xref: /linux-6.15/include/linux/blkdev.h (revision 2676eb4b)
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/minmax.h>
12 #include <linux/timer.h>
13 #include <linux/workqueue.h>
14 #include <linux/pagemap.h>
15 #include <linux/backing-dev-defs.h>
16 #include <linux/wait.h>
17 #include <linux/mempool.h>
18 #include <linux/pfn.h>
19 #include <linux/bio.h>
20 #include <linux/stringify.h>
21 #include <linux/gfp.h>
22 #include <linux/bsg.h>
23 #include <linux/smp.h>
24 #include <linux/rcupdate.h>
25 #include <linux/percpu-refcount.h>
26 #include <linux/scatterlist.h>
27 #include <linux/blkzoned.h>
28 #include <linux/pm.h>
29 
30 struct module;
31 struct scsi_ioctl_command;
32 
33 struct request_queue;
34 struct elevator_queue;
35 struct blk_trace;
36 struct request;
37 struct sg_io_hdr;
38 struct bsg_job;
39 struct blkcg_gq;
40 struct blk_flush_queue;
41 struct pr_ops;
42 struct rq_qos;
43 struct blk_queue_stats;
44 struct blk_stat_callback;
45 struct blk_keyslot_manager;
46 
47 #define BLKDEV_MIN_RQ	4
48 #define BLKDEV_MAX_RQ	128	/* Default maximum */
49 
50 /* Must be consistent with blk_mq_poll_stats_bkt() */
51 #define BLK_MQ_POLL_STATS_BKTS 16
52 
53 /* Doing classic polling */
54 #define BLK_MQ_POLL_CLASSIC -1
55 
56 /*
57  * Maximum number of blkcg policies allowed to be registered concurrently.
58  * Defined here to simplify include dependency.
59  */
60 #define BLKCG_MAX_POLS		5
61 
62 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
63 
64 /*
65  * request flags */
66 typedef __u32 __bitwise req_flags_t;
67 
68 /* elevator knows about this request */
69 #define RQF_SORTED		((__force req_flags_t)(1 << 0))
70 /* drive already may have started this one */
71 #define RQF_STARTED		((__force req_flags_t)(1 << 1))
72 /* may not be passed by ioscheduler */
73 #define RQF_SOFTBARRIER		((__force req_flags_t)(1 << 3))
74 /* request for flush sequence */
75 #define RQF_FLUSH_SEQ		((__force req_flags_t)(1 << 4))
76 /* merge of different types, fail separately */
77 #define RQF_MIXED_MERGE		((__force req_flags_t)(1 << 5))
78 /* track inflight for MQ */
79 #define RQF_MQ_INFLIGHT		((__force req_flags_t)(1 << 6))
80 /* don't call prep for this one */
81 #define RQF_DONTPREP		((__force req_flags_t)(1 << 7))
82 /* vaguely specified driver internal error.  Ignored by the block layer */
83 #define RQF_FAILED		((__force req_flags_t)(1 << 10))
84 /* don't warn about errors */
85 #define RQF_QUIET		((__force req_flags_t)(1 << 11))
86 /* elevator private data attached */
87 #define RQF_ELVPRIV		((__force req_flags_t)(1 << 12))
88 /* account into disk and partition IO statistics */
89 #define RQF_IO_STAT		((__force req_flags_t)(1 << 13))
90 /* request came from our alloc pool */
91 #define RQF_ALLOCED		((__force req_flags_t)(1 << 14))
92 /* runtime pm request */
93 #define RQF_PM			((__force req_flags_t)(1 << 15))
94 /* on IO scheduler merge hash */
95 #define RQF_HASHED		((__force req_flags_t)(1 << 16))
96 /* track IO completion time */
97 #define RQF_STATS		((__force req_flags_t)(1 << 17))
98 /* Look at ->special_vec for the actual data payload instead of the
99    bio chain. */
100 #define RQF_SPECIAL_PAYLOAD	((__force req_flags_t)(1 << 18))
101 /* The per-zone write lock is held for this request */
102 #define RQF_ZONE_WRITE_LOCKED	((__force req_flags_t)(1 << 19))
103 /* already slept for hybrid poll */
104 #define RQF_MQ_POLL_SLEPT	((__force req_flags_t)(1 << 20))
105 /* ->timeout has been called, don't expire again */
106 #define RQF_TIMED_OUT		((__force req_flags_t)(1 << 21))
107 
108 /* flags that prevent us from merging requests: */
109 #define RQF_NOMERGE_FLAGS \
110 	(RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
111 
112 /*
113  * Request state for blk-mq.
114  */
115 enum mq_rq_state {
116 	MQ_RQ_IDLE		= 0,
117 	MQ_RQ_IN_FLIGHT		= 1,
118 	MQ_RQ_COMPLETE		= 2,
119 };
120 
121 /*
122  * Try to put the fields that are referenced together in the same cacheline.
123  *
124  * If you modify this structure, make sure to update blk_rq_init() and
125  * especially blk_mq_rq_ctx_init() to take care of the added fields.
126  */
127 struct request {
128 	struct request_queue *q;
129 	struct blk_mq_ctx *mq_ctx;
130 	struct blk_mq_hw_ctx *mq_hctx;
131 
132 	unsigned int cmd_flags;		/* op and common flags */
133 	req_flags_t rq_flags;
134 
135 	int tag;
136 	int internal_tag;
137 
138 	/* the following two fields are internal, NEVER access directly */
139 	unsigned int __data_len;	/* total data len */
140 	sector_t __sector;		/* sector cursor */
141 
142 	struct bio *bio;
143 	struct bio *biotail;
144 
145 	struct list_head queuelist;
146 
147 	/*
148 	 * The hash is used inside the scheduler, and killed once the
149 	 * request reaches the dispatch list. The ipi_list is only used
150 	 * to queue the request for softirq completion, which is long
151 	 * after the request has been unhashed (and even removed from
152 	 * the dispatch list).
153 	 */
154 	union {
155 		struct hlist_node hash;	/* merge hash */
156 		struct llist_node ipi_list;
157 	};
158 
159 	/*
160 	 * The rb_node is only used inside the io scheduler, requests
161 	 * are pruned when moved to the dispatch queue. So let the
162 	 * completion_data share space with the rb_node.
163 	 */
164 	union {
165 		struct rb_node rb_node;	/* sort/lookup */
166 		struct bio_vec special_vec;
167 		void *completion_data;
168 		int error_count; /* for legacy drivers, don't use */
169 	};
170 
171 	/*
172 	 * Three pointers are available for the IO schedulers, if they need
173 	 * more they have to dynamically allocate it.  Flush requests are
174 	 * never put on the IO scheduler. So let the flush fields share
175 	 * space with the elevator data.
