390 struct request_queue {
391 	struct request		*last_merge;
392 	struct elevator_queue	*elevator;
393 
394 	struct percpu_ref	q_usage_counter;
395 
396 	struct blk_queue_stats	*stats;
397 	struct rq_qos		*rq_qos;
398 
399 	const struct blk_mq_ops	*mq_ops;
400 
401 	/* sw queues */
402 	struct blk_mq_ctx __percpu	*queue_ctx;
403 
404 	unsigned int		queue_depth;
405 
406 	/* hw dispatch queues */
407 	struct xarray		hctx_table;
408 	unsigned int		nr_hw_queues;
409 
410 	/*
411 	 * The queue owner gets to use this for whatever they like.
412 	 * ll_rw_blk doesn't touch it.
413 	 */
414 	void			*queuedata;
415 
416 	/*
417 	 * various queue flags, see QUEUE_* below
418 	 */
419 	unsigned long		queue_flags;
420 	/*
421 	 * Number of contexts that have called blk_set_pm_only(). If this
422 	 * counter is above zero then only RQF_PM requests are processed.
423 	 */
424 	atomic_t		pm_only;
425 
426 	/*
427 	 * ida allocated id for this queue.  Used to index queues from
428 	 * ioctx.
429 	 */
430 	int			id;
431 
432 	spinlock_t		queue_lock;
433 
434 	struct gendisk		*disk;
435 
436 	refcount_t		refs;
437 
438 	/*
439 	 * mq queue kobject
440 	 */
441 	struct kobject *mq_kobj;
442 
443 #ifdef  CONFIG_BLK_DEV_INTEGRITY
444 	struct blk_integrity integrity;
445 #endif	/* CONFIG_BLK_DEV_INTEGRITY */
446 
447 #ifdef CONFIG_PM
448 	struct device		*dev;
449 	enum rpm_status		rpm_status;
450 #endif
451 
452 	/*
453 	 * queue settings
454 	 */
455 	unsigned long		nr_requests;	/* Max # of requests */
456 
457 	unsigned int		dma_pad_mask;
458 
459 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
460 	struct blk_crypto_profile *crypto_profile;
461 	struct kobject *crypto_kobject;
462 #endif
463 
464 	unsigned int		rq_timeout;
465 
466 	struct timer_list	timeout;
467 	struct work_struct	timeout_work;
468 
469 	atomic_t		nr_active_requests_shared_tags;
470 
471 	struct blk_mq_tags	*sched_shared_tags;
472 
473 	struct list_head	icq_list;
474 #ifdef CONFIG_BLK_CGROUP
475 	DECLARE_BITMAP		(blkcg_pols, BLKCG_MAX_POLS);
476 	struct blkcg_gq		*root_blkg;
477 	struct list_head	blkg_list;
478 	struct mutex		blkcg_mutex;
479 #endif
480 
481 	struct queue_limits	limits;
482 
483 	unsigned int		required_elevator_features;
484 
485 	int			node;
486 #ifdef CONFIG_BLK_DEV_IO_TRACE
487 	struct blk_trace __rcu	*blk_trace;
488 #endif
489 	/*
490 	 * for flush operations
491 	 */
492 	struct blk_flush_queue	*fq;
493 
494 	struct list_head	requeue_list;
495 	spinlock_t		requeue_lock;
496 	struct delayed_work	requeue_work;
497 
498 	struct mutex		sysfs_lock;
499 	struct mutex		sysfs_dir_lock;
500 
501 	/*
502 	 * for reusing dead hctx instance in case of updating
503 	 * nr_hw_queues
504 	 */
505 	struct list_head	unused_hctx_list;
506 	spinlock_t		unused_hctx_lock;
507 
508 	int			mq_freeze_depth;
509 
510 #ifdef CONFIG_BLK_DEV_THROTTLING
511 	/* Throttle data */
512 	struct throtl_data *td;
513 #endif
514 	struct rcu_head		rcu_head;
515 	wait_queue_head_t	mq_freeze_wq;
516 	/*
517 	 * Protect concurrent access to q_usage_counter by
518 	 * percpu_ref_kill() and percpu_ref_reinit().
519 	 */
520 	struct mutex		mq_freeze_lock;
521 
522 	int			quiesce_depth;
523 
524 	struct blk_mq_tag_set	*tag_set;
525 	struct list_head	tag_set_list;
526 
527 	struct dentry		*debugfs_dir;
528 	struct dentry		*sched_debugfs_dir;
529 	struct dentry		*rqos_debugfs_dir;
530 	/*
531 	 * Serializes all debugfs metadata operations using the above dentries.
