xref: /linux-6.15/include/linux/blkdev.h (revision 7af6fbdd)
1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef _LINUX_BLKDEV_H
3 #define _LINUX_BLKDEV_H
4 
5 #include <linux/sched.h>
6 #include <linux/sched/clock.h>
7 #include <linux/major.h>
8 #include <linux/genhd.h>
9 #include <linux/list.h>
10 #include <linux/llist.h>
11 #include <linux/timer.h>
12 #include <linux/workqueue.h>
13 #include <linux/pagemap.h>
14 #include <linux/backing-dev-defs.h>
15 #include <linux/wait.h>
16 #include <linux/mempool.h>
17 #include <linux/pfn.h>
18 #include <linux/bio.h>
19 #include <linux/stringify.h>
20 #include <linux/gfp.h>
21 #include <linux/bsg.h>
22 #include <linux/smp.h>
23 #include <linux/rcupdate.h>
24 #include <linux/percpu-refcount.h>
25 #include <linux/scatterlist.h>
26 #include <linux/blkzoned.h>
27 #include <linux/pm.h>
28 
29 struct module;
30 struct scsi_ioctl_command;
31 
32 struct request_queue;
33 struct elevator_queue;
34 struct blk_trace;
35 struct request;
36 struct sg_io_hdr;
37 struct bsg_job;
38 struct blkcg_gq;
39 struct blk_flush_queue;
40 struct pr_ops;
41 struct rq_qos;
42 struct blk_queue_stats;
43 struct blk_stat_callback;
44 struct blk_keyslot_manager;
45 
46 #define BLKDEV_MIN_RQ	4
47 #define BLKDEV_MAX_RQ	128	/* Default maximum */
48 
49 /* Must be consistent with blk_mq_poll_stats_bkt() */
50 #define BLK_MQ_POLL_STATS_BKTS 16
51 
52 /* Doing classic polling */
53 #define BLK_MQ_POLL_CLASSIC -1
54 
55 /*
56  * Maximum number of blkcg policies allowed to be registered concurrently.
57  * Defined here to simplify include dependency.
58  */
59 #define BLKCG_MAX_POLS		5
60 
61 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
62 
63 /*
64  * request flags */
65 typedef __u32 __bitwise req_flags_t;
66 
67 /* elevator knows about this request */
68 #define RQF_SORTED		((__force req_flags_t)(1 << 0))
69 /* drive already may have started this one */
70 #define RQF_STARTED		((__force req_flags_t)(1 << 1))
71 /* may not be passed by ioscheduler */
72 #define RQF_SOFTBARRIER		((__force req_flags_t)(1 << 3))
73 /* request for flush sequence */
74 #define RQF_FLUSH_SEQ		((__force req_flags_t)(1 << 4))
75 /* merge of different types, fail separately */
76 #define RQF_MIXED_MERGE		((__force req_flags_t)(1 << 5))
77 /* track inflight for MQ */
78 #define RQF_MQ_INFLIGHT		((__force req_flags_t)(1 << 6))
79 /* don't call prep for this one */
80 #define RQF_DONTPREP		((__force req_flags_t)(1 << 7))
81 /* set for "ide_preempt" requests and also for requests for which the SCSI
82    "quiesce" state must be ignored. */
83 #define RQF_PREEMPT		((__force req_flags_t)(1 << 8))
84 /* vaguely specified driver internal error.  Ignored by the block layer */
85 #define RQF_FAILED		((__force req_flags_t)(1 << 10))
86 /* don't warn about errors */
87 #define RQF_QUIET		((__force req_flags_t)(1 << 11))
88 /* elevator private data attached */
89 #define RQF_ELVPRIV		((__force req_flags_t)(1 << 12))
90 /* account into disk and partition IO statistics */
91 #define RQF_IO_STAT		((__force req_flags_t)(1 << 13))
92 /* request came from our alloc pool */
93 #define RQF_ALLOCED		((__force req_flags_t)(1 << 14))
94 /* runtime pm request */
95 #define RQF_PM			((__force req_flags_t)(1 << 15))
96 /* on IO scheduler merge hash */
97 #define RQF_HASHED		((__force req_flags_t)(1 << 16))
98 /* track IO completion time */
99 #define RQF_STATS		((__force req_flags_t)(1 << 17))
100 /* Look at ->special_vec for the actual data payload instead of the
101    bio chain. */
102 #define RQF_SPECIAL_PAYLOAD	((__force req_flags_t)(1 << 18))
103 /* The per-zone write lock is held for this request */
104 #define RQF_ZONE_WRITE_LOCKED	((__force req_flags_t)(1 << 19))
105 /* already slept for hybrid poll */
106 #define RQF_MQ_POLL_SLEPT	((__force req_flags_t)(1 << 20))
107 /* ->timeout has been called, don't expire again */
108 #define RQF_TIMED_OUT		((__force req_flags_t)(1 << 21))
109 
110 /* flags that prevent us from merging requests: */
111 #define RQF_NOMERGE_FLAGS \
112 	(RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
113 
114 /*
115  * Request state for blk-mq.
116  */
117 enum mq_rq_state {
118 	MQ_RQ_IDLE		= 0,
119 	MQ_RQ_IN_FLIGHT		= 1,
120 	MQ_RQ_COMPLETE		= 2,
121 };
122 
123 /*
124  * Try to put the fields that are referenced together in the same cacheline.
125  *
126  * If you modify this structure, make sure to update blk_rq_init() and
127  * especially blk_mq_rq_ctx_init() to take care of the added fields.
128  */
129 struct request {
130 	struct request_queue *q;
131 	struct blk_mq_ctx *mq_ctx;
132 	struct blk_mq_hw_ctx *mq_hctx;
133 
134 	unsigned int cmd_flags;		/* op and common flags */
135 	req_flags_t rq_flags;
136 
137 	int tag;
138 	int internal_tag;
139 
140 	/* the following two fields are internal, NEVER access directly */
141 	unsigned int __data_len;	/* total data len */
142 	sector_t __sector;		/* sector cursor */
143 
144 	struct bio *bio;
145 	struct bio *biotail;
146 
147 	struct list_head queuelist;
148 
149 	/*
150 	 * The hash is used inside the scheduler, and killed once the
151 	 * request reaches the dispatch list. The ipi_list is only used
152 	 * to queue the request for softirq completion, which is long
153 	 * after the request has been unhashed (and even removed from
154 	 * the dispatch list).
155 	 */
156 	union {
157 		struct hlist_node hash;	/* merge hash */
158 		struct list_head ipi_list;
159 	};
160 
161 	/*
162 	 * The rb_node is only used inside the io scheduler, requests
163 	 * are pruned when moved to the dispatch queue. So let the
164 	 * completion_data share space with the rb_node.
165 	 */
166 	union {
167 		struct rb_node rb_node;	/* sort/lookup */
168 		struct bio_vec special_vec;
169 		void *completion_data;
170 		int error_count; /* for legacy drivers, don't use */
171 	};
172 
173 	/*
174 	 * Three pointers are available for the IO schedulers, if they need
175 	 * more they have to dynamically allocate it.  Flush requests are
176 	 * never put on the IO scheduler. So let the flush fields share
177 	 * space with the elevator data.
178 	 */
179 	union {
180 		struct {
181 			struct io_cq		*icq;
182 			void			*priv[2];
183 		} elv;
184 
185 		struct {
186 			unsigned int		seq;
187 			struct list_head	list;
188 			rq_end_io_fn		*saved_end_io;
189 		} flush;
190 	};
191 
192 	struct gendisk *rq_disk;
193 	struct hd_struct *part;
194 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
195 	/* Time that the first bio started allocating this request. */
196 	u64 alloc_time_ns;
197 #endif
198 	/* Time that this request was allocated for this IO. */
199 	u64 start_time_ns;
200 	/* Time that I/O was submitted to the device. */
201 	u64 io_start_time_ns;
202 
203 #ifdef CONFIG_BLK_WBT
204 	unsigned short wbt_flags;
205 #endif
206 	/*
207 	 * rq sectors used for blk stats. It has the same value
208 	 * with blk_rq_sectors(rq), except that it never be zeroed
209 	 * by completion.