176 	 */
177 	union {
178 		struct {
179 			struct io_cq		*icq;
180 			void			*priv[2];
181 		} elv;
182 
183 		struct {
184 			unsigned int		seq;
185 			struct list_head	list;
186 			rq_end_io_fn		*saved_end_io;
187 		} flush;
188 	};
189 
190 	struct gendisk *rq_disk;
191 	struct block_device *part;
192 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
193 	/* Time that the first bio started allocating this request. */
194 	u64 alloc_time_ns;
195 #endif
196 	/* Time that this request was allocated for this IO. */
197 	u64 start_time_ns;
198 	/* Time that I/O was submitted to the device. */
199 	u64 io_start_time_ns;
200 
201 #ifdef CONFIG_BLK_WBT
202 	unsigned short wbt_flags;
203 #endif
204 	/*
205 	 * rq sectors used for blk stats. It has the same value
206 	 * with blk_rq_sectors(rq), except that it never be zeroed
207 	 * by completion.
208 	 */
209 	unsigned short stats_sectors;
210 
211 	/*
212 	 * Number of scatter-gather DMA addr+len pairs after
213 	 * physical address coalescing is performed.
214 	 */
215 	unsigned short nr_phys_segments;
216 
217 #if defined(CONFIG_BLK_DEV_INTEGRITY)
218 	unsigned short nr_integrity_segments;
219 #endif
220 
221 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
222 	struct bio_crypt_ctx *crypt_ctx;
223 	struct blk_ksm_keyslot *crypt_keyslot;
224 #endif
225 
226 	unsigned short write_hint;
227 	unsigned short ioprio;
228 
229 	enum mq_rq_state state;
230 	refcount_t ref;
231 
232 	unsigned int timeout;
233 	unsigned long deadline;
234 
235 	union {
236 		struct __call_single_data csd;
237 		u64 fifo_time;
238 	};
239 
240 	/*
241 	 * completion callback.
242 	 */
243 	rq_end_io_fn *end_io;
244 	void *end_io_data;
245 };
246 
247 static inline bool blk_op_is_scsi(unsigned int op)
248 {
249 	return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
250 }
251 
252 static inline bool blk_op_is_private(unsigned int op)
253 {
254 	return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
255 }
256 
257 static inline bool blk_rq_is_scsi(struct request *rq)
258 {
259 	return blk_op_is_scsi(req_op(rq));
260 }
261 
262 static inline bool blk_rq_is_private(struct request *rq)
263 {
264 	return blk_op_is_private(req_op(rq));
265 }
266 
267 static inline bool blk_rq_is_passthrough(struct request *rq)
268 {
269 	return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
270 }
271 
272 static inline bool bio_is_passthrough(struct bio *bio)
273 {
274 	unsigned op = bio_op(bio);
275 
276 	return blk_op_is_scsi(op) || blk_op_is_private(op);
277 }
278 
279 static inline unsigned short req_get_ioprio(struct request *req)
280 {
281 	return req->ioprio;
282 }
283 
284 #include <linux/elevator.h>
285 
286 struct blk_queue_ctx;
287 
288 struct bio_vec;
289 
290 enum blk_eh_timer_return {
291 	BLK_EH_DONE,		/* drivers has completed the command */
292 	BLK_EH_RESET_TIMER,	/* reset timer and try again */
293 };
294 
295 enum blk_queue_state {
296 	Queue_down,
297 	Queue_up,
298 };
299 
300 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
301 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
302 
303 #define BLK_SCSI_MAX_CMDS	(256)
304 #define BLK_SCSI_CMD_PER_LONG	(BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
305 
306 /*
307  * Zoned block device models (zoned limit).
308  *
309  * Note: This needs to be ordered from the least to the most severe
310  * restrictions for the inheritance in blk_stack_limits() to work.
311  */
312 enum blk_zoned_model {
313 	BLK_ZONED_NONE = 0,	/* Regular block device */
314 	BLK_ZONED_HA,		/* Host-aware zoned block device */
315 	BLK_ZONED_HM,		/* Host-managed zoned block device */
316 };
317 
318 struct queue_limits {
319 	unsigned long		bounce_pfn;
320 	unsigned long		seg_boundary_mask;
321 	unsigned long		virt_boundary_mask;
322 
323 	unsigned int		max_hw_sectors;
324 	unsigned int		max_dev_sectors;
325 	unsigned int		chunk_sectors;
326 	unsigned int		max_sectors;
327 	unsigned int		max_segment_size;
328 	unsigned int		physical_block_size;
329 	unsigned int		logical_block_size;
330 	unsigned int		alignment_offset;
331 	unsigned int		io_min;
332 	unsigned int		io_opt;
333 	unsigned int		max_discard_sectors;
334 	unsigned int		max_hw_discard_sectors;
335 	unsigned int		max_write_same_sectors;
336 	unsigned int		max_write_zeroes_sectors;
337 	unsigned int		max_zone_append_sectors;
338 	unsigned int		discard_granularity;
339 	unsigned int		discard_alignment;
340 	unsigned int		zone_write_granularity;
341 
342 	unsigned short		max_segments;
343 	unsigned short		max_integrity_segments;
344 	unsigned short		max_discard_segments;
345 
346 	unsigned char		misaligned;
347 	unsigned char		discard_misaligned;
348 	unsigned char		raid_partial_stripes_expensive;
349 	enum blk_zoned_model	zoned;
350 };
351 
352 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
353 			       void *data);
354 
355 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
356 
357 #ifdef CONFIG_BLK_DEV_ZONED
358 
359 #define BLK_ALL_ZONES  ((unsigned int)-1)
360 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
361 			unsigned int nr_zones, report_zones_cb cb, void *data);
362 unsigned int blkdev_nr_zones(struct gendisk *disk);
363 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
364 			    sector_t sectors, sector_t nr_sectors,
365 			    gfp_t gfp_mask);
366 int blk_revalidate_disk_zones(struct gendisk *disk,
367 			      void (*update_driver_data)(struct gendisk *disk));
368 
369 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
370 				     unsigned int cmd, unsigned long arg);
371 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
372 				  unsigned int cmd, unsigned long arg);
373 
374 #else /* CONFIG_BLK_DEV_ZONED */
375 
376 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
377 {
378 	return 0;
379 }
380 
381 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
382 					    fmode_t mode, unsigned int cmd,
383 					    unsigned long arg)
384 {
385 	return -ENOTTY;
386 }
387 
388 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
389 					 fmode_t mode, unsigned int cmd,
390 					 unsigned long arg)
391 {
392 	return -ENOTTY;
393 }
394 
395 #endif /* CONFIG_BLK_DEV_ZONED */
396 
397 struct request_queue {
398 	struct request		*last_merge;
399 	struct elevator_queue	*elevator;
400 
401 	struct percpu_ref	q_usage_counter;
402 
403 	struct blk_queue_stats	*stats;
404 	struct rq_qos		*rq_qos;
405 
406 	const struct blk_mq_ops	*mq_ops;
407 
408 	/* sw queues */
409 	struct blk_mq_ctx __percpu	*queue_ctx;
410 
411 	unsigned int		queue_depth;
412 
413 	/* hw dispatch queues */
414 	struct blk_mq_hw_ctx	**queue_hw_ctx;
415 	unsigned int		nr_hw_queues;
416 
417 	struct backing_dev_info	*backing_dev_info;
418 
419 	/*
420 	 * The queue owner gets to use this for whatever they like.