532 	 */
533 	struct mutex		debugfs_mutex;
534 
535 	bool			mq_sysfs_init_done;
536 };
537 
538 /* Keep blk_queue_flag_name[] in sync with the definitions below */
539 #define QUEUE_FLAG_STOPPED	0	/* queue is stopped */
540 #define QUEUE_FLAG_DYING	1	/* queue being torn down */
541 #define QUEUE_FLAG_NOMERGES     3	/* disable merge attempts */
542 #define QUEUE_FLAG_SAME_COMP	4	/* complete on same CPU-group */
543 #define QUEUE_FLAG_FAIL_IO	5	/* fake timeout */
544 #define QUEUE_FLAG_NONROT	6	/* non-rotational device (SSD) */
545 #define QUEUE_FLAG_VIRT		QUEUE_FLAG_NONROT /* paravirt device */
546 #define QUEUE_FLAG_IO_STAT	7	/* do disk/partitions IO accounting */
547 #define QUEUE_FLAG_NOXMERGES	9	/* No extended merges */
548 #define QUEUE_FLAG_ADD_RANDOM	10	/* Contributes to random pool */
549 #define QUEUE_FLAG_SYNCHRONOUS	11	/* always completes in submit context */
550 #define QUEUE_FLAG_SAME_FORCE	12	/* force complete on same CPU */
551 #define QUEUE_FLAG_INIT_DONE	14	/* queue is initialized */
552 #define QUEUE_FLAG_STABLE_WRITES 15	/* don't modify blks until WB is done */
553 #define QUEUE_FLAG_POLL		16	/* IO polling enabled if set */
554 #define QUEUE_FLAG_WC		17	/* Write back caching */
555 #define QUEUE_FLAG_FUA		18	/* device supports FUA writes */
556 #define QUEUE_FLAG_DAX		19	/* device supports DAX */
557 #define QUEUE_FLAG_STATS	20	/* track IO start and completion times */
558 #define QUEUE_FLAG_REGISTERED	22	/* queue has been registered to a disk */
559 #define QUEUE_FLAG_QUIESCED	24	/* queue has been quiesced */
560 #define QUEUE_FLAG_PCI_P2PDMA	25	/* device supports PCI p2p requests */
561 #define QUEUE_FLAG_ZONE_RESETALL 26	/* supports Zone Reset All */
562 #define QUEUE_FLAG_RQ_ALLOC_TIME 27	/* record rq->alloc_time_ns */
563 #define QUEUE_FLAG_HCTX_ACTIVE	28	/* at least one blk-mq hctx is active */
564 #define QUEUE_FLAG_NOWAIT       29	/* device supports NOWAIT */
565 #define QUEUE_FLAG_SQ_SCHED     30	/* single queue style io dispatch */
566 #define QUEUE_FLAG_SKIP_TAGSET_QUIESCE	31 /* quiesce_tagset skip the queue*/
567 
568 #define QUEUE_FLAG_MQ_DEFAULT	((1UL << QUEUE_FLAG_IO_STAT) |		\
569 				 (1UL << QUEUE_FLAG_SAME_COMP) |	\
570 				 (1UL << QUEUE_FLAG_NOWAIT))
571 
572 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
573 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
574 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
575 
576 #define blk_queue_stopped(q)	test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
577 #define blk_queue_dying(q)	test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
578 #define blk_queue_init_done(q)	test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
579 #define blk_queue_nomerges(q)	test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
580 #define blk_queue_noxmerges(q)	\
581 	test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
582 #define blk_queue_nonrot(q)	test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
583 #define blk_queue_stable_writes(q) \
584 	test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
585 #define blk_queue_io_stat(q)	test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
586 #define blk_queue_add_random(q)	test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
587 #define blk_queue_zone_resetall(q)	\
588 	test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
589 #define blk_queue_dax(q)	test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
590 #define blk_queue_pci_p2pdma(q)	\
591 	test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
592 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
593 #define blk_queue_rq_alloc_time(q)	\
594 	test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
595 #else
596 #define blk_queue_rq_alloc_time(q)	false
597 #endif
598 
599 #define blk_noretry_request(rq) \
600 	((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
601 			     REQ_FAILFAST_DRIVER))
602 #define blk_queue_quiesced(q)	test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
603 #define blk_queue_pm_only(q)	atomic_read(&(q)->pm_only)
604 #define blk_queue_registered(q)	test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
605 #define blk_queue_sq_sched(q)	test_bit(QUEUE_FLAG_SQ_SCHED, &(q)->queue_flags)
606 #define blk_queue_skip_tagset_quiesce(q) \
607 	test_bit(QUEUE_FLAG_SKIP_TAGSET_QUIESCE, &(q)->queue_flags)
608 
609 extern void blk_set_pm_only(struct request_queue *q);
610 extern void blk_clear_pm_only(struct request_queue *q);
611 
612 #define list_entry_rq(ptr)	list_entry((ptr), struct request, queuelist)
613 
614 #define dma_map_bvec(dev, bv, dir, attrs) \
615 	dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
616 	(dir), (attrs))
617 
618 static inline bool queue_is_mq(struct request_queue *q)
619 {
620 	return q->mq_ops;
621 }
622 
623 #ifdef CONFIG_PM
624 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
625 {
626 	return q->rpm_status;
627 }
628 #else
629 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