210 	 */
211 	unsigned short stats_sectors;
212 
213 	/*
214 	 * Number of scatter-gather DMA addr+len pairs after
215 	 * physical address coalescing is performed.
216 	 */
217 	unsigned short nr_phys_segments;
218 
219 #if defined(CONFIG_BLK_DEV_INTEGRITY)
220 	unsigned short nr_integrity_segments;
221 #endif
222 
223 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
224 	struct bio_crypt_ctx *crypt_ctx;
225 	struct blk_ksm_keyslot *crypt_keyslot;
226 #endif
227 
228 	unsigned short write_hint;
229 	unsigned short ioprio;
230 
231 	enum mq_rq_state state;
232 	refcount_t ref;
233 
234 	unsigned int timeout;
235 	unsigned long deadline;
236 
237 	union {
238 		struct __call_single_data csd;
239 		u64 fifo_time;
240 	};
241 
242 	/*
243 	 * completion callback.
244 	 */
245 	rq_end_io_fn *end_io;
246 	void *end_io_data;
247 };
248 
249 static inline bool blk_op_is_scsi(unsigned int op)
250 {
251 	return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
252 }
253 
254 static inline bool blk_op_is_private(unsigned int op)
255 {
256 	return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
257 }
258 
259 static inline bool blk_rq_is_scsi(struct request *rq)
260 {
261 	return blk_op_is_scsi(req_op(rq));
262 }
263 
264 static inline bool blk_rq_is_private(struct request *rq)
265 {
266 	return blk_op_is_private(req_op(rq));
267 }
268 
269 static inline bool blk_rq_is_passthrough(struct request *rq)
270 {
271 	return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
272 }
273 
274 static inline bool bio_is_passthrough(struct bio *bio)
275 {
276 	unsigned op = bio_op(bio);
277 
278 	return blk_op_is_scsi(op) || blk_op_is_private(op);
279 }
280 
281 static inline unsigned short req_get_ioprio(struct request *req)
282 {
283 	return req->ioprio;
284 }
285 
286 #include <linux/elevator.h>
287 
288 struct blk_queue_ctx;
289 
290 struct bio_vec;
291 
292 enum blk_eh_timer_return {
293 	BLK_EH_DONE,		/* drivers has completed the command */
294 	BLK_EH_RESET_TIMER,	/* reset timer and try again */
295 };
296 
297 enum blk_queue_state {
298 	Queue_down,
299 	Queue_up,
300 };
301 
302 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
303 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
304 
305 #define BLK_SCSI_MAX_CMDS	(256)
306 #define BLK_SCSI_CMD_PER_LONG	(BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
307 
308 /*
309  * Zoned block device models (zoned limit).
310  *
311  * Note: This needs to be ordered from the least to the most severe
312  * restrictions for the inheritance in blk_stack_limits() to work.
313  */
314 enum blk_zoned_model {
315 	BLK_ZONED_NONE = 0,	/* Regular block device */
316 	BLK_ZONED_HA,		/* Host-aware zoned block device */
317 	BLK_ZONED_HM,		/* Host-managed zoned block device */
318 };
319 
320 struct queue_limits {
321 	unsigned long		bounce_pfn;
322 	unsigned long		seg_boundary_mask;
323 	unsigned long		virt_boundary_mask;
324 
325 	unsigned int		max_hw_sectors;
326 	unsigned int		max_dev_sectors;
327 	unsigned int		chunk_sectors;
328 	unsigned int		max_sectors;
329 	unsigned int		max_segment_size;
330 	unsigned int		physical_block_size;
331 	unsigned int		logical_block_size;
332 	unsigned int		alignment_offset;
333 	unsigned int		io_min;
334 	unsigned int		io_opt;
335 	unsigned int		max_discard_sectors;
336 	unsigned int		max_hw_discard_sectors;
337 	unsigned int		max_write_same_sectors;
338 	unsigned int		max_write_zeroes_sectors;
339 	unsigned int		max_zone_append_sectors;
340 	unsigned int		discard_granularity;
341 	unsigned int		discard_alignment;
342 
343 	unsigned short		max_segments;
344 	unsigned short		max_integrity_segments;
345 	unsigned short		max_discard_segments;
346 
347 	unsigned char		misaligned;
348 	unsigned char		discard_misaligned;
349 	unsigned char		raid_partial_stripes_expensive;
350 	enum blk_zoned_model	zoned;
351 };
352 
353 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
354 			       void *data);
355 
356 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
357 
358 #ifdef CONFIG_BLK_DEV_ZONED
359 
360 #define BLK_ALL_ZONES  ((unsigned int)-1)
361 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
362 			unsigned int nr_zones, report_zones_cb cb, void *data);
363 unsigned int blkdev_nr_zones(struct gendisk *disk);
364 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
365 			    sector_t sectors, sector_t nr_sectors,
366 			    gfp_t gfp_mask);
367 int blk_revalidate_disk_zones(struct gendisk *disk,
368 			      void (*update_driver_data)(struct gendisk *disk));
369 
370 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
371 				     unsigned int cmd, unsigned long arg);
372 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
373 				  unsigned int cmd, unsigned long arg);
374 
375 #else /* CONFIG_BLK_DEV_ZONED */
376 
377 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
378 {
379 	return 0;
380 }
381 
382 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
383 					    fmode_t mode, unsigned int cmd,
384 					    unsigned long arg)
385 {
386 	return -ENOTTY;
387 }
388 
389 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
390 					 fmode_t mode, unsigned int cmd,
391 					 unsigned long arg)
392 {
393 	return -ENOTTY;
394 }
395 
396 #endif /* CONFIG_BLK_DEV_ZONED */
397 
398 struct request_queue {
399 	struct request		*last_merge;
400 	struct elevator_queue	*elevator;
401 
402 	struct percpu_ref	q_usage_counter;
403 
404 	struct blk_queue_stats	*stats;
405 	struct rq_qos		*rq_qos;
406 
407 	const struct blk_mq_ops	*mq_ops;
408 
409 	/* sw queues */
410 	struct blk_mq_ctx __percpu	*queue_ctx;
411 
412 	unsigned int		queue_depth;
413 
414 	/* hw dispatch queues */
415 	struct blk_mq_hw_ctx	**queue_hw_ctx;
416 	unsigned int		nr_hw_queues;
417 
418 	struct backing_dev_info	*backing_dev_info;
419 
420 	/*
421 	 * The queue owner gets to use this for whatever they like.
422 	 * ll_rw_blk doesn't touch it.
423 	 */
424 	void			*queuedata;
425 
426 	/*
427 	 * various queue flags, see QUEUE_* below
428 	 */
429 	unsigned long		queue_flags;
430 	/*
431 	 * Number of contexts that have called blk_set_pm_only(). If this
432 	 * counter is above zero then only RQF_PM and RQF_PREEMPT requests are
433 	 * processed.
434 	 */
435 	atomic_t		pm_only;
436 
437 	/*
438 	 * ida allocated id for this queue.  Used to index queues from
439 	 * ioctx.
440 	 */
441 	int			id;
442 
443 	/*
444 	 * queue needs bounce pages for pages above this limit
445 	 */
446 	gfp_t			bounce_gfp;
447 
448 	spinlock_t		queue_lock;
449 
450 	/*
451 	 * queue kobject
452 	 */
453 	struct kobject kobj;
454 
455 	/*
456 	 * mq queue kobject
457 	 */
458 	struct kobject *mq_kobj;
459 
460 #ifdef  CONFIG_BLK_DEV_INTEGRITY
461 	struct blk_integrity integrity;
462 #endif	/* CONFIG_BLK_DEV_INTEGRITY */
463 
464 #ifdef CONFIG_PM
465 	struct device		*dev;
466 	enum rpm_status		rpm_status;
467 	unsigned int		nr_pending;
468 #endif
469 
470 	/*
471 	 * queue settings
472 	 */
473 	unsigned long		nr_requests;	/* Max # of requests */
474 
475 	unsigned int		dma_pad_mask;
476 	unsigned int		dma_alignment;
477 
478 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
479 	/* Inline crypto capabilities */
480 	struct blk_keyslot_manager *ksm;
481 #endif
482 
483 	unsigned int		rq_timeout;
484 	int			poll_nsec;
485 
486 	struct blk_stat_callback	*poll_cb;
487 	struct blk_rq_stat	poll_stat[BLK_MQ_POLL_STATS_BKTS];
488 
489 	struct timer_list	timeout;
490 	struct work_struct	timeout_work;
491 
492 	atomic_t		nr_active_requests_shared_sbitmap;
493 
494 	struct list_head	icq_list;
495 #ifdef CONFIG_BLK_CGROUP
496 	DECLARE_BITMAP		(blkcg_pols, BLKCG_MAX_POLS);
497 	struct blkcg_gq		*root_blkg;
498 	struct list_head	blkg_list;
499 #endif
500 
501 	struct queue_limits	limits;
502 
503 	unsigned int		required_elevator_features;
504 
505 #ifdef CONFIG_BLK_DEV_ZONED
506 	/*
507 	 * Zoned block device information for request dispatch control.