421 	 * ll_rw_blk doesn't touch it.
422 	 */
423 	void			*queuedata;
424 
425 	/*
426 	 * various queue flags, see QUEUE_* below
427 	 */
428 	unsigned long		queue_flags;
429 	/*
430 	 * Number of contexts that have called blk_set_pm_only(). If this
431 	 * counter is above zero then only RQF_PM requests are processed.
432 	 */
433 	atomic_t		pm_only;
434 
435 	/*
436 	 * ida allocated id for this queue.  Used to index queues from
437 	 * ioctx.
438 	 */
439 	int			id;
440 
441 	/*
442 	 * queue needs bounce pages for pages above this limit
443 	 */
444 	gfp_t			bounce_gfp;
445 
446 	spinlock_t		queue_lock;
447 
448 	/*
449 	 * queue kobject
450 	 */
451 	struct kobject kobj;
452 
453 	/*
454 	 * mq queue kobject
455 	 */
456 	struct kobject *mq_kobj;
457 
458 #ifdef  CONFIG_BLK_DEV_INTEGRITY
459 	struct blk_integrity integrity;
460 #endif	/* CONFIG_BLK_DEV_INTEGRITY */
461 
462 #ifdef CONFIG_PM
463 	struct device		*dev;
464 	enum rpm_status		rpm_status;
465 #endif
466 
467 	/*
468 	 * queue settings
469 	 */
470 	unsigned long		nr_requests;	/* Max # of requests */
471 
472 	unsigned int		dma_pad_mask;
473 	unsigned int		dma_alignment;
474 
475 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
476 	/* Inline crypto capabilities */
477 	struct blk_keyslot_manager *ksm;
478 #endif
479 
480 	unsigned int		rq_timeout;
481 	int			poll_nsec;
482 
483 	struct blk_stat_callback	*poll_cb;
484 	struct blk_rq_stat	poll_stat[BLK_MQ_POLL_STATS_BKTS];
485 
486 	struct timer_list	timeout;
487 	struct work_struct	timeout_work;
488 
489 	atomic_t		nr_active_requests_shared_sbitmap;
490 
491 	struct list_head	icq_list;
492 #ifdef CONFIG_BLK_CGROUP
493 	DECLARE_BITMAP		(blkcg_pols, BLKCG_MAX_POLS);
494 	struct blkcg_gq		*root_blkg;
495 	struct list_head	blkg_list;
496 #endif
497 
498 	struct queue_limits	limits;
499 
500 	unsigned int		required_elevator_features;
501 
502 #ifdef CONFIG_BLK_DEV_ZONED
503 	/*
504 	 * Zoned block device information for request dispatch control.
505 	 * nr_zones is the total number of zones of the device. This is always
506 	 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
507 	 * bits which indicates if a zone is conventional (bit set) or
508 	 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
509 	 * bits which indicates if a zone is write locked, that is, if a write
510 	 * request targeting the zone was dispatched. All three fields are
511 	 * initialized by the low level device driver (e.g. scsi/sd.c).
512 	 * Stacking drivers (device mappers) may or may not initialize
513 	 * these fields.
514 	 *
515 	 * Reads of this information must be protected with blk_queue_enter() /
516 	 * blk_queue_exit(). Modifying this information is only allowed while
517 	 * no requests are being processed. See also blk_mq_freeze_queue() and
518 	 * blk_mq_unfreeze_queue().
519 	 */
520 	unsigned int		nr_zones;
521 	unsigned long		*conv_zones_bitmap;
522 	unsigned long		*seq_zones_wlock;
523 	unsigned int		max_open_zones;
524 	unsigned int		max_active_zones;
525 #endif /* CONFIG_BLK_DEV_ZONED */
526 
527 	/*
528 	 * sg stuff
529 	 */
530 	unsigned int		sg_timeout;
531 	unsigned int		sg_reserved_size;
532 	int			node;
533 	struct mutex		debugfs_mutex;
534 #ifdef CONFIG_BLK_DEV_IO_TRACE
535 	struct blk_trace __rcu	*blk_trace;
536 #endif
537 	/*
538 	 * for flush operations
539 	 */
540 	struct blk_flush_queue	*fq;
541 
542 	struct list_head	requeue_list;
543 	spinlock_t		requeue_lock;
544 	struct delayed_work	requeue_work;
545 
546 	struct mutex		sysfs_lock;
547 	struct mutex		sysfs_dir_lock;
548 
549 	/*
550 	 * for reusing dead hctx instance in case of updating
551 	 * nr_hw_queues
552 	 */
553 	struct list_head	unused_hctx_list;
554 	spinlock_t		unused_hctx_lock;
555 
556 	int			mq_freeze_depth;
557 
558 #if defined(CONFIG_BLK_DEV_BSG)
559 	struct bsg_class_device bsg_dev;
560 #endif
561 
562 #ifdef CONFIG_BLK_DEV_THROTTLING
563 	/* Throttle data */
564 	struct throtl_data *td;
565 #endif
566 	struct rcu_head		rcu_head;
567 	wait_queue_head_t	mq_freeze_wq;
568 	/*
569 	 * Protect concurrent access to q_usage_counter by
570 	 * percpu_ref_kill() and percpu_ref_reinit().