630 {
631 	return RPM_ACTIVE;
632 }
633 #endif
634 
635 static inline enum blk_zoned_model
636 blk_queue_zoned_model(struct request_queue *q)
637 {
638 	if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
639 		return q->limits.zoned;
640 	return BLK_ZONED_NONE;
641 }
642 
643 static inline bool blk_queue_is_zoned(struct request_queue *q)
644 {
645 	switch (blk_queue_zoned_model(q)) {
646 	case BLK_ZONED_HA:
647 	case BLK_ZONED_HM:
648 		return true;
649 	default:
650 		return false;
651 	}
652 }
653 
654 #ifdef CONFIG_BLK_DEV_ZONED
655 static inline unsigned int disk_nr_zones(struct gendisk *disk)
656 {
657 	return blk_queue_is_zoned(disk->queue) ? disk->nr_zones : 0;
658 }
659 
660 static inline unsigned int disk_zone_no(struct gendisk *disk, sector_t sector)
661 {
662 	if (!blk_queue_is_zoned(disk->queue))
663 		return 0;
664 	return sector >> ilog2(disk->queue->limits.chunk_sectors);
665 }
666 
667 static inline bool disk_zone_is_seq(struct gendisk *disk, sector_t sector)
668 {
669 	if (!blk_queue_is_zoned(disk->queue))
670 		return false;
671 	if (!disk->conv_zones_bitmap)
672 		return true;
673 	return !test_bit(disk_zone_no(disk, sector), disk->conv_zones_bitmap);
674 }
675 
676 static inline void disk_set_max_open_zones(struct gendisk *disk,
677 		unsigned int max_open_zones)
678 {
679 	disk->max_open_zones = max_open_zones;
680 }
681 
682 static inline void disk_set_max_active_zones(struct gendisk *disk,
683 		unsigned int max_active_zones)
684 {
685 	disk->max_active_zones = max_active_zones;
686 }
687 
688 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
689 {
690 	return bdev->bd_disk->max_open_zones;
691 }
692 
693 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
694 {
695 	return bdev->bd_disk->max_active_zones;
696 }
697 
698 #else /* CONFIG_BLK_DEV_ZONED */
699 static inline unsigned int disk_nr_zones(struct gendisk *disk)
700 {
701 	return 0;
702 }
703 static inline bool disk_zone_is_seq(struct gendisk *disk, sector_t sector)
704 {
705 	return false;
706 }
707 static inline unsigned int disk_zone_no(struct gendisk *disk, sector_t sector)
708 {
709 	return 0;
710 }
711 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
712 {
713 	return 0;
714 }
715 
716 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
717 {
718 	return 0;
719 }
720 #endif /* CONFIG_BLK_DEV_ZONED */
721 
722 static inline unsigned int blk_queue_depth(struct request_queue *q)
723 {
724 	if (q->queue_depth)
725 		return q->queue_depth;
726 
727 	return q->nr_requests;
728 }
729 
730 /*
731  * default timeout for SG_IO if none specified
732  */
733 #define BLK_DEFAULT_SG_TIMEOUT	(60 * HZ)
734 #define BLK_MIN_SG_TIMEOUT	(7 * HZ)
735 
736 /* This should not be used directly - use rq_for_each_segment */
737 #define for_each_bio(_bio)		\
738 	for (; _bio; _bio = _bio->bi_next)
739 
740 int __must_check device_add_disk(struct device *parent, struct gendisk *disk,
741 				 const struct attribute_group **groups);
742 static inline int __must_check add_disk(struct gendisk *disk)
743 {
744 	return device_add_disk(NULL, disk, NULL);
745 }
746 void del_gendisk(struct gendisk *gp);
747 void invalidate_disk(struct gendisk *disk);
748 void set_disk_ro(struct gendisk *disk, bool read_only);
749 void disk_uevent(struct gendisk *disk, enum kobject_action action);
750 
751 static inline int get_disk_ro(struct gendisk *disk)
752 {
753 	return disk->part0->bd_read_only ||
754 		test_bit(GD_READ_ONLY, &disk->state);
755 }
756 
757 static inline int bdev_read_only(struct block_device *bdev)
758 {
759 	return bdev->bd_read_only || get_disk_ro(bdev->bd_disk);
760 }
761 
762 bool set_capacity_and_notify(struct gendisk *disk, sector_t size);
763 bool disk_force_media_change(struct gendisk *disk, unsigned int events);
764 
765 void add_disk_randomness(struct gendisk *disk) __latent_entropy;
766 void rand_initialize_disk(struct gendisk *disk);
767 
768 static inline sector_t get_start_sect(struct block_device *bdev)
769 {
770 	return bdev->bd_start_sect;
771 }
772 
773 static inline sector_t bdev_nr_sectors(struct block_device *bdev)
774 {
775 	return bdev->bd_nr_sectors;
776 }
777 
778 static inline loff_t bdev_nr_bytes(struct block_device *bdev)
779 {
780 	return (loff_t)bdev_nr_sectors(bdev) << SECTOR_SHIFT;
781 }
782 
783 static inline sector_t get_capacity(struct gendisk *disk)
784 {
785 	return bdev_nr_sectors(disk->part0);
786 }
787 
788 static inline u64 sb_bdev_nr_blocks(struct super_block *sb)
789 {
790 	return bdev_nr_sectors(sb->s_bdev) >>
791 		(sb->s_blocksize_bits - SECTOR_SHIFT);
792 }
793 
794 int bdev_disk_changed(struct gendisk *disk, bool invalidate);
795 
796 void put_disk(struct gendisk *disk);
797 struct gendisk *__blk_alloc_disk(int node, struct lock_class_key *lkclass);
798 
799 /**
800  * blk_alloc_disk - allocate a gendisk structure
801  * @node_id: numa node to allocate on
802  *
803  * Allocate and pre-initialize a gendisk structure for use with BIO based
804  * drivers.