508 	 * nr_zones is the total number of zones of the device. This is always
509 	 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
510 	 * bits which indicates if a zone is conventional (bit set) or
511 	 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
512 	 * bits which indicates if a zone is write locked, that is, if a write
513 	 * request targeting the zone was dispatched. All three fields are
514 	 * initialized by the low level device driver (e.g. scsi/sd.c).
515 	 * Stacking drivers (device mappers) may or may not initialize
516 	 * these fields.
517 	 *
518 	 * Reads of this information must be protected with blk_queue_enter() /
519 	 * blk_queue_exit(). Modifying this information is only allowed while
520 	 * no requests are being processed. See also blk_mq_freeze_queue() and
521 	 * blk_mq_unfreeze_queue().
522 	 */
523 	unsigned int		nr_zones;
524 	unsigned long		*conv_zones_bitmap;
525 	unsigned long		*seq_zones_wlock;
526 	unsigned int		max_open_zones;
527 	unsigned int		max_active_zones;
528 #endif /* CONFIG_BLK_DEV_ZONED */
529 
530 	/*
531 	 * sg stuff
532 	 */
533 	unsigned int		sg_timeout;
534 	unsigned int		sg_reserved_size;
535 	int			node;
536 	struct mutex		debugfs_mutex;
537 #ifdef CONFIG_BLK_DEV_IO_TRACE
538 	struct blk_trace __rcu	*blk_trace;
539 #endif
540 	/*
541 	 * for flush operations
542 	 */
543 	struct blk_flush_queue	*fq;
544 
545 	struct list_head	requeue_list;
546 	spinlock_t		requeue_lock;
547 	struct delayed_work	requeue_work;
548 
549 	struct mutex		sysfs_lock;
550 	struct mutex		sysfs_dir_lock;
551 
552 	/*
553 	 * for reusing dead hctx instance in case of updating
554 	 * nr_hw_queues
555 	 */
556 	struct list_head	unused_hctx_list;
557 	spinlock_t		unused_hctx_lock;
558 
559 	int			mq_freeze_depth;
560 
561 #if defined(CONFIG_BLK_DEV_BSG)
562 	struct bsg_class_device bsg_dev;
563 #endif
564 
565 #ifdef CONFIG_BLK_DEV_THROTTLING
566 	/* Throttle data */
567 	struct throtl_data *td;
568 #endif
569 	struct rcu_head		rcu_head;
570 	wait_queue_head_t	mq_freeze_wq;
571 	/*
572 	 * Protect concurrent access to q_usage_counter by
573 	 * percpu_ref_kill() and percpu_ref_reinit().
574 	 */
575 	struct mutex		mq_freeze_lock;
576 
577 	struct blk_mq_tag_set	*tag_set;
578 	struct list_head	tag_set_list;
579 	struct bio_set		bio_split;
580 
581 	struct dentry		*debugfs_dir;
582 
583 #ifdef CONFIG_BLK_DEBUG_FS
584 	struct dentry		*sched_debugfs_dir;
585 	struct dentry		*rqos_debugfs_dir;
586 #endif
587 
588 	bool			mq_sysfs_init_done;
589 
590 	size_t			cmd_size;
591 
592 #define BLK_MAX_WRITE_HINTS	5
593 	u64			write_hints[BLK_MAX_WRITE_HINTS];
594 };
595 
596 /* Keep blk_queue_flag_name[] in sync with the definitions below */
597 #define QUEUE_FLAG_STOPPED	0	/* queue is stopped */
598 #define QUEUE_FLAG_DYING	1	/* queue being torn down */
599 #define QUEUE_FLAG_NOMERGES     3	/* disable merge attempts */
600 #define QUEUE_FLAG_SAME_COMP	4	/* complete on same CPU-group */
601 #define QUEUE_FLAG_FAIL_IO	5	/* fake timeout */
602 #define QUEUE_FLAG_NONROT	6	/* non-rotational device (SSD) */
603 #define QUEUE_FLAG_VIRT		QUEUE_FLAG_NONROT /* paravirt device */
604 #define QUEUE_FLAG_IO_STAT	7	/* do disk/partitions IO accounting */
605 #define QUEUE_FLAG_DISCARD	8	/* supports DISCARD */
606 #define QUEUE_FLAG_NOXMERGES	9	/* No extended merges */
607 #define QUEUE_FLAG_ADD_RANDOM	10	/* Contributes to random pool */
608 #define QUEUE_FLAG_SECERASE	11	/* supports secure erase */
609 #define QUEUE_FLAG_SAME_FORCE	12	/* force complete on same CPU */
610 #define QUEUE_FLAG_DEAD		13	/* queue tear-down finished */
611 #define QUEUE_FLAG_INIT_DONE	14	/* queue is initialized */
612 #define QUEUE_FLAG_STABLE_WRITES 15	/* don't modify blks until WB is done */
613 #define QUEUE_FLAG_POLL		16	/* IO polling enabled if set */
614 #define QUEUE_FLAG_WC		17	/* Write back caching */
615 #define QUEUE_FLAG_FUA		18	/* device supports FUA writes */
616 #define QUEUE_FLAG_DAX		19	/* device supports DAX */
617 #define QUEUE_FLAG_STATS	20	/* track IO start and completion times */
618 #define QUEUE_FLAG_POLL_STATS	21	/* collecting stats for hybrid polling */
619 #define QUEUE_FLAG_REGISTERED	22	/* queue has been registered to a disk */
620 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23	/* queue supports SCSI commands */
621 #define QUEUE_FLAG_QUIESCED	24	/* queue has been quiesced */
622 #define QUEUE_FLAG_PCI_P2PDMA	25	/* device supports PCI p2p requests */
623 #define QUEUE_FLAG_ZONE_RESETALL 26	/* supports Zone Reset All */
624 #define QUEUE_FLAG_RQ_ALLOC_TIME 27	/* record rq->alloc_time_ns */
625 #define QUEUE_FLAG_HCTX_ACTIVE	28	/* at least one blk-mq hctx is active */
626 #define QUEUE_FLAG_NOWAIT       29	/* device supports NOWAIT */
627 
628 #define QUEUE_FLAG_MQ_DEFAULT	((1 << QUEUE_FLAG_IO_STAT) |		\
629 				 (1 << QUEUE_FLAG_SAME_COMP) |		\
630 				 (1 << QUEUE_FLAG_NOWAIT))
631 
632 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
633 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
634 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
635 
636 #define blk_queue_stopped(q)	test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
637 #define blk_queue_dying(q)	test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
638 #define blk_queue_dead(q)	test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
639 #define blk_queue_init_done(q)	test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
640 #define blk_queue_nomerges(q)	test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
641 #define blk_queue_noxmerges(q)	\
642 	test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
643 #define blk_queue_nonrot(q)	test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
644 #define blk_queue_stable_writes(q) \
645 	test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
646 #define blk_queue_io_stat(q)	test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
647 #define blk_queue_add_random(q)	test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
648 #define blk_queue_discard(q)	test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
649 #define blk_queue_zone_resetall(q)	\
650 	test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
651 #define blk_queue_secure_erase(q) \
652 	(test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
653 #define blk_queue_dax(q)	test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
654 #define blk_queue_scsi_passthrough(q)	\
655 	test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
656 #define blk_queue_pci_p2pdma(q)	\
657 	test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
658 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
659 #define blk_queue_rq_alloc_time(q)	\
660 	test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
661 #else
662 #define blk_queue_rq_alloc_time(q)	false
663 #endif
664 
665 #define blk_noretry_request(rq) \
666 	((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
667 			     REQ_FAILFAST_DRIVER))
668 #define blk_queue_quiesced(q)	test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
669 #define blk_queue_pm_only(q)	atomic_read(&(q)->pm_only)
670 #define blk_queue_fua(q)	test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
671 #define blk_queue_registered(q)	test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
672 #define blk_queue_nowait(q)	test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
673 
674 extern void blk_set_pm_only(struct request_queue *q);
675 extern void blk_clear_pm_only(struct request_queue *q);
676 
677 static inline bool blk_account_rq(struct request *rq)
678 {
679 	return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
680 }
681 
682 #define list_entry_rq(ptr)	list_entry((ptr), struct request, queuelist)
683 
684 #define rq_data_dir(rq)		(op_is_write(req_op(rq)) ? WRITE : READ)
685 
686 #define rq_dma_dir(rq) \
687 	(op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
688 
689 #define dma_map_bvec(dev, bv, dir, attrs) \
690 	dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
691 	(dir), (attrs))
692 
693 static inline bool queue_is_mq(struct request_queue *q)
694 {
695 	return q->mq_ops;
696 }
697 
698 static inline enum blk_zoned_model