571 	 */
572 	struct mutex		mq_freeze_lock;
573 
574 	struct blk_mq_tag_set	*tag_set;
575 	struct list_head	tag_set_list;
576 	struct bio_set		bio_split;
577 
578 	struct dentry		*debugfs_dir;
579 
580 #ifdef CONFIG_BLK_DEBUG_FS
581 	struct dentry		*sched_debugfs_dir;
582 	struct dentry		*rqos_debugfs_dir;
583 #endif
584 
585 	bool			mq_sysfs_init_done;
586 
587 	size_t			cmd_size;
588 
589 #define BLK_MAX_WRITE_HINTS	5
590 	u64			write_hints[BLK_MAX_WRITE_HINTS];
591 };
592 
593 /* Keep blk_queue_flag_name[] in sync with the definitions below */
594 #define QUEUE_FLAG_STOPPED	0	/* queue is stopped */
595 #define QUEUE_FLAG_DYING	1	/* queue being torn down */
596 #define QUEUE_FLAG_NOMERGES     3	/* disable merge attempts */
597 #define QUEUE_FLAG_SAME_COMP	4	/* complete on same CPU-group */
598 #define QUEUE_FLAG_FAIL_IO	5	/* fake timeout */
599 #define QUEUE_FLAG_NONROT	6	/* non-rotational device (SSD) */
600 #define QUEUE_FLAG_VIRT		QUEUE_FLAG_NONROT /* paravirt device */
601 #define QUEUE_FLAG_IO_STAT	7	/* do disk/partitions IO accounting */
602 #define QUEUE_FLAG_DISCARD	8	/* supports DISCARD */
603 #define QUEUE_FLAG_NOXMERGES	9	/* No extended merges */
604 #define QUEUE_FLAG_ADD_RANDOM	10	/* Contributes to random pool */
605 #define QUEUE_FLAG_SECERASE	11	/* supports secure erase */
606 #define QUEUE_FLAG_SAME_FORCE	12	/* force complete on same CPU */
607 #define QUEUE_FLAG_DEAD		13	/* queue tear-down finished */
608 #define QUEUE_FLAG_INIT_DONE	14	/* queue is initialized */
609 #define QUEUE_FLAG_STABLE_WRITES 15	/* don't modify blks until WB is done */
610 #define QUEUE_FLAG_POLL		16	/* IO polling enabled if set */
611 #define QUEUE_FLAG_WC		17	/* Write back caching */
612 #define QUEUE_FLAG_FUA		18	/* device supports FUA writes */
613 #define QUEUE_FLAG_DAX		19	/* device supports DAX */
614 #define QUEUE_FLAG_STATS	20	/* track IO start and completion times */
615 #define QUEUE_FLAG_POLL_STATS	21	/* collecting stats for hybrid polling */
616 #define QUEUE_FLAG_REGISTERED	22	/* queue has been registered to a disk */
617 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23	/* queue supports SCSI commands */
618 #define QUEUE_FLAG_QUIESCED	24	/* queue has been quiesced */
619 #define QUEUE_FLAG_PCI_P2PDMA	25	/* device supports PCI p2p requests */
620 #define QUEUE_FLAG_ZONE_RESETALL 26	/* supports Zone Reset All */
621 #define QUEUE_FLAG_RQ_ALLOC_TIME 27	/* record rq->alloc_time_ns */
622 #define QUEUE_FLAG_HCTX_ACTIVE	28	/* at least one blk-mq hctx is active */
623 #define QUEUE_FLAG_NOWAIT       29	/* device supports NOWAIT */
624 
625 #define QUEUE_FLAG_MQ_DEFAULT	((1 << QUEUE_FLAG_IO_STAT) |		\
626 				 (1 << QUEUE_FLAG_SAME_COMP) |		\
627 				 (1 << QUEUE_FLAG_NOWAIT))
628 
629 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
630 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
631 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
632 
633 #define blk_queue_stopped(q)	test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
634 #define blk_queue_dying(q)	test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
635 #define blk_queue_dead(q)	test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
636 #define blk_queue_init_done(q)	test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
637 #define blk_queue_nomerges(q)	test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
638 #define blk_queue_noxmerges(q)	\
639 	test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
640 #define blk_queue_nonrot(q)	test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
641 #define blk_queue_stable_writes(q) \
642 	test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
643 #define blk_queue_io_stat(q)	test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
644 #define blk_queue_add_random(q)	test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
645 #define blk_queue_discard(q)	test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
646 #define blk_queue_zone_resetall(q)	\
647 	test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
648 #define blk_queue_secure_erase(q) \
649 	(test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
650 #define blk_queue_dax(q)	test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
651 #define blk_queue_scsi_passthrough(q)	\
652 	test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
653 #define blk_queue_pci_p2pdma(q)	\
654 	test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
655 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
656 #define blk_queue_rq_alloc_time(q)	\
657 	test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
658 #else
659 #define blk_queue_rq_alloc_time(q)	false
660 #endif
661 
662 #define blk_noretry_request(rq) \
663 	((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
664 			     REQ_FAILFAST_DRIVER))
665 #define blk_queue_quiesced(q)	test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
666 #define blk_queue_pm_only(q)	atomic_read(&(q)->pm_only)
667 #define blk_queue_fua(q)	test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
668 #define blk_queue_registered(q)	test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
669 #define blk_queue_nowait(q)	test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
670 
671 extern void blk_set_pm_only(struct request_queue *q);
672 extern void blk_clear_pm_only(struct request_queue *q);
673 
674 static inline bool blk_account_rq(struct request *rq)
675 {
676 	return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
677 }
678 
679 #define list_entry_rq(ptr)	list_entry((ptr), struct request, queuelist)
680 
681 #define rq_data_dir(rq)		(op_is_write(req_op(rq)) ? WRITE : READ)
682 
683 #define rq_dma_dir(rq) \
684 	(op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
685 
686 #define dma_map_bvec(dev, bv, dir, attrs) \
687 	dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
688 	(dir), (attrs))
689 
690 static inline bool queue_is_mq(struct request_queue *q)
691 {
692 	return q->mq_ops;
693 }
694 
695 #ifdef CONFIG_PM
696 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
697 {
698 	return q->rpm_status;
699 }
700 #else
701 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
702 {
703 	return RPM_ACTIVE;
704 }
705 #endif
706 
707 static inline enum blk_zoned_model
708 blk_queue_zoned_model(struct request_queue *q)
709 {
710 	if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
711 		return q->limits.zoned;
712 	return BLK_ZONED_NONE;
713 }
714 
715 static inline bool blk_queue_is_zoned(struct request_queue *q)
716 {
717 	switch (blk_queue_zoned_model(q)) {
718 	case BLK_ZONED_HA:
719 	case BLK_ZONED_HM:
720 		return true;
721 	default:
722 		return false;
723 	}
724 }
725 
726 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
727 {
728 	return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
729 }
730 
731 #ifdef CONFIG_BLK_DEV_ZONED
732 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
733 {
734 	return blk_queue_is_zoned(q) ? q->nr_zones : 0;
735 }
736 
737 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
738 					     sector_t sector)
739 {
740 	if (!blk_queue_is_zoned(q))
741 		return 0;
742 	return sector >> ilog2(q->limits.chunk_sectors);
743 }
744 
745 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
746 					 sector_t sector)
747 {
748 	if (!blk_queue_is_zoned(q))
749 		return false;
750 	if (!q->conv_zones_bitmap)
751 		return true;
752 	return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
753 }
754 
755 static inline void blk_queue_max_open_zones(struct request_queue *q,
756 		unsigned int max_open_zones)
757 {
758 	q->max_open_zones = max_open_zones;
759 }
760 
761 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
762 {
763 	return q->max_open_zones;
764 }
765 
766 static inline void blk_queue_max_active_zones(struct request_queue *q,
767 		unsigned int max_active_zones)
768 {
769 	q->max_active_zones = max_active_zones;
770 }
771 
772 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
773 {
774 	return q->max_active_zones;
775 }
776 #else /* CONFIG_BLK_DEV_ZONED */
777 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
778 {
779 	return 0;
780 }
781 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
782 					 sector_t sector)
783 {
784 	return false;
785 }
786 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