805  *
806  * Context: can sleep
807  */
808 #define blk_alloc_disk(node_id)						\
809 ({									\
810 	static struct lock_class_key __key;				\
811 									\
812 	__blk_alloc_disk(node_id, &__key);				\
813 })
814 
815 int __register_blkdev(unsigned int major, const char *name,
816 		void (*probe)(dev_t devt));
817 #define register_blkdev(major, name) \
818 	__register_blkdev(major, name, NULL)
819 void unregister_blkdev(unsigned int major, const char *name);
820 
821 bool bdev_check_media_change(struct block_device *bdev);
822 int __invalidate_device(struct block_device *bdev, bool kill_dirty);
823 void set_capacity(struct gendisk *disk, sector_t size);
824 
825 #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED
826 int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
827 void bd_unlink_disk_holder(struct block_device *bdev, struct gendisk *disk);
828 #else
829 static inline int bd_link_disk_holder(struct block_device *bdev,
830 				      struct gendisk *disk)
831 {
832 	return 0;
833 }
834 static inline void bd_unlink_disk_holder(struct block_device *bdev,
835 					 struct gendisk *disk)
836 {
837 }
838 #endif /* CONFIG_BLOCK_HOLDER_DEPRECATED */
839 
840 dev_t part_devt(struct gendisk *disk, u8 partno);
841 void inc_diskseq(struct gendisk *disk);
842 dev_t blk_lookup_devt(const char *name, int partno);
843 void blk_request_module(dev_t devt);
844 
845 extern int blk_register_queue(struct gendisk *disk);
846 extern void blk_unregister_queue(struct gendisk *disk);
847 void submit_bio_noacct(struct bio *bio);
848 struct bio *bio_split_to_limits(struct bio *bio);
849 
850 extern int blk_lld_busy(struct request_queue *q);
851 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
852 extern void blk_queue_exit(struct request_queue *q);
853 extern void blk_sync_queue(struct request_queue *q);
854 
855 /* Helper to convert REQ_OP_XXX to its string format XXX */
856 extern const char *blk_op_str(enum req_op op);
857 
858 int blk_status_to_errno(blk_status_t status);
859 blk_status_t errno_to_blk_status(int errno);
860 
861 /* only poll the hardware once, don't continue until a completion was found */
862 #define BLK_POLL_ONESHOT		(1 << 0)
863 int bio_poll(struct bio *bio, struct io_comp_batch *iob, unsigned int flags);
864 int iocb_bio_iopoll(struct kiocb *kiocb, struct io_comp_batch *iob,
865 			unsigned int flags);
866 
867 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
868 {
869 	return bdev->bd_queue;	/* this is never NULL */
870 }
871 
872 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
873 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
874 
875 static inline unsigned int bio_zone_no(struct bio *bio)
876 {
877 	return disk_zone_no(bio->bi_bdev->bd_disk, bio->bi_iter.bi_sector);
878 }
879 
880 static inline unsigned int bio_zone_is_seq(struct bio *bio)
881 {
882 	return disk_zone_is_seq(bio->bi_bdev->bd_disk, bio->bi_iter.bi_sector);
883 }
884 
885 /*
886  * Return how much of the chunk is left to be used for I/O at a given offset.
887  */
888 static inline unsigned int blk_chunk_sectors_left(sector_t offset,
889 		unsigned int chunk_sectors)
890 {
891 	if (unlikely(!is_power_of_2(chunk_sectors)))
892 		return chunk_sectors - sector_div(offset, chunk_sectors);
893 	return chunk_sectors - (offset & (chunk_sectors - 1));
894 }
895 
896 /*
897  * Access functions for manipulating queue properties
898  */
899 void blk_queue_bounce_limit(struct request_queue *q, enum blk_bounce limit);
900 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
901 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
902 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
903 extern void blk_queue_max_discard_segments(struct request_queue *,
904 		unsigned short);
905 void blk_queue_max_secure_erase_sectors(struct request_queue *q,
906 		unsigned int max_sectors);
907 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
908 extern void blk_queue_max_discard_sectors(struct request_queue *q,
909 		unsigned int max_discard_sectors);
910 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
911 		unsigned int max_write_same_sectors);
912 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
913 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
914 		unsigned int max_zone_append_sectors);
915 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
916 void blk_queue_zone_write_granularity(struct request_queue *q,
917 				      unsigned int size);
918 extern void blk_queue_alignment_offset(struct request_queue *q,
919 				       unsigned int alignment);
920 void disk_update_readahead(struct gendisk *disk);
921 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
922 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
923 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
924 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
925 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
926 extern void blk_set_stacking_limits(struct queue_limits *lim);
927 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
928 			    sector_t offset);
929 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
930 			      sector_t offset);
931 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
932 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
933 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
934 extern void blk_queue_dma_alignment(struct request_queue *, int);
935 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
936 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
937 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
938 
939 struct blk_independent_access_ranges *
940 disk_alloc_independent_access_ranges(struct gendisk *disk, int nr_ia_ranges);
941 void disk_set_independent_access_ranges(struct gendisk *disk,
942 				struct blk_independent_access_ranges *iars);
943 
944 /*
945  * Elevator features for blk_queue_required_elevator_features:
946  */
947 /* Supports zoned block devices sequential write constraint */
948 #define ELEVATOR_F_ZBD_SEQ_WRITE	(1U << 0)
949 
950 extern void blk_queue_required_elevator_features(struct request_queue *q,
951 						 unsigned int features);
952 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
953 					      struct device *dev);
954 
955 bool __must_check blk_get_queue(struct request_queue *);
956 extern void blk_put_queue(struct request_queue *);
957 
958 void blk_mark_disk_dead(struct gendisk *disk);
959 
960 #ifdef CONFIG_BLOCK
961 /*
962  * blk_plug permits building a queue of related requests by holding the I/O
963  * fragments for a short period. This allows merging of sequential requests
964  * into single larger request. As the requests are moved from a per-task list to
965  * the device's request_queue in a batch, this results in improved scalability
966  * as the lock contention for request_queue lock is reduced.