699 blk_queue_zoned_model(struct request_queue *q)
700 {
701 	if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
702 		return q->limits.zoned;
703 	return BLK_ZONED_NONE;
704 }
705 
706 static inline bool blk_queue_is_zoned(struct request_queue *q)
707 {
708 	switch (blk_queue_zoned_model(q)) {
709 	case BLK_ZONED_HA:
710 	case BLK_ZONED_HM:
711 		return true;
712 	default:
713 		return false;
714 	}
715 }
716 
717 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
718 {
719 	return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
720 }
721 
722 #ifdef CONFIG_BLK_DEV_ZONED
723 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
724 {
725 	return blk_queue_is_zoned(q) ? q->nr_zones : 0;
726 }
727 
728 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
729 					     sector_t sector)
730 {
731 	if (!blk_queue_is_zoned(q))
732 		return 0;
733 	return sector >> ilog2(q->limits.chunk_sectors);
734 }
735 
736 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
737 					 sector_t sector)
738 {
739 	if (!blk_queue_is_zoned(q))
740 		return false;
741 	if (!q->conv_zones_bitmap)
742 		return true;
743 	return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
744 }
745 
746 static inline void blk_queue_max_open_zones(struct request_queue *q,
747 		unsigned int max_open_zones)
748 {
749 	q->max_open_zones = max_open_zones;
750 }
751 
752 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
753 {
754 	return q->max_open_zones;
755 }
756 
757 static inline void blk_queue_max_active_zones(struct request_queue *q,
758 		unsigned int max_active_zones)
759 {
760 	q->max_active_zones = max_active_zones;
761 }
762 
763 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
764 {
765 	return q->max_active_zones;
766 }
767 #else /* CONFIG_BLK_DEV_ZONED */
768 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
769 {
770 	return 0;
771 }
772 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
773 					 sector_t sector)
774 {
775 	return false;
776 }
777 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
778 					     sector_t sector)
779 {
780 	return 0;
781 }
782 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
783 {
784 	return 0;
785 }
786 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
787 {
788 	return 0;
789 }
790 #endif /* CONFIG_BLK_DEV_ZONED */
791 
792 static inline bool rq_is_sync(struct request *rq)
793 {
794 	return op_is_sync(rq->cmd_flags);
795 }
796 
797 static inline bool rq_mergeable(struct request *rq)
798 {
799 	if (blk_rq_is_passthrough(rq))
800 		return false;
801 
802 	if (req_op(rq) == REQ_OP_FLUSH)
803 		return false;
804 
805 	if (req_op(rq) == REQ_OP_WRITE_ZEROES)
806 		return false;
807 
808 	if (req_op(rq) == REQ_OP_ZONE_APPEND)
809 		return false;
810 
811 	if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
812 		return false;
813 	if (rq->rq_flags & RQF_NOMERGE_FLAGS)
814 		return false;
815 
816 	return true;
817 }
818 
819 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
820 {
821 	if (bio_page(a) == bio_page(b) &&
822 	    bio_offset(a) == bio_offset(b))
823 		return true;
824 
825 	return false;
826 }
827 
828 static inline unsigned int blk_queue_depth(struct request_queue *q)
829 {
830 	if (q->queue_depth)
831 		return q->queue_depth;
832 
833 	return q->nr_requests;
834 }
835 
836 extern unsigned long blk_max_low_pfn, blk_max_pfn;
837 
838 /*
839  * standard bounce addresses:
840  *
841  * BLK_BOUNCE_HIGH	: bounce all highmem pages
842  * BLK_BOUNCE_ANY	: don't bounce anything
843  * BLK_BOUNCE_ISA	: bounce pages above ISA DMA boundary
844  */
845 
846 #if BITS_PER_LONG == 32
847 #define BLK_BOUNCE_HIGH		((u64)blk_max_low_pfn << PAGE_SHIFT)
848 #else
849 #define BLK_BOUNCE_HIGH		-1ULL
850 #endif
851 #define BLK_BOUNCE_ANY		(-1ULL)
852 #define BLK_BOUNCE_ISA		(DMA_BIT_MASK(24))
853 
854 /*
855  * default timeout for SG_IO if none specified
856  */
857 #define BLK_DEFAULT_SG_TIMEOUT	(60 * HZ)
858 #define BLK_MIN_SG_TIMEOUT	(7 * HZ)
859 
860 struct rq_map_data {
861 	struct page **pages;
862 	int page_order;
863 	int nr_entries;
864 	unsigned long offset;
865 	int null_mapped;
866 	int from_user;
867 };
868 
869 struct req_iterator {
870 	struct bvec_iter iter;
871 	struct bio *bio;
872 };
873 
874 /* This should not be used directly - use rq_for_each_segment */
875 #define for_each_bio(_bio)		\
876 	for (; _bio; _bio = _bio->bi_next)
877 #define __rq_for_each_bio(_bio, rq)	\
878 	if ((rq->bio))			\
879 		for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
880 
881 #define rq_for_each_segment(bvl, _rq, _iter)			\
882 	__rq_for_each_bio(_iter.bio, _rq)			\
883 		bio_for_each_segment(bvl, _iter.bio, _iter.iter)
884 
885 #define rq_for_each_bvec(bvl, _rq, _iter)			\
886 	__rq_for_each_bio(_iter.bio, _rq)			\
887 		bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
888 
889 #define rq_iter_last(bvec, _iter)				\
890 		(_iter.bio->bi_next == NULL &&			\
891 		 bio_iter_last(bvec, _iter.iter))
892 
893 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
894 # error	"You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
895 #endif
896 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
897 extern void rq_flush_dcache_pages(struct request *rq);
898 #else
899 static inline void rq_flush_dcache_pages(struct request *rq)
900 {
901 }
902 #endif
903 
904 extern int blk_register_queue(struct gendisk *disk);
905 extern void blk_unregister_queue(struct gendisk *disk);
906 blk_qc_t submit_bio_noacct(struct bio *bio);
907 extern void blk_rq_init(struct request_queue *q, struct request *rq);
908 extern void blk_put_request(struct request *);
909 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
910 				       blk_mq_req_flags_t flags);
911 extern int blk_lld_busy(struct request_queue *q);
912 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
913 			     struct bio_set *bs, gfp_t gfp_mask,
914 			     int (*bio_ctr)(struct bio *, struct bio *, void *),
915 			     void *data);
916 extern void blk_rq_unprep_clone(struct request *rq);
917 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
918 				     struct request *rq);
919 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
920 extern void blk_queue_split(struct bio **);
921 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
922 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
923 			      unsigned int, void __user *);
924 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
925 			  unsigned int, void __user *);
926 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
927 			 struct scsi_ioctl_command __user *);
928 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
929 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
930 
931 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
932 extern void blk_queue_exit(struct request_queue *q);
933 extern void blk_sync_queue(struct request_queue *q);
934 extern int blk_rq_map_user(struct request_queue *, struct request *,
935 			   struct rq_map_data *, void __user *, unsigned long,
936 			   gfp_t);
937 extern int blk_rq_unmap_user(struct bio *);
938 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
939 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
940 			       struct rq_map_data *, const struct iov_iter *,
941 			       gfp_t);
942 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
943 			  struct request *, int);
944 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
945 				  struct request *, int, rq_end_io_fn *);
946 
947 /* Helper to convert REQ_OP_XXX to its string format XXX */
948 extern const char *blk_op_str(unsigned int op);
949 
950 int blk_status_to_errno(blk_status_t status);
951 blk_status_t errno_to_blk_status(int errno);
952 
953 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
954 
955 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
956 {
957 	return bdev->bd_disk->queue;	/* this is never NULL */
958 }
959 
960 /*
961  * The basic unit of block I/O is a sector. It is used in a number of contexts
962  * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
963  * bytes. Variables of type sector_t represent an offset or size that is a
964  * multiple of 512 bytes. Hence these two constants.