787 					     sector_t sector)
788 {
789 	return 0;
790 }
791 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
792 {
793 	return 0;
794 }
795 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
796 {
797 	return 0;
798 }
799 #endif /* CONFIG_BLK_DEV_ZONED */
800 
801 static inline bool rq_is_sync(struct request *rq)
802 {
803 	return op_is_sync(rq->cmd_flags);
804 }
805 
806 static inline bool rq_mergeable(struct request *rq)
807 {
808 	if (blk_rq_is_passthrough(rq))
809 		return false;
810 
811 	if (req_op(rq) == REQ_OP_FLUSH)
812 		return false;
813 
814 	if (req_op(rq) == REQ_OP_WRITE_ZEROES)
815 		return false;
816 
817 	if (req_op(rq) == REQ_OP_ZONE_APPEND)
818 		return false;
819 
820 	if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
821 		return false;
822 	if (rq->rq_flags & RQF_NOMERGE_FLAGS)
823 		return false;
824 
825 	return true;
826 }
827 
828 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
829 {
830 	if (bio_page(a) == bio_page(b) &&
831 	    bio_offset(a) == bio_offset(b))
832 		return true;
833 
834 	return false;
835 }
836 
837 static inline unsigned int blk_queue_depth(struct request_queue *q)
838 {
839 	if (q->queue_depth)
840 		return q->queue_depth;
841 
842 	return q->nr_requests;
843 }
844 
845 extern unsigned long blk_max_low_pfn, blk_max_pfn;
846 
847 /*
848  * standard bounce addresses:
849  *
850  * BLK_BOUNCE_HIGH	: bounce all highmem pages
851  * BLK_BOUNCE_ANY	: don't bounce anything
852  * BLK_BOUNCE_ISA	: bounce pages above ISA DMA boundary
853  */
854 
855 #if BITS_PER_LONG == 32
856 #define BLK_BOUNCE_HIGH		((u64)blk_max_low_pfn << PAGE_SHIFT)
857 #else
858 #define BLK_BOUNCE_HIGH		-1ULL
859 #endif
860 #define BLK_BOUNCE_ANY		(-1ULL)
861 #define BLK_BOUNCE_ISA		(DMA_BIT_MASK(24))
862 
863 /*
864  * default timeout for SG_IO if none specified
865  */
866 #define BLK_DEFAULT_SG_TIMEOUT	(60 * HZ)
867 #define BLK_MIN_SG_TIMEOUT	(7 * HZ)
868 
869 struct rq_map_data {
870 	struct page **pages;
871 	int page_order;
872 	int nr_entries;
873 	unsigned long offset;
874 	int null_mapped;
875 	int from_user;
876 };
877 
878 struct req_iterator {
879 	struct bvec_iter iter;
880 	struct bio *bio;
881 };
882 
883 /* This should not be used directly - use rq_for_each_segment */
884 #define for_each_bio(_bio)		\
885 	for (; _bio; _bio = _bio->bi_next)
886 #define __rq_for_each_bio(_bio, rq)	\
887 	if ((rq->bio))			\
888 		for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
889 
890 #define rq_for_each_segment(bvl, _rq, _iter)			\
891 	__rq_for_each_bio(_iter.bio, _rq)			\
892 		bio_for_each_segment(bvl, _iter.bio, _iter.iter)
893 
894 #define rq_for_each_bvec(bvl, _rq, _iter)			\
895 	__rq_for_each_bio(_iter.bio, _rq)			\
896 		bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
897 
898 #define rq_iter_last(bvec, _iter)				\
899 		(_iter.bio->bi_next == NULL &&			\
900 		 bio_iter_last(bvec, _iter.iter))
901 
902 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
903 # error	"You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
904 #endif
905 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
906 extern void rq_flush_dcache_pages(struct request *rq);
907 #else
908 static inline void rq_flush_dcache_pages(struct request *rq)
909 {
910 }
911 #endif
912 
913 extern int blk_register_queue(struct gendisk *disk);
914 extern void blk_unregister_queue(struct gendisk *disk);
915 blk_qc_t submit_bio_noacct(struct bio *bio);
916 extern void blk_rq_init(struct request_queue *q, struct request *rq);
917 extern void blk_put_request(struct request *);
918 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
919 				       blk_mq_req_flags_t flags);
920 extern int blk_lld_busy(struct request_queue *q);
921 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
922 			     struct bio_set *bs, gfp_t gfp_mask,
923 			     int (*bio_ctr)(struct bio *, struct bio *, void *),
924 			     void *data);
925 extern void blk_rq_unprep_clone(struct request *rq);
926 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
927 				     struct request *rq);
928 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
929 extern void blk_queue_split(struct bio **);
930 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
931 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
932 			      unsigned int, void __user *);
933 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
934 			  unsigned int, void __user *);
935 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
936 			 struct scsi_ioctl_command __user *);
937 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
938 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
939 
940 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
941 extern void blk_queue_exit(struct request_queue *q);
942 extern void blk_sync_queue(struct request_queue *q);
943 extern int blk_rq_map_user(struct request_queue *, struct request *,
944 			   struct rq_map_data *, void __user *, unsigned long,
945 			   gfp_t);
946 extern int blk_rq_unmap_user(struct bio *);
947 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
948 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
949 			       struct rq_map_data *, const struct iov_iter *,
950 			       gfp_t);
951 extern void blk_execute_rq(struct gendisk *, struct request *, int);
952 extern void blk_execute_rq_nowait(struct gendisk *,
953 				  struct request *, int, rq_end_io_fn *);
954 
955 /* Helper to convert REQ_OP_XXX to its string format XXX */
956 extern const char *blk_op_str(unsigned int op);
957 
958 int blk_status_to_errno(blk_status_t status);
959 blk_status_t errno_to_blk_status(int errno);
960 
961 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
962 
963 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
964 {
965 	return bdev->bd_disk->queue;	/* this is never NULL */
966 }
967 
968 /*
969  * The basic unit of block I/O is a sector. It is used in a number of contexts
970  * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
971  * bytes. Variables of type sector_t represent an offset or size that is a
972  * multiple of 512 bytes. Hence these two constants.
973  */
974 #ifndef SECTOR_SHIFT
975 #define SECTOR_SHIFT 9
976 #endif
977 #ifndef SECTOR_SIZE
978 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
979 #endif
980 
981 /*
982  * blk_rq_pos()			: the current sector
983  * blk_rq_bytes()		: bytes left in the entire request
984  * blk_rq_cur_bytes()		: bytes left in the current segment
985  * blk_rq_err_bytes()		: bytes left till the next error boundary
986  * blk_rq_sectors()		: sectors left in the entire request
987  * blk_rq_cur_sectors()		: sectors left in the current segment
988  * blk_rq_stats_sectors()	: sectors of the entire request used for stats
989  */
990 static inline sector_t blk_rq_pos(const struct request *rq)
991 {
992 	return rq->__sector;
993 }
994 
995 static inline unsigned int blk_rq_bytes(const struct request *rq)
996 {
997 	return rq->__data_len;
998 }
999 
1000 static inline int blk_rq_cur_bytes(const struct request *rq)
1001 {
1002 	return rq->bio ? bio_cur_bytes(rq->bio) : 0;
1003 }
1004 
1005 extern unsigned int blk_rq_err_bytes(const struct request *rq);
1006 
1007 static inline unsigned int blk_rq_sectors(const struct request *rq)
1008 {
1009 	return blk_rq_bytes(rq) >> SECTOR_SHIFT;
1010 }
1011 
1012 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
1013 {
1014 	return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
1015 }
1016 
1017 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
1018 {
1019 	return rq->stats_sectors;
1020 }
1021 
1022 #ifdef CONFIG_BLK_DEV_ZONED
1023 
1024 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
1025 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
1026 
1027 static inline unsigned int blk_rq_zone_no(struct request *rq)
1028 {
1029 	return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1030 }
1031 
1032 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1033 {
1034 	return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1035 }
1036 #endif /* CONFIG_BLK_DEV_ZONED */
1037 
1038 /*
1039  * Some commands like WRITE SAME have a payload or data transfer size which
1040  * is different from the size of the request.  Any driver that supports such
1041  * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1042  * calculate the data transfer size.