967  *
968  * It is ok not to disable preemption when adding the request to the plug list
969  * or when attempting a merge. For details, please see schedule() where
970  * blk_flush_plug() is called.
971  */
972 struct blk_plug {
973 	struct request *mq_list; /* blk-mq requests */
974 
975 	/* if ios_left is > 1, we can batch tag/rq allocations */
976 	struct request *cached_rq;
977 	unsigned short nr_ios;
978 
979 	unsigned short rq_count;
980 
981 	bool multiple_queues;
982 	bool has_elevator;
983 	bool nowait;
984 
985 	struct list_head cb_list; /* md requires an unplug callback */
986 };
987 
988 struct blk_plug_cb;
989 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
990 struct blk_plug_cb {
991 	struct list_head list;
992 	blk_plug_cb_fn callback;
993 	void *data;
994 };
995 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
996 					     void *data, int size);
997 extern void blk_start_plug(struct blk_plug *);
998 extern void blk_start_plug_nr_ios(struct blk_plug *, unsigned short);
999 extern void blk_finish_plug(struct blk_plug *);
1000 
1001 void __blk_flush_plug(struct blk_plug *plug, bool from_schedule);
1002 static inline void blk_flush_plug(struct blk_plug *plug, bool async)
1003 {
1004 	if (plug)
1005 		__blk_flush_plug(plug, async);
1006 }
1007 
1008 int blkdev_issue_flush(struct block_device *bdev);
1009 long nr_blockdev_pages(void);
1010 #else /* CONFIG_BLOCK */
1011 struct blk_plug {
1012 };
1013 
1014 static inline void blk_start_plug_nr_ios(struct blk_plug *plug,
1015 					 unsigned short nr_ios)
1016 {
1017 }
1018 
1019 static inline void blk_start_plug(struct blk_plug *plug)
1020 {
1021 }
1022 
1023 static inline void blk_finish_plug(struct blk_plug *plug)
1024 {
1025 }
1026 
1027 static inline void blk_flush_plug(struct blk_plug *plug, bool async)
1028 {
1029 }
1030 
1031 static inline int blkdev_issue_flush(struct block_device *bdev)
1032 {
1033 	return 0;
1034 }
1035 
1036 static inline long nr_blockdev_pages(void)
1037 {
1038 	return 0;
1039 }
1040 #endif /* CONFIG_BLOCK */
1041 
1042 extern void blk_io_schedule(void);
1043 
1044 int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1045 		sector_t nr_sects, gfp_t gfp_mask);
1046 int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1047 		sector_t nr_sects, gfp_t gfp_mask, struct bio **biop);
1048 int blkdev_issue_secure_erase(struct block_device *bdev, sector_t sector,
1049 		sector_t nr_sects, gfp_t gfp);
1050 
1051 #define BLKDEV_ZERO_NOUNMAP	(1 << 0)  /* do not free blocks */
1052 #define BLKDEV_ZERO_NOFALLBACK	(1 << 1)  /* don't write explicit zeroes */
1053 
1054 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1055 		sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1056 		unsigned flags);
1057 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1058 		sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1059 
1060 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1061 		sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1062 {
1063 	return blkdev_issue_discard(sb->s_bdev,
1064 				    block << (sb->s_blocksize_bits -
1065 					      SECTOR_SHIFT),
1066 				    nr_blocks << (sb->s_blocksize_bits -
1067 						  SECTOR_SHIFT),
1068 				    gfp_mask);
1069 }
1070 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1071 		sector_t nr_blocks, gfp_t gfp_mask)
1072 {
1073 	return blkdev_issue_zeroout(sb->s_bdev,
1074 				    block << (sb->s_blocksize_bits -
1075 					      SECTOR_SHIFT),
1076 				    nr_blocks << (sb->s_blocksize_bits -
1077 						  SECTOR_SHIFT),
1078 				    gfp_mask, 0);
1079 }
1080 
1081 static inline bool bdev_is_partition(struct block_device *bdev)
1082 {
1083 	return bdev->bd_partno;
1084 }
1085 
1086 enum blk_default_limits {
1087 	BLK_MAX_SEGMENTS	= 128,
1088 	BLK_SAFE_MAX_SECTORS	= 255,
1089 	BLK_MAX_SEGMENT_SIZE	= 65536,
1090 	BLK_SEG_BOUNDARY_MASK	= 0xFFFFFFFFUL,
1091 };
1092 
1093 #define BLK_DEF_MAX_SECTORS 2560u
1094 
1095 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1096 {
1097 	return q->limits.seg_boundary_mask;
1098 }
1099 
1100 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1101 {
1102 	return q->limits.virt_boundary_mask;
1103 }
1104 
1105 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1106 {
1107 	return q->limits.max_sectors;
1108 }
1109 
1110 static inline unsigned int queue_max_bytes(struct request_queue *q)
1111 {
1112 	return min_t(unsigned int, queue_max_sectors(q), INT_MAX >> 9) << 9;
1113 }
1114 
1115 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1116 {