965  */
966 #ifndef SECTOR_SHIFT
967 #define SECTOR_SHIFT 9
968 #endif
969 #ifndef SECTOR_SIZE
970 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
971 #endif
972 
973 /*
974  * blk_rq_pos()			: the current sector
975  * blk_rq_bytes()		: bytes left in the entire request
976  * blk_rq_cur_bytes()		: bytes left in the current segment
977  * blk_rq_err_bytes()		: bytes left till the next error boundary
978  * blk_rq_sectors()		: sectors left in the entire request
979  * blk_rq_cur_sectors()		: sectors left in the current segment
980  * blk_rq_stats_sectors()	: sectors of the entire request used for stats
981  */
982 static inline sector_t blk_rq_pos(const struct request *rq)
983 {
984 	return rq->__sector;
985 }
986 
987 static inline unsigned int blk_rq_bytes(const struct request *rq)
988 {
989 	return rq->__data_len;
990 }
991 
992 static inline int blk_rq_cur_bytes(const struct request *rq)
993 {
994 	return rq->bio ? bio_cur_bytes(rq->bio) : 0;
995 }
996 
997 extern unsigned int blk_rq_err_bytes(const struct request *rq);
998 
999 static inline unsigned int blk_rq_sectors(const struct request *rq)
1000 {
1001 	return blk_rq_bytes(rq) >> SECTOR_SHIFT;
1002 }
1003 
1004 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
1005 {
1006 	return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
1007 }
1008 
1009 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
1010 {
1011 	return rq->stats_sectors;
1012 }
1013 
1014 #ifdef CONFIG_BLK_DEV_ZONED
1015 
1016 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
1017 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
1018 
1019 static inline unsigned int blk_rq_zone_no(struct request *rq)
1020 {
1021 	return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1022 }
1023 
1024 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1025 {
1026 	return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1027 }
1028 #endif /* CONFIG_BLK_DEV_ZONED */
1029 
1030 /*
1031  * Some commands like WRITE SAME have a payload or data transfer size which
1032  * is different from the size of the request.  Any driver that supports such
1033  * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1034  * calculate the data transfer size.
1035  */
1036 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1037 {
1038 	if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1039 		return rq->special_vec.bv_len;
1040 	return blk_rq_bytes(rq);
1041 }
1042 
1043 /*
1044  * Return the first full biovec in the request.  The caller needs to check that
1045  * there are any bvecs before calling this helper.
1046  */
1047 static inline struct bio_vec req_bvec(struct request *rq)
1048 {
1049 	if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1050 		return rq->special_vec;
1051 	return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1052 }
1053 
1054 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1055 						     int op)
1056 {
1057 	if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1058 		return min(q->limits.max_discard_sectors,
1059 			   UINT_MAX >> SECTOR_SHIFT);
1060 
1061 	if (unlikely(op == REQ_OP_WRITE_SAME))
1062 		return q->limits.max_write_same_sectors;
1063 
1064 	if (unlikely(op == REQ_OP_WRITE_ZEROES))
1065 		return q->limits.max_write_zeroes_sectors;
1066 
1067 	return q->limits.max_sectors;
1068 }
1069 
1070 /*
1071  * Return maximum size of a request at given offset. Only valid for
1072  * file system requests.
1073  */
1074 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1075 					       sector_t offset)
1076 {
1077 	unsigned int chunk_sectors = q->limits.chunk_sectors;
1078 
1079 	if (!chunk_sectors)
1080 		return q->limits.max_sectors;
1081 
1082 	if (likely(is_power_of_2(chunk_sectors)))
1083 		chunk_sectors -= offset & (chunk_sectors - 1);
1084 	else
1085 		chunk_sectors -= sector_div(offset, chunk_sectors);
1086 
1087 	return min(q->limits.max_sectors, chunk_sectors);
1088 }
1089 
1090 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1091 						  sector_t offset)
1092 {
1093 	struct request_queue *q = rq->q;
1094 
1095 	if (blk_rq_is_passthrough(rq))
1096 		return q->limits.max_hw_sectors;
1097 
1098 	if (!q->limits.chunk_sectors ||
1099 	    req_op(rq) == REQ_OP_DISCARD ||
1100 	    req_op(rq) == REQ_OP_SECURE_ERASE)
1101 		return blk_queue_get_max_sectors(q, req_op(rq));
1102 
1103 	return min(blk_max_size_offset(q, offset),
1104 			blk_queue_get_max_sectors(q, req_op(rq)));
1105 }
1106 
1107 static inline unsigned int blk_rq_count_bios(struct request *rq)
1108 {
1109 	unsigned int nr_bios = 0;
1110 	struct bio *bio;
1111 
1112 	__rq_for_each_bio(bio, rq)
1113 		nr_bios++;
1114 
1115 	return nr_bios;
1116 }
1117 
1118 void blk_steal_bios(struct bio_list *list, struct request *rq);
1119 
1120 /*
1121  * Request completion related functions.
1122  *
1123  * blk_update_request() completes given number of bytes and updates
1124  * the request without completing it.
1125  */
1126 extern bool blk_update_request(struct request *rq, blk_status_t error,
1127 			       unsigned int nr_bytes);
1128 
1129 extern void blk_abort_request(struct request *);
1130 
1131 /*
1132  * Access functions for manipulating queue properties
1133  */
1134 extern void blk_cleanup_queue(struct request_queue *);
1135 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1136 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1137 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1138 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1139 extern void blk_queue_max_discard_segments(struct request_queue *,
1140 		unsigned short);
1141 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1142 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1143 		unsigned int max_discard_sectors);
1144 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1145 		unsigned int max_write_same_sectors);
1146 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1147 		unsigned int max_write_same_sectors);
1148 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1149 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1150 		unsigned int max_zone_append_sectors);
1151 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1152 extern void blk_queue_alignment_offset(struct request_queue *q,
1153 				       unsigned int alignment);
1154 void blk_queue_update_readahead(struct request_queue *q);
1155 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1156 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1157 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1158 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1159 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1160 extern void blk_set_default_limits(struct queue_limits *lim);
1161 extern void blk_set_stacking_limits(struct queue_limits *lim);
1162 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1163 			    sector_t offset);
1164 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1165 			      sector_t offset);
1166 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1167 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1168 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1169 extern void blk_queue_dma_alignment(struct request_queue *, int);
1170 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1171 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1172 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1173 extern void blk_queue_required_elevator_features(struct request_queue *q,
1174 						 unsigned int features);
1175 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1176 					      struct device *dev);
1177 
1178 /*
1179  * Number of physical segments as sent to the device.
1180  *
1181  * Normally this is the number of discontiguous data segments sent by the
1182  * submitter.  But for data-less command like discard we might have no
1183  * actual data segments submitted, but the driver might have to add it's
1184  * own special payload.  In that case we still return 1 here so that this
1185  * special payload will be mapped.
1186  */
1187 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1188 {
1189 	if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1190 		return 1;
1191 	return rq->nr_phys_segments;
1192 }
1193 
1194 /*
1195  * Number of discard segments (or ranges) the driver needs to fill in.
1196  * Each discard bio merged into a request is counted as one segment.