1043  */
1044 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1045 {
1046 	if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1047 		return rq->special_vec.bv_len;
1048 	return blk_rq_bytes(rq);
1049 }
1050 
1051 /*
1052  * Return the first full biovec in the request.  The caller needs to check that
1053  * there are any bvecs before calling this helper.
1054  */
1055 static inline struct bio_vec req_bvec(struct request *rq)
1056 {
1057 	if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1058 		return rq->special_vec;
1059 	return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1060 }
1061 
1062 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1063 						     int op)
1064 {
1065 	if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1066 		return min(q->limits.max_discard_sectors,
1067 			   UINT_MAX >> SECTOR_SHIFT);
1068 
1069 	if (unlikely(op == REQ_OP_WRITE_SAME))
1070 		return q->limits.max_write_same_sectors;
1071 
1072 	if (unlikely(op == REQ_OP_WRITE_ZEROES))
1073 		return q->limits.max_write_zeroes_sectors;
1074 
1075 	return q->limits.max_sectors;
1076 }
1077 
1078 /*
1079  * Return maximum size of a request at given offset. Only valid for
1080  * file system requests.
1081  */
1082 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1083 					       sector_t offset,
1084 					       unsigned int chunk_sectors)
1085 {
1086 	if (!chunk_sectors) {
1087 		if (q->limits.chunk_sectors)
1088 			chunk_sectors = q->limits.chunk_sectors;
1089 		else
1090 			return q->limits.max_sectors;
1091 	}
1092 
1093 	if (likely(is_power_of_2(chunk_sectors)))
1094 		chunk_sectors -= offset & (chunk_sectors - 1);
1095 	else
1096 		chunk_sectors -= sector_div(offset, chunk_sectors);
1097 
1098 	return min(q->limits.max_sectors, chunk_sectors);
1099 }
1100 
1101 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1102 						  sector_t offset)
1103 {
1104 	struct request_queue *q = rq->q;
1105 
1106 	if (blk_rq_is_passthrough(rq))
1107 		return q->limits.max_hw_sectors;
1108 
1109 	if (!q->limits.chunk_sectors ||
1110 	    req_op(rq) == REQ_OP_DISCARD ||
1111 	    req_op(rq) == REQ_OP_SECURE_ERASE)
1112 		return blk_queue_get_max_sectors(q, req_op(rq));
1113 
1114 	return min(blk_max_size_offset(q, offset, 0),
1115 			blk_queue_get_max_sectors(q, req_op(rq)));
1116 }
1117 
1118 static inline unsigned int blk_rq_count_bios(struct request *rq)
1119 {
1120 	unsigned int nr_bios = 0;
1121 	struct bio *bio;
1122 
1123 	__rq_for_each_bio(bio, rq)
1124 		nr_bios++;
1125 
1126 	return nr_bios;
1127 }
1128 
1129 void blk_steal_bios(struct bio_list *list, struct request *rq);
1130 
1131 /*
1132  * Request completion related functions.
1133  *
1134  * blk_update_request() completes given number of bytes and updates
1135  * the request without completing it.
1136  */
1137 extern bool blk_update_request(struct request *rq, blk_status_t error,
1138 			       unsigned int nr_bytes);
1139 
1140 extern void blk_abort_request(struct request *);
1141 
1142 /*
1143  * Access functions for manipulating queue properties
1144  */
1145 extern void blk_cleanup_queue(struct request_queue *);
1146 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1147 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1148 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1149 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1150 extern void blk_queue_max_discard_segments(struct request_queue *,
1151 		unsigned short);
1152 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1153 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1154 		unsigned int max_discard_sectors);
1155 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1156 		unsigned int max_write_same_sectors);
1157 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1158 		unsigned int max_write_same_sectors);
1159 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1160 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1161 		unsigned int max_zone_append_sectors);
1162 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1163 void blk_queue_zone_write_granularity(struct request_queue *q,
1164 				      unsigned int size);
1165 extern void blk_queue_alignment_offset(struct request_queue *q,
1166 				       unsigned int alignment);
1167 void blk_queue_update_readahead(struct request_queue *q);
1168 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1169 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1170 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1171 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1172 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1173 extern void blk_set_default_limits(struct queue_limits *lim);
1174 extern void blk_set_stacking_limits(struct queue_limits *lim);
1175 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1176 			    sector_t offset);
1177 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1178 			      sector_t offset);
1179 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1180 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1181 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1182 extern void blk_queue_dma_alignment(struct request_queue *, int);
1183 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1184 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1185 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1186 extern void blk_queue_required_elevator_features(struct request_queue *q,
1187 						 unsigned int features);
1188 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1189 					      struct device *dev);
1190 
1191 /*
1192  * Number of physical segments as sent to the device.
1193  *
1194  * Normally this is the number of discontiguous data segments sent by the
1195  * submitter.  But for data-less command like discard we might have no
1196  * actual data segments submitted, but the driver might have to add it's
1197  * own special payload.  In that case we still return 1 here so that this
1198  * special payload will be mapped.
1199  */
1200 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1201 {
1202 	if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1203 		return 1;
1204 	return rq->nr_phys_segments;
1205 }
1206 
1207 /*
1208  * Number of discard segments (or ranges) the driver needs to fill in.
1209  * Each discard bio merged into a request is counted as one segment.
1210  */
1211 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1212 {
1213 	return max_t(unsigned short, rq->nr_phys_segments, 1);
1214 }
1215 
1216 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1217 		struct scatterlist *sglist, struct scatterlist **last_sg);
1218 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1219 		struct scatterlist *sglist)
1220 {
1221 	struct scatterlist *last_sg = NULL;
1222 
1223 	return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1224 }
1225 extern void blk_dump_rq_flags(struct request *, char *);
1226 
1227 bool __must_check blk_get_queue(struct request_queue *);
1228 struct request_queue *blk_alloc_queue(int node_id);
1229 extern void blk_put_queue(struct request_queue *);
1230 extern void blk_set_queue_dying(struct request_queue *);
1231 
1232 #ifdef CONFIG_BLOCK
1233 /*
1234  * blk_plug permits building a queue of related requests by holding the I/O
1235  * fragments for a short period. This allows merging of sequential requests
1236  * into single larger request. As the requests are moved from a per-task list to
1237  * the device's request_queue in a batch, this results in improved scalability
1238  * as the lock contention for request_queue lock is reduced.