1117 	return q->limits.max_hw_sectors;
1118 }
1119 
1120 static inline unsigned short queue_max_segments(const struct request_queue *q)
1121 {
1122 	return q->limits.max_segments;
1123 }
1124 
1125 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1126 {
1127 	return q->limits.max_discard_segments;
1128 }
1129 
1130 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1131 {
1132 	return q->limits.max_segment_size;
1133 }
1134 
1135 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1136 {
1137 
1138 	const struct queue_limits *l = &q->limits;
1139 
1140 	return min(l->max_zone_append_sectors, l->max_sectors);
1141 }
1142 
1143 static inline unsigned int
1144 bdev_max_zone_append_sectors(struct block_device *bdev)
1145 {
1146 	return queue_max_zone_append_sectors(bdev_get_queue(bdev));
1147 }
1148 
1149 static inline unsigned int bdev_max_segments(struct block_device *bdev)
1150 {
1151 	return queue_max_segments(bdev_get_queue(bdev));
1152 }
1153 
1154 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1155 {
1156 	int retval = 512;
1157 
1158 	if (q && q->limits.logical_block_size)
1159 		retval = q->limits.logical_block_size;
1160 
1161 	return retval;
1162 }
1163 
1164 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1165 {
1166 	return queue_logical_block_size(bdev_get_queue(bdev));
1167 }
1168 
1169 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1170 {
1171 	return q->limits.physical_block_size;
1172 }
1173 
1174 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1175 {
1176 	return queue_physical_block_size(bdev_get_queue(bdev));
1177 }
1178 
1179 static inline unsigned int queue_io_min(const struct request_queue *q)
1180 {
1181 	return q->limits.io_min;
1182 }
1183 
1184 static inline int bdev_io_min(struct block_device *bdev)
1185 {
1186 	return queue_io_min(bdev_get_queue(bdev));
1187 }
1188 
1189 static inline unsigned int queue_io_opt(const struct request_queue *q)
1190 {
1191 	return q->limits.io_opt;
1192 }
1193 
1194 static inline int bdev_io_opt(struct block_device *bdev)
1195 {
1196 	return queue_io_opt(bdev_get_queue(bdev));
1197 }
1198 
1199 static inline unsigned int
1200 queue_zone_write_granularity(const struct request_queue *q)
1201 {
1202 	return q->limits.zone_write_granularity;
1203 }
1204 
1205 static inline unsigned int
1206 bdev_zone_write_granularity(struct block_device *bdev)
1207 {
1208 	return queue_zone_write_granularity(bdev_get_queue(bdev));
1209 }
1210 
1211 int bdev_alignment_offset(struct block_device *bdev);
1212 unsigned int bdev_discard_alignment(struct block_device *bdev);
1213 
1214 static inline unsigned int bdev_max_discard_sectors(struct block_device *bdev)
1215 {
1216 	return bdev_get_queue(bdev)->limits.max_discard_sectors;
1217 }
1218 
1219 static inline unsigned int bdev_discard_granularity(struct block_device *bdev)
1220 {
1221 	return bdev_get_queue(bdev)->limits.discard_granularity;
1222 }
1223 
1224 static inline unsigned int
1225 bdev_max_secure_erase_sectors(struct block_device *bdev)
1226 {
1227 	return bdev_get_queue(bdev)->limits.max_secure_erase_sectors;
1228 }
1229 
1230 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1231 {
1232 	struct request_queue *q = bdev_get_queue(bdev);
1233 
1234 	if (q)
1235 		return q->limits.max_write_zeroes_sectors;
1236 
1237 	return 0;
1238 }
1239 
1240 static inline bool bdev_nonrot(struct block_device *bdev)
1241 {
1242 	return blk_queue_nonrot(bdev_get_queue(bdev));
1243 }
1244 
1245 static inline bool bdev_synchronous(struct block_device *bdev)
1246 {
1247 	return test_bit(QUEUE_FLAG_SYNCHRONOUS,
1248 			&bdev_get_queue(bdev)->queue_flags);
1249 }
1250 
1251 static inline bool bdev_stable_writes(struct block_device *bdev)
1252 {
1253 	return test_bit(QUEUE_FLAG_STABLE_WRITES,
1254 			&bdev_get_queue(bdev)->queue_flags);
1255 }
1256 
1257 static inline bool bdev_write_cache(struct block_device *bdev)
1258 {
1259 	return test_bit(QUEUE_FLAG_WC, &bdev_get_queue(bdev)->queue_flags);
1260 }
1261 
1262 static inline bool bdev_fua(struct block_device *bdev)
1263 {
1264 	return test_bit(QUEUE_FLAG_FUA, &bdev_get_queue(bdev)->queue_flags);
1265 }
1266 
1267 static inline bool bdev_nowait(struct block_device *bdev)
1268 {
1269 	return test_bit(QUEUE_FLAG_NOWAIT, &bdev_get_queue(bdev)->queue_flags);
1270 }
1271 
1272 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1273 {
1274 	return blk_queue_zoned_model(bdev_get_queue(bdev));
1275 }
1276 
1277 static inline bool bdev_is_zoned(struct block_device *bdev)
1278 {