1197  */
1198 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1199 {
1200 	return max_t(unsigned short, rq->nr_phys_segments, 1);
1201 }
1202 
1203 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1204 		struct scatterlist *sglist, struct scatterlist **last_sg);
1205 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1206 		struct scatterlist *sglist)
1207 {
1208 	struct scatterlist *last_sg = NULL;
1209 
1210 	return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1211 }
1212 extern void blk_dump_rq_flags(struct request *, char *);
1213 
1214 bool __must_check blk_get_queue(struct request_queue *);
1215 struct request_queue *blk_alloc_queue(int node_id);
1216 extern void blk_put_queue(struct request_queue *);
1217 extern void blk_set_queue_dying(struct request_queue *);
1218 
1219 #ifdef CONFIG_BLOCK
1220 /*
1221  * blk_plug permits building a queue of related requests by holding the I/O
1222  * fragments for a short period. This allows merging of sequential requests
1223  * into single larger request. As the requests are moved from a per-task list to
1224  * the device's request_queue in a batch, this results in improved scalability
1225  * as the lock contention for request_queue lock is reduced.
1226  *
1227  * It is ok not to disable preemption when adding the request to the plug list
1228  * or when attempting a merge, because blk_schedule_flush_list() will only flush
1229  * the plug list when the task sleeps by itself. For details, please see
1230  * schedule() where blk_schedule_flush_plug() is called.
1231  */
1232 struct blk_plug {
1233 	struct list_head mq_list; /* blk-mq requests */
1234 	struct list_head cb_list; /* md requires an unplug callback */
1235 	unsigned short rq_count;
1236 	bool multiple_queues;
1237 	bool nowait;
1238 };
1239 #define BLK_MAX_REQUEST_COUNT 16
1240 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1241 
1242 struct blk_plug_cb;
1243 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1244 struct blk_plug_cb {
1245 	struct list_head list;
1246 	blk_plug_cb_fn callback;
1247 	void *data;
1248 };
1249 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1250 					     void *data, int size);
1251 extern void blk_start_plug(struct blk_plug *);
1252 extern void blk_finish_plug(struct blk_plug *);
1253 extern void blk_flush_plug_list(struct blk_plug *, bool);
1254 
1255 static inline void blk_flush_plug(struct task_struct *tsk)
1256 {
1257 	struct blk_plug *plug = tsk->plug;
1258 
1259 	if (plug)
1260 		blk_flush_plug_list(plug, false);
1261 }
1262 
1263 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1264 {
1265 	struct blk_plug *plug = tsk->plug;
1266 
1267 	if (plug)
1268 		blk_flush_plug_list(plug, true);
1269 }
1270 
1271 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1272 {
1273 	struct blk_plug *plug = tsk->plug;
1274 
1275 	return plug &&
1276 		 (!list_empty(&plug->mq_list) ||
1277 		 !list_empty(&plug->cb_list));
1278 }
1279 
1280 int blkdev_issue_flush(struct block_device *, gfp_t);
1281 long nr_blockdev_pages(void);
1282 #else /* CONFIG_BLOCK */
1283 struct blk_plug {
1284 };
1285 
1286 static inline void blk_start_plug(struct blk_plug *plug)
1287 {
1288 }
1289 
1290 static inline void blk_finish_plug(struct blk_plug *plug)
1291 {
1292 }
1293 
1294 static inline void blk_flush_plug(struct task_struct *task)
1295 {
1296 }
1297 
1298 static inline void blk_schedule_flush_plug(struct task_struct *task)
1299 {
1300 }
1301 
1302 
1303 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1304 {
1305 	return false;
1306 }
1307 
1308 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask)
1309 {
1310 	return 0;
1311 }
1312 
1313 static inline long nr_blockdev_pages(void)
1314 {
1315 	return 0;
1316 }
1317 #endif /* CONFIG_BLOCK */
1318 
1319 extern void blk_io_schedule(void);
1320 
1321 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1322 		sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1323 
1324 #define BLKDEV_DISCARD_SECURE	(1 << 0)	/* issue a secure erase */
1325 
1326 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1327 		sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1328 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1329 		sector_t nr_sects, gfp_t gfp_mask, int flags,
1330 		struct bio **biop);
1331 
1332 #define BLKDEV_ZERO_NOUNMAP	(1 << 0)  /* do not free blocks */
1333 #define BLKDEV_ZERO_NOFALLBACK	(1 << 1)  /* don't write explicit zeroes */
1334 
1335 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1336 		sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1337 		unsigned flags);
1338 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1339 		sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1340 
1341 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1342 		sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1343 {
1344 	return blkdev_issue_discard(sb->s_bdev,
1345 				    block << (sb->s_blocksize_bits -
1346 					      SECTOR_SHIFT),
1347 				    nr_blocks << (sb->s_blocksize_bits -
1348 						  SECTOR_SHIFT),
1349 				    gfp_mask, flags);
1350 }
1351 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1352 		sector_t nr_blocks, gfp_t gfp_mask)
1353 {
1354 	return blkdev_issue_zeroout(sb->s_bdev,
1355 				    block << (sb->s_blocksize_bits -
1356 					      SECTOR_SHIFT),
1357 				    nr_blocks << (sb->s_blocksize_bits -
1358 						  SECTOR_SHIFT),
1359 				    gfp_mask, 0);
1360 }
1361 
1362 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1363 
1364 static inline bool bdev_is_partition(struct block_device *bdev)
1365 {
1366 	return bdev->bd_partno;
1367 }
1368 
1369 enum blk_default_limits {
1370 	BLK_MAX_SEGMENTS	= 128,
1371 	BLK_SAFE_MAX_SECTORS	= 255,
1372 	BLK_DEF_MAX_SECTORS	= 2560,
1373 	BLK_MAX_SEGMENT_SIZE	= 65536,
1374 	BLK_SEG_BOUNDARY_MASK	= 0xFFFFFFFFUL,
1375 };
1376 
1377 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1378 {
1379 	return q->limits.seg_boundary_mask;
1380 }
1381 
1382 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1383 {
1384 	return q->limits.virt_boundary_mask;
1385 }
1386 
1387 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1388 {
1389 	return q->limits.max_sectors;
1390 }
1391 
1392 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1393 {
1394 	return q->limits.max_hw_sectors;
1395 }
1396 
1397 static inline unsigned short queue_max_segments(const struct request_queue *q)
1398 {
1399 	return q->limits.max_segments;
1400 }
1401 
1402 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1403 {
1404 	return q->limits.max_discard_segments;
1405 }
1406 
1407 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1408 {
1409 	return q->limits.max_segment_size;
1410 }
1411 
1412 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1413 {
1414 
1415 	const struct queue_limits *l = &q->limits;
1416 
1417 	return min(l->max_zone_append_sectors, l->max_sectors);
1418 }
1419 
1420 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1421 {
1422 	int retval = 512;
1423 
1424 	if (q && q->limits.logical_block_size)
1425 		retval = q->limits.logical_block_size;
1426 
1427 	return retval;
1428 }
1429 
1430 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1431 {
1432 	return queue_logical_block_size(bdev_get_queue(bdev));
1433 }
1434 
1435 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1436 {
1437 	return q->limits.physical_block_size;
1438 }
1439 
1440 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1441 {
1442 	return queue_physical_block_size(bdev_get_queue(bdev));
1443 }
1444 
1445 static inline unsigned int queue_io_min(const struct request_queue *q)
1446 {
1447 	return q->limits.io_min;
1448 }
1449 
1450 static inline int bdev_io_min(struct block_device *bdev)
1451 {
1452 	return queue_io_min(bdev_get_queue(bdev));
1453 }
1454 
1455 static inline unsigned int queue_io_opt(const struct request_queue *q)
1456 {
1457 	return q->limits.io_opt;
1458 }
1459 
1460 static inline int bdev_io_opt(struct block_device *bdev)
1461 {
1462 	return queue_io_opt(bdev_get_queue(bdev));