1239  *
1240  * It is ok not to disable preemption when adding the request to the plug list
1241  * or when attempting a merge, because blk_schedule_flush_list() will only flush
1242  * the plug list when the task sleeps by itself. For details, please see
1243  * schedule() where blk_schedule_flush_plug() is called.
1244  */
1245 struct blk_plug {
1246 	struct list_head mq_list; /* blk-mq requests */
1247 	struct list_head cb_list; /* md requires an unplug callback */
1248 	unsigned short rq_count;
1249 	bool multiple_queues;
1250 	bool nowait;
1251 };
1252 #define BLK_MAX_REQUEST_COUNT 16
1253 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1254 
1255 struct blk_plug_cb;
1256 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1257 struct blk_plug_cb {
1258 	struct list_head list;
1259 	blk_plug_cb_fn callback;
1260 	void *data;
1261 };
1262 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1263 					     void *data, int size);
1264 extern void blk_start_plug(struct blk_plug *);
1265 extern void blk_finish_plug(struct blk_plug *);
1266 extern void blk_flush_plug_list(struct blk_plug *, bool);
1267 
1268 static inline void blk_flush_plug(struct task_struct *tsk)
1269 {
1270 	struct blk_plug *plug = tsk->plug;
1271 
1272 	if (plug)
1273 		blk_flush_plug_list(plug, false);
1274 }
1275 
1276 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1277 {
1278 	struct blk_plug *plug = tsk->plug;
1279 
1280 	if (plug)
1281 		blk_flush_plug_list(plug, true);
1282 }
1283 
1284 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1285 {
1286 	struct blk_plug *plug = tsk->plug;
1287 
1288 	return plug &&
1289 		 (!list_empty(&plug->mq_list) ||
1290 		 !list_empty(&plug->cb_list));
1291 }
1292 
1293 int blkdev_issue_flush(struct block_device *bdev);
1294 long nr_blockdev_pages(void);
1295 #else /* CONFIG_BLOCK */
1296 struct blk_plug {
1297 };
1298 
1299 static inline void blk_start_plug(struct blk_plug *plug)
1300 {
1301 }
1302 
1303 static inline void blk_finish_plug(struct blk_plug *plug)
1304 {
1305 }
1306 
1307 static inline void blk_flush_plug(struct task_struct *task)
1308 {
1309 }
1310 
1311 static inline void blk_schedule_flush_plug(struct task_struct *task)
1312 {
1313 }
1314 
1315 
1316 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1317 {
1318 	return false;
1319 }
1320 
1321 static inline int blkdev_issue_flush(struct block_device *bdev)
1322 {
1323 	return 0;
1324 }
1325 
1326 static inline long nr_blockdev_pages(void)
1327 {
1328 	return 0;
1329 }
1330 #endif /* CONFIG_BLOCK */
1331 
1332 extern void blk_io_schedule(void);
1333 
1334 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1335 		sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1336 
1337 #define BLKDEV_DISCARD_SECURE	(1 << 0)	/* issue a secure erase */
1338 
1339 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1340 		sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1341 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1342 		sector_t nr_sects, gfp_t gfp_mask, int flags,
1343 		struct bio **biop);
1344 
1345 #define BLKDEV_ZERO_NOUNMAP	(1 << 0)  /* do not free blocks */
1346 #define BLKDEV_ZERO_NOFALLBACK	(1 << 1)  /* don't write explicit zeroes */
1347 
1348 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1349 		sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1350 		unsigned flags);
1351 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1352 		sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1353 
1354 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1355 		sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1356 {
1357 	return blkdev_issue_discard(sb->s_bdev,
1358 				    block << (sb->s_blocksize_bits -
1359 					      SECTOR_SHIFT),
1360 				    nr_blocks << (sb->s_blocksize_bits -
1361 						  SECTOR_SHIFT),
1362 				    gfp_mask, flags);
1363 }
1364 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1365 		sector_t nr_blocks, gfp_t gfp_mask)
1366 {
1367 	return blkdev_issue_zeroout(sb->s_bdev,
1368 				    block << (sb->s_blocksize_bits -
1369 					      SECTOR_SHIFT),
1370 				    nr_blocks << (sb->s_blocksize_bits -
1371 						  SECTOR_SHIFT),
1372 				    gfp_mask, 0);
1373 }
1374 
1375 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1376 
1377 static inline bool bdev_is_partition(struct block_device *bdev)
1378 {
1379 	return bdev->bd_partno;
1380 }
1381 
1382 enum blk_default_limits {
1383 	BLK_MAX_SEGMENTS	= 128,
1384 	BLK_SAFE_MAX_SECTORS	= 255,
1385 	BLK_DEF_MAX_SECTORS	= 2560,
1386 	BLK_MAX_SEGMENT_SIZE	= 65536,
1387 	BLK_SEG_BOUNDARY_MASK	= 0xFFFFFFFFUL,
1388 };
1389 
1390 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1391 {
1392 	return q->limits.seg_boundary_mask;
1393 }
1394 
1395 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1396 {
1397 	return q->limits.virt_boundary_mask;
1398 }
1399 
1400 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1401 {
1402 	return q->limits.max_sectors;
1403 }
1404 
1405 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1406 {
1407 	return q->limits.max_hw_sectors;
1408 }
1409 
1410 static inline unsigned short queue_max_segments(const struct request_queue *q)
1411 {
1412 	return q->limits.max_segments;
1413 }
1414 
1415 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1416 {
1417 	return q->limits.max_discard_segments;
1418 }
1419 
1420 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1421 {
1422 	return q->limits.max_segment_size;
1423 }
1424 
1425 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1426 {
1427 
1428 	const struct queue_limits *l = &q->limits;
1429 
1430 	return min(l->max_zone_append_sectors, l->max_sectors);
1431 }
1432 
1433 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1434 {
1435 	int retval = 512;
1436 
1437 	if (q && q->limits.logical_block_size)
1438 		retval = q->limits.logical_block_size;
1439 
1440 	return retval;
1441 }
1442 
1443 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1444 {
1445 	return queue_logical_block_size(bdev_get_queue(bdev));
1446 }
1447 
1448 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1449 {
1450 	return q->limits.physical_block_size;
1451 }
1452 
1453 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1454 {
1455 	return queue_physical_block_size(bdev_get_queue(bdev));
1456 }
1457 
1458 static inline unsigned int queue_io_min(const struct request_queue *q)
1459 {
1460 	return q->limits.io_min;
1461 }
1462 
1463 static inline int bdev_io_min(struct block_device *bdev)
1464 {
1465 	return queue_io_min(bdev_get_queue(bdev));
1466 }
1467 
1468 static inline unsigned int queue_io_opt(const struct request_queue *q)
1469 {
1470 	return q->limits.io_opt;
1471 }
1472 
1473 static inline int bdev_io_opt(struct block_device *bdev)
1474 {
1475 	return queue_io_opt(bdev_get_queue(bdev));
1476 }
1477 
1478 static inline unsigned int