1279 	return blk_queue_is_zoned(bdev_get_queue(bdev));
1280 }
1281 
1282 static inline unsigned int bdev_zone_no(struct block_device *bdev, sector_t sec)
1283 {
1284 	return disk_zone_no(bdev->bd_disk, sec);
1285 }
1286 
1287 static inline bool bdev_op_is_zoned_write(struct block_device *bdev,
1288 					  blk_opf_t op)
1289 {
1290 	if (!bdev_is_zoned(bdev))
1291 		return false;
1292 
1293 	return op == REQ_OP_WRITE || op == REQ_OP_WRITE_ZEROES;
1294 }
1295 
1296 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1297 {
1298 	struct request_queue *q = bdev_get_queue(bdev);
1299 
1300 	if (!blk_queue_is_zoned(q))
1301 		return 0;
1302 	return q->limits.chunk_sectors;
1303 }
1304 
1305 static inline sector_t bdev_offset_from_zone_start(struct block_device *bdev,
1306 						   sector_t sector)
1307 {
1308 	return sector & (bdev_zone_sectors(bdev) - 1);
1309 }
1310 
1311 static inline bool bdev_is_zone_start(struct block_device *bdev,
1312 				      sector_t sector)
1313 {
1314 	return bdev_offset_from_zone_start(bdev, sector) == 0;
1315 }
1316 
1317 static inline int queue_dma_alignment(const struct request_queue *q)
1318 {
1319 	return q ? q->limits.dma_alignment : 511;
1320 }
1321 
1322 static inline unsigned int bdev_dma_alignment(struct block_device *bdev)
1323 {
1324 	return queue_dma_alignment(bdev_get_queue(bdev));
1325 }
1326 
1327 static inline bool bdev_iter_is_aligned(struct block_device *bdev,
1328 					struct iov_iter *iter)
1329 {
1330 	return iov_iter_is_aligned(iter, bdev_dma_alignment(bdev),
1331 				   bdev_logical_block_size(bdev) - 1);
1332 }
1333 
1334 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1335 				 unsigned int len)
1336 {
1337 	unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1338 	return !(addr & alignment) && !(len & alignment);
1339 }
1340 
1341 /* assumes size > 256 */
1342 static inline unsigned int blksize_bits(unsigned int size)
1343 {
1344 	return order_base_2(size >> SECTOR_SHIFT) + SECTOR_SHIFT;
1345 }
1346 
1347 static inline unsigned int block_size(struct block_device *bdev)
1348 {
1349 	return 1 << bdev->bd_inode->i_blkbits;
1350 }
1351 
1352 int kblockd_schedule_work(struct work_struct *work);
1353 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1354 
1355 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1356 	MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1357 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1358 	MODULE_ALIAS("block-major-" __stringify(major) "-*")
1359 
1360 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1361 
1362 bool blk_crypto_register(struct blk_crypto_profile *profile,
1363 			 struct request_queue *q);
1364 
1365 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1366 
1367 static inline bool blk_crypto_register(struct blk_crypto_profile *profile,
1368 				       struct request_queue *q)
1369 {
1370 	return true;
1371 }
1372 
1373 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1374 
1375 enum blk_unique_id {
1376 	/* these match the Designator Types specified in SPC */
1377 	BLK_UID_T10	= 1,
1378 	BLK_UID_EUI64	= 2,
1379 	BLK_UID_NAA	= 3,
1380 };
1381 
1382 #define NFL4_UFLG_MASK			0x0000003F
1383 
1384 struct block_device_operations {
1385 	void (*submit_bio)(struct bio *bio);
1386 	int (*poll_bio)(struct bio *bio, struct io_comp_batch *iob,
1387 			unsigned int flags);
1388 	int (*open) (struct block_device *, fmode_t);
1389 	void (*release) (struct gendisk *, fmode_t);
1390 	int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1391 	int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1392 	unsigned int (*check_events) (struct gendisk *disk,
1393 				      unsigned int clearing);
1394 	void (*unlock_native_capacity) (struct gendisk *);
1395 	int (*getgeo)(struct block_device *, struct hd_geometry *);
1396 	int (*set_read_only)(struct block_device *bdev, bool ro);
1397 	void (*free_disk)(struct gendisk *disk);
1398 	/* this callback is with swap_lock and sometimes page table lock held */
1399 	void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1400 	int (*report_zones)(struct gendisk *, sector_t sector,
1401 			unsigned int nr_zones, report_zones_cb cb, void *data);
1402 	char *(*devnode)(struct gendisk *disk, umode_t *mode);
1403 	/* returns the length of the identifier or a negative errno: */
1404 	int (*get_unique_id)(struct gendisk *disk, u8 id[16],
1405 			enum blk_unique_id id_type);
1406 	struct module *owner;
1407 	const struct pr_ops *pr_ops;
1408 
1409 	/*
1410 	 * Special callback for probing GPT entry at a given sector.