1463 }
1464 
1465 static inline int queue_alignment_offset(const struct request_queue *q)
1466 {
1467 	if (q->limits.misaligned)
1468 		return -1;
1469 
1470 	return q->limits.alignment_offset;
1471 }
1472 
1473 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1474 {
1475 	unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1476 	unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1477 		<< SECTOR_SHIFT;
1478 
1479 	return (granularity + lim->alignment_offset - alignment) % granularity;
1480 }
1481 
1482 static inline int bdev_alignment_offset(struct block_device *bdev)
1483 {
1484 	struct request_queue *q = bdev_get_queue(bdev);
1485 
1486 	if (q->limits.misaligned)
1487 		return -1;
1488 	if (bdev_is_partition(bdev))
1489 		return queue_limit_alignment_offset(&q->limits,
1490 				bdev->bd_part->start_sect);
1491 	return q->limits.alignment_offset;
1492 }
1493 
1494 static inline int queue_discard_alignment(const struct request_queue *q)
1495 {
1496 	if (q->limits.discard_misaligned)
1497 		return -1;
1498 
1499 	return q->limits.discard_alignment;
1500 }
1501 
1502 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1503 {
1504 	unsigned int alignment, granularity, offset;
1505 
1506 	if (!lim->max_discard_sectors)
1507 		return 0;
1508 
1509 	/* Why are these in bytes, not sectors? */
1510 	alignment = lim->discard_alignment >> SECTOR_SHIFT;
1511 	granularity = lim->discard_granularity >> SECTOR_SHIFT;
1512 	if (!granularity)
1513 		return 0;
1514 
1515 	/* Offset of the partition start in 'granularity' sectors */
1516 	offset = sector_div(sector, granularity);
1517 
1518 	/* And why do we do this modulus *again* in blkdev_issue_discard()? */
1519 	offset = (granularity + alignment - offset) % granularity;
1520 
1521 	/* Turn it back into bytes, gaah */
1522 	return offset << SECTOR_SHIFT;
1523 }
1524 
1525 static inline int bdev_discard_alignment(struct block_device *bdev)
1526 {
1527 	struct request_queue *q = bdev_get_queue(bdev);
1528 
1529 	if (bdev_is_partition(bdev))
1530 		return queue_limit_discard_alignment(&q->limits,
1531 				bdev->bd_part->start_sect);
1532 	return q->limits.discard_alignment;
1533 }
1534 
1535 static inline unsigned int bdev_write_same(struct block_device *bdev)
1536 {
1537 	struct request_queue *q = bdev_get_queue(bdev);
1538 
1539 	if (q)
1540 		return q->limits.max_write_same_sectors;
1541 
1542 	return 0;
1543 }
1544 
1545 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1546 {
1547 	struct request_queue *q = bdev_get_queue(bdev);
1548 
1549 	if (q)
1550 		return q->limits.max_write_zeroes_sectors;
1551 
1552 	return 0;
1553 }
1554 
1555 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1556 {
1557 	struct request_queue *q = bdev_get_queue(bdev);
1558 
1559 	if (q)
1560 		return blk_queue_zoned_model(q);
1561 
1562 	return BLK_ZONED_NONE;
1563 }
1564 
1565 static inline bool bdev_is_zoned(struct block_device *bdev)
1566 {
1567 	struct request_queue *q = bdev_get_queue(bdev);
1568 
1569 	if (q)
1570 		return blk_queue_is_zoned(q);
1571 
1572 	return false;
1573 }
1574 
1575 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1576 {
1577 	struct request_queue *q = bdev_get_queue(bdev);
1578 
1579 	if (q)
1580 		return blk_queue_zone_sectors(q);
1581 	return 0;
1582 }
1583 
1584 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1585 {
1586 	struct request_queue *q = bdev_get_queue(bdev);
1587 
1588 	if (q)
1589 		return queue_max_open_zones(q);
1590 	return 0;
1591 }
1592 
1593 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1594 {
1595 	struct request_queue *q = bdev_get_queue(bdev);
1596 
1597 	if (q)
1598 		return queue_max_active_zones(q);
1599 	return 0;
1600 }
1601 
1602 static inline int queue_dma_alignment(const struct request_queue *q)
1603 {
1604 	return q ? q->dma_alignment : 511;
1605 }
1606 
1607 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1608 				 unsigned int len)
1609 {
1610 	unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1611 	return !(addr & alignment) && !(len & alignment);
1612 }
1613 
1614 /* assumes size > 256 */
1615 static inline unsigned int blksize_bits(unsigned int size)
1616 {
1617 	unsigned int bits = 8;
1618 	do {
1619 		bits++;
1620 		size >>= 1;
1621 	} while (size > 256);
1622 	return bits;
1623 }
1624 
1625 static inline unsigned int block_size(struct block_device *bdev)
1626 {
1627 	return 1 << bdev->bd_inode->i_blkbits;
1628 }
1629 
1630 int kblockd_schedule_work(struct work_struct *work);
1631 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1632 
1633 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1634 	MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1635 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1636 	MODULE_ALIAS("block-major-" __stringify(major) "-*")
1637 
1638 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1639 
1640 enum blk_integrity_flags {
1641 	BLK_INTEGRITY_VERIFY		= 1 << 0,
1642 	BLK_INTEGRITY_GENERATE		= 1 << 1,
1643 	BLK_INTEGRITY_DEVICE_CAPABLE	= 1 << 2,
1644 	BLK_INTEGRITY_IP_CHECKSUM	= 1 << 3,
1645 };
1646 
1647 struct blk_integrity_iter {
1648 	void			*prot_buf;
1649 	void			*data_buf;
1650 	sector_t		seed;
1651 	unsigned int		data_size;
1652 	unsigned short		interval;
1653 	const char		*disk_name;
1654 };
1655 
1656 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1657 typedef void (integrity_prepare_fn) (struct request *);
1658 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1659 
1660 struct blk_integrity_profile {
1661 	integrity_processing_fn		*generate_fn;
1662 	integrity_processing_fn		*verify_fn;
1663 	integrity_prepare_fn		*prepare_fn;
1664 	integrity_complete_fn		*complete_fn;
1665 	const char			*name;
1666 };
1667 
1668 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1669 extern void blk_integrity_unregister(struct gendisk *);
1670 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1671 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1672 				   struct scatterlist *);
1673 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1674 
1675 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1676 {
1677 	struct blk_integrity *bi = &disk->queue->integrity;
1678 
1679 	if (!bi->profile)
1680 		return NULL;
1681 
1682 	return bi;
1683 }
1684 
1685 static inline
1686 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1687 {
1688 	return blk_get_integrity(bdev->bd_disk);
1689 }
1690 
1691 static inline bool
1692 blk_integrity_queue_supports_integrity(struct request_queue *q)
1693 {
1694 	return q->integrity.profile;
1695 }
1696 
1697 static inline bool blk_integrity_rq(struct request *rq)
1698 {
1699 	return rq->cmd_flags & REQ_INTEGRITY;
1700 }
1701 
1702 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1703 						    unsigned int segs)
1704 {
1705 	q->limits.max_integrity_segments = segs;
1706 }
1707 
1708 static inline unsigned short
1709 queue_max_integrity_segments(const struct request_queue *q)
1710 {
1711 	return q->limits.max_integrity_segments;
1712 }
1713 
1714 /**
1715  * bio_integrity_intervals - Return number of integrity intervals for a bio
1716  * @bi:		blk_integrity profile for device
1717  * @sectors:	Size of the bio in 512-byte sectors
1718  *
1719  * Description: The block layer calculates everything in 512 byte
1720  * sectors but integrity metadata is done in terms of the data integrity
1721  * interval size of the storage device.  Convert the block layer sectors
1722  * to the appropriate number of integrity intervals.
1723  */
1724 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1725 						   unsigned int sectors)
1726 {
1727 	return sectors >> (bi->interval_exp - 9);
1728 }
1729 
1730 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1731 					       unsigned int sectors)
1732 {
1733 	return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1734 }
1735 
1736 /*
1737  * Return the first bvec that contains integrity data.  Only drivers that are
1738  * limited to a single integrity segment should use this helper.