1411 	 * Needed by Android devices, used by GPT scanner and MMC blk
1412 	 * driver.
1413 	 */
1414 	int (*alternative_gpt_sector)(struct gendisk *disk, sector_t *sector);
1415 };
1416 
1417 #ifdef CONFIG_COMPAT
1418 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1419 				      unsigned int, unsigned long);
1420 #else
1421 #define blkdev_compat_ptr_ioctl NULL
1422 #endif
1423 
1424 static inline void blk_wake_io_task(struct task_struct *waiter)
1425 {
1426 	/*
1427 	 * If we're polling, the task itself is doing the completions. For
1428 	 * that case, we don't need to signal a wakeup, it's enough to just
1429 	 * mark us as RUNNING.
1430 	 */
1431 	if (waiter == current)
1432 		__set_current_state(TASK_RUNNING);
1433 	else
1434 		wake_up_process(waiter);
1435 }
1436 
1437 unsigned long bdev_start_io_acct(struct block_device *bdev, enum req_op op,
1438 				 unsigned long start_time);
1439 void bdev_end_io_acct(struct block_device *bdev, enum req_op op,
1440 		      unsigned int sectors, unsigned long start_time);
1441 
1442 unsigned long bio_start_io_acct(struct bio *bio);
1443 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1444 		struct block_device *orig_bdev);
1445 
1446 /**
1447  * bio_end_io_acct - end I/O accounting for bio based drivers
1448  * @bio:	bio to end account for
1449  * @start_time:	start time returned by bio_start_io_acct()
1450  */
1451 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1452 {
1453 	return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1454 }
1455 
1456 int bdev_read_only(struct block_device *bdev);
1457 int set_blocksize(struct block_device *bdev, int size);
1458 
1459 int lookup_bdev(const char *pathname, dev_t *dev);
1460 
1461 void blkdev_show(struct seq_file *seqf, off_t offset);
1462 
1463 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
1464 #define BDEVT_SIZE	10	/* Largest string for MAJ:MIN for blkdev */
1465 #ifdef CONFIG_BLOCK
1466 #define BLKDEV_MAJOR_MAX	512
1467 #else
1468 #define BLKDEV_MAJOR_MAX	0
1469 #endif
1470 
1471 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1472 		void *holder);
1473 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1474 int bd_prepare_to_claim(struct block_device *bdev, void *holder);
1475 void bd_abort_claiming(struct block_device *bdev, void *holder);
1476 void blkdev_put(struct block_device *bdev, fmode_t mode);
1477 
1478 /* just for blk-cgroup, don't use elsewhere */
1479 struct block_device *blkdev_get_no_open(dev_t dev);
1480 void blkdev_put_no_open(struct block_device *bdev);
1481 
1482 struct block_device *bdev_alloc(struct gendisk *disk, u8 partno);
1483 void bdev_add(struct block_device *bdev, dev_t dev);
1484 struct block_device *I_BDEV(struct inode *inode);
1485 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
1486 		loff_t lend);
1487 
1488 #ifdef CONFIG_BLOCK
1489 void invalidate_bdev(struct block_device *bdev);
1490 int sync_blockdev(struct block_device *bdev);
1491 int sync_blockdev_range(struct block_device *bdev, loff_t lstart, loff_t lend);
1492 int sync_blockdev_nowait(struct block_device *bdev);
1493 void sync_bdevs(bool wait);
1494 void bdev_statx_dioalign(struct inode *inode, struct kstat *stat);
1495 void printk_all_partitions(void);
1496 #else
1497 static inline void invalidate_bdev(struct block_device *bdev)
1498 {
1499 }
1500 static inline int sync_blockdev(struct block_device *bdev)
1501 {
1502 	return 0;
1503 }
1504 static inline int sync_blockdev_nowait(struct block_device *bdev)
1505 {
1506 	return 0;
1507 }
1508 static inline void sync_bdevs(bool wait)
1509 {
1510 }
1511 static inline void bdev_statx_dioalign(struct inode *inode, struct kstat *stat)
1512 {
1513 }
1514 static inline void printk_all_partitions(void)
1515 {
1516 }
1517 #endif /* CONFIG_BLOCK */
1518 
1519 int fsync_bdev(struct block_device *bdev);
1520 
1521 int freeze_bdev(struct block_device *bdev);
1522 int thaw_bdev(struct block_device *bdev);
1523 
1524 struct io_comp_batch {
1525 	struct request *req_list;
1526 	bool need_ts;
1527 	void (*complete)(struct io_comp_batch *);
1528 };
1529 
1530 #define DEFINE_IO_COMP_BATCH(name)	struct io_comp_batch name = { }
1531 
1532 #endif /* _LINUX_BLKDEV_H */
1533