1739  */
1740 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1741 {
1742 	if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1743 		return NULL;
1744 	return rq->bio->bi_integrity->bip_vec;
1745 }
1746 
1747 #else /* CONFIG_BLK_DEV_INTEGRITY */
1748 
1749 struct bio;
1750 struct block_device;
1751 struct gendisk;
1752 struct blk_integrity;
1753 
1754 static inline int blk_integrity_rq(struct request *rq)
1755 {
1756 	return 0;
1757 }
1758 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1759 					    struct bio *b)
1760 {
1761 	return 0;
1762 }
1763 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1764 					  struct bio *b,
1765 					  struct scatterlist *s)
1766 {
1767 	return 0;
1768 }
1769 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1770 {
1771 	return NULL;
1772 }
1773 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1774 {
1775 	return NULL;
1776 }
1777 static inline bool
1778 blk_integrity_queue_supports_integrity(struct request_queue *q)
1779 {
1780 	return false;
1781 }
1782 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1783 {
1784 	return 0;
1785 }
1786 static inline void blk_integrity_register(struct gendisk *d,
1787 					 struct blk_integrity *b)
1788 {
1789 }
1790 static inline void blk_integrity_unregister(struct gendisk *d)
1791 {
1792 }
1793 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1794 						    unsigned int segs)
1795 {
1796 }
1797 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1798 {
1799 	return 0;
1800 }
1801 
1802 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1803 						   unsigned int sectors)
1804 {
1805 	return 0;
1806 }
1807 
1808 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1809 					       unsigned int sectors)
1810 {
1811 	return 0;
1812 }
1813 
1814 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1815 {
1816 	return NULL;
1817 }
1818 
1819 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1820 
1821 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1822 
1823 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1824 
1825 void blk_ksm_unregister(struct request_queue *q);
1826 
1827 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1828 
1829 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1830 				    struct request_queue *q)
1831 {
1832 	return true;
1833 }
1834 
1835 static inline void blk_ksm_unregister(struct request_queue *q) { }
1836 
1837 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1838 
1839 
1840 struct block_device_operations {
1841 	blk_qc_t (*submit_bio) (struct bio *bio);
1842 	int (*open) (struct block_device *, fmode_t);
1843 	void (*release) (struct gendisk *, fmode_t);
1844 	int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1845 	int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1846 	int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1847 	unsigned int (*check_events) (struct gendisk *disk,
1848 				      unsigned int clearing);
1849 	void (*unlock_native_capacity) (struct gendisk *);
1850 	int (*revalidate_disk) (struct gendisk *);
1851 	int (*getgeo)(struct block_device *, struct hd_geometry *);
1852 	/* this callback is with swap_lock and sometimes page table lock held */
1853 	void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1854 	int (*report_zones)(struct gendisk *, sector_t sector,
1855 			unsigned int nr_zones, report_zones_cb cb, void *data);
1856 	char *(*devnode)(struct gendisk *disk, umode_t *mode);
1857 	struct module *owner;
1858 	const struct pr_ops *pr_ops;
1859 };
1860 
1861 #ifdef CONFIG_COMPAT
1862 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1863 				      unsigned int, unsigned long);
1864 #else
1865 #define blkdev_compat_ptr_ioctl NULL
1866 #endif
1867 
1868 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1869 				 unsigned long);
1870 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1871 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1872 						struct writeback_control *);
1873 
1874 #ifdef CONFIG_BLK_DEV_ZONED
1875 bool blk_req_needs_zone_write_lock(struct request *rq);
1876 bool blk_req_zone_write_trylock(struct request *rq);
1877 void __blk_req_zone_write_lock(struct request *rq);
1878 void __blk_req_zone_write_unlock(struct request *rq);
1879 
1880 static inline void blk_req_zone_write_lock(struct request *rq)
1881 {
1882 	if (blk_req_needs_zone_write_lock(rq))
1883 		__blk_req_zone_write_lock(rq);
1884 }
1885 
1886 static inline void blk_req_zone_write_unlock(struct request *rq)
1887 {
1888 	if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1889 		__blk_req_zone_write_unlock(rq);
1890 }
1891 
1892 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1893 {
1894 	return rq->q->seq_zones_wlock &&
1895 		test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1896 }
1897 
1898 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1899 {
1900 	if (!blk_req_needs_zone_write_lock(rq))
1901 		return true;
1902 	return !blk_req_zone_is_write_locked(rq);
1903 }
1904 #else
1905 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1906 {
1907 	return false;
1908 }
1909 
1910 static inline void blk_req_zone_write_lock(struct request *rq)
1911 {
1912 }
1913 
1914 static inline void blk_req_zone_write_unlock(struct request *rq)
1915 {
1916 }
1917 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1918 {
1919 	return false;
1920 }
1921 
1922 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1923 {
1924 	return true;
1925 }
1926 #endif /* CONFIG_BLK_DEV_ZONED */
1927 
1928 static inline void blk_wake_io_task(struct task_struct *waiter)
1929 {
1930 	/*
1931 	 * If we're polling, the task itself is doing the completions. For
1932 	 * that case, we don't need to signal a wakeup, it's enough to just
1933 	 * mark us as RUNNING.
1934 	 */
1935 	if (waiter == current)
1936 		__set_current_state(TASK_RUNNING);
1937 	else
1938 		wake_up_process(waiter);
1939 }
1940 
1941 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1942 		unsigned int op);
1943 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1944 		unsigned long start_time);
1945 
1946 unsigned long part_start_io_acct(struct gendisk *disk, struct hd_struct **part,
1947 				 struct bio *bio);
1948 void part_end_io_acct(struct hd_struct *part, struct bio *bio,
1949 		      unsigned long start_time);
1950 
1951 /**
1952  * bio_start_io_acct - start I/O accounting for bio based drivers
1953  * @bio:	bio to start account for
1954  *
1955  * Returns the start time that should be passed back to bio_end_io_acct().
1956  */
1957 static inline unsigned long bio_start_io_acct(struct bio *bio)
1958 {
1959 	return disk_start_io_acct(bio->bi_disk, bio_sectors(bio), bio_op(bio));
1960 }
1961 
1962 /**
1963  * bio_end_io_acct - end I/O accounting for bio based drivers
1964  * @bio:	bio to end account for
1965  * @start:	start time returned by bio_start_io_acct()
1966  */
1967 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1968 {
1969 	return disk_end_io_acct(bio->bi_disk, bio_op(bio), start_time);
1970 }
1971 
1972 int bdev_read_only(struct block_device *bdev);
1973 int set_blocksize(struct block_device *bdev, int size);
1974 
1975 const char *bdevname(struct block_device *bdev, char *buffer);
1976 struct block_device *lookup_bdev(const char *);
1977 
1978 void blkdev_show(struct seq_file *seqf, off_t offset);
1979 
1980 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
1981 #define BDEVT_SIZE	10	/* Largest string for MAJ:MIN for blkdev */
1982 #ifdef CONFIG_BLOCK
1983 #define BLKDEV_MAJOR_MAX	512
1984 #else
1985 #define BLKDEV_MAJOR_MAX	0
1986 #endif
1987 
1988 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1989 		void *holder);
1990 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1991 int bd_prepare_to_claim(struct block_device *bdev, struct block_device *whole,
1992 		void *holder);
1993 void bd_abort_claiming(struct block_device *bdev, struct block_device *whole,
1994 		void *holder);
1995 void blkdev_put(struct block_device *bdev, fmode_t mode);
1996 
1997 struct block_device *I_BDEV(struct inode *inode);
1998 struct block_device *bdget_part(struct hd_struct *part);
1999 struct block_device *bdgrab(struct block_device *bdev);
2000 void bdput(struct block_device *);
2001 
2002 #ifdef CONFIG_BLOCK
2003 void invalidate_bdev(struct block_device *bdev);
2004 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
2005 			loff_t lend);
2006 int sync_blockdev(struct block_device *bdev);
2007 #else
2008 static inline void invalidate_bdev(struct block_device *bdev)
2009 {
2010 }
2011 static inline int truncate_bdev_range(struct block_device *bdev, fmode_t mode,
2012 				      loff_t lstart, loff_t lend)
2013 {
2014 	return 0;
2015 }
2016 static inline int sync_blockdev(struct block_device *bdev)
2017 {
2018 	return 0;
2019 }
2020 #endif
2021 int fsync_bdev(struct block_device *bdev);
2022 
2023 struct super_block *freeze_bdev(struct block_device *bdev);
2024 int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2025 
2026 #endif /* _LINUX_BLKDEV_H */
2027