1479 queue_zone_write_granularity(const struct request_queue *q)
1480 {
1481 	return q->limits.zone_write_granularity;
1482 }
1483 
1484 static inline unsigned int
1485 bdev_zone_write_granularity(struct block_device *bdev)
1486 {
1487 	return queue_zone_write_granularity(bdev_get_queue(bdev));
1488 }
1489 
1490 static inline int queue_alignment_offset(const struct request_queue *q)
1491 {
1492 	if (q->limits.misaligned)
1493 		return -1;
1494 
1495 	return q->limits.alignment_offset;
1496 }
1497 
1498 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1499 {
1500 	unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1501 	unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1502 		<< SECTOR_SHIFT;
1503 
1504 	return (granularity + lim->alignment_offset - alignment) % granularity;
1505 }
1506 
1507 static inline int bdev_alignment_offset(struct block_device *bdev)
1508 {
1509 	struct request_queue *q = bdev_get_queue(bdev);
1510 
1511 	if (q->limits.misaligned)
1512 		return -1;
1513 	if (bdev_is_partition(bdev))
1514 		return queue_limit_alignment_offset(&q->limits,
1515 				bdev->bd_start_sect);
1516 	return q->limits.alignment_offset;
1517 }
1518 
1519 static inline int queue_discard_alignment(const struct request_queue *q)
1520 {
1521 	if (q->limits.discard_misaligned)
1522 		return -1;
1523 
1524 	return q->limits.discard_alignment;
1525 }
1526 
1527 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1528 {
1529 	unsigned int alignment, granularity, offset;
1530 
1531 	if (!lim->max_discard_sectors)
1532 		return 0;
1533 
1534 	/* Why are these in bytes, not sectors? */
1535 	alignment = lim->discard_alignment >> SECTOR_SHIFT;
1536 	granularity = lim->discard_granularity >> SECTOR_SHIFT;
1537 	if (!granularity)
1538 		return 0;
1539 
1540 	/* Offset of the partition start in 'granularity' sectors */
1541 	offset = sector_div(sector, granularity);
1542 
1543 	/* And why do we do this modulus *again* in blkdev_issue_discard()? */
1544 	offset = (granularity + alignment - offset) % granularity;
1545 
1546 	/* Turn it back into bytes, gaah */
1547 	return offset << SECTOR_SHIFT;
1548 }
1549 
1550 static inline int bdev_discard_alignment(struct block_device *bdev)
1551 {
1552 	struct request_queue *q = bdev_get_queue(bdev);
1553 
1554 	if (bdev_is_partition(bdev))
1555 		return queue_limit_discard_alignment(&q->limits,
1556 				bdev->bd_start_sect);
1557 	return q->limits.discard_alignment;
1558 }
1559 
1560 static inline unsigned int bdev_write_same(struct block_device *bdev)
1561 {
1562 	struct request_queue *q = bdev_get_queue(bdev);
1563 
1564 	if (q)
1565 		return q->limits.max_write_same_sectors;
1566 
1567 	return 0;
1568 }
1569 
1570 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1571 {
1572 	struct request_queue *q = bdev_get_queue(bdev);
1573 
1574 	if (q)
1575 		return q->limits.max_write_zeroes_sectors;
1576 
1577 	return 0;
1578 }
1579 
1580 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1581 {
1582 	struct request_queue *q = bdev_get_queue(bdev);
1583 
1584 	if (q)
1585 		return blk_queue_zoned_model(q);
1586 
1587 	return BLK_ZONED_NONE;
1588 }
1589 
1590 static inline bool bdev_is_zoned(struct block_device *bdev)
1591 {
1592 	struct request_queue *q = bdev_get_queue(bdev);
1593 
1594 	if (q)
1595 		return blk_queue_is_zoned(q);
1596 
1597 	return false;
1598 }
1599 
1600 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1601 {
1602 	struct request_queue *q = bdev_get_queue(bdev);
1603 
1604 	if (q)
1605 		return blk_queue_zone_sectors(q);
1606 	return 0;
1607 }
1608 
1609 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1610 {
1611 	struct request_queue *q = bdev_get_queue(bdev);
1612 
1613 	if (q)
1614 		return queue_max_open_zones(q);
1615 	return 0;
1616 }
1617 
1618 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1619 {
1620 	struct request_queue *q = bdev_get_queue(bdev);
1621 
1622 	if (q)
1623 		return queue_max_active_zones(q);
1624 	return 0;
1625 }
1626 
1627 static inline int queue_dma_alignment(const struct request_queue *q)
1628 {
1629 	return q ? q->dma_alignment : 511;
1630 }
1631 
1632 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1633 				 unsigned int len)
1634 {
1635 	unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1636 	return !(addr & alignment) && !(len & alignment);
1637 }
1638 
1639 /* assumes size > 256 */
1640 static inline unsigned int blksize_bits(unsigned int size)
1641 {
1642 	unsigned int bits = 8;
1643 	do {
1644 		bits++;
1645 		size >>= 1;
1646 	} while (size > 256);
1647 	return bits;
1648 }
1649 
1650 static inline unsigned int block_size(struct block_device *bdev)
1651 {
1652 	return 1 << bdev->bd_inode->i_blkbits;
1653 }
1654 
1655 int kblockd_schedule_work(struct work_struct *work);
1656 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1657 
1658 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1659 	MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1660 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1661 	MODULE_ALIAS("block-major-" __stringify(major) "-*")
1662 
1663 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1664 
1665 enum blk_integrity_flags {
1666 	BLK_INTEGRITY_VERIFY		= 1 << 0,
1667 	BLK_INTEGRITY_GENERATE		= 1 << 1,
1668 	BLK_INTEGRITY_DEVICE_CAPABLE	= 1 << 2,
1669 	BLK_INTEGRITY_IP_CHECKSUM	= 1 << 3,
1670 };
1671 
1672 struct blk_integrity_iter {
1673 	void			*prot_buf;
1674 	void			*data_buf;
1675 	sector_t		seed;
1676 	unsigned int		data_size;
1677 	unsigned short		interval;
1678 	const char		*disk_name;
1679 };
1680 
1681 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1682 typedef void (integrity_prepare_fn) (struct request *);
1683 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1684 
1685 struct blk_integrity_profile {
1686 	integrity_processing_fn		*generate_fn;
1687 	integrity_processing_fn		*verify_fn;
1688 	integrity_prepare_fn		*prepare_fn;
1689 	integrity_complete_fn		*complete_fn;
1690 	const char			*name;
1691 };
1692 
1693 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1694 extern void blk_integrity_unregister(struct gendisk *);
1695 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1696 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1697 				   struct scatterlist *);
1698 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1699 
1700 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1701 {
1702 	struct blk_integrity *bi = &disk->queue->integrity;
1703 
1704 	if (!bi->profile)
1705 		return NULL;
1706 
1707 	return bi;
1708 }
1709 
1710 static inline
1711 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1712 {
1713 	return blk_get_integrity(bdev->bd_disk);
1714 }
1715 
1716 static inline bool
1717 blk_integrity_queue_supports_integrity(struct request_queue *q)
1718 {
1719 	return q->integrity.profile;
1720 }
1721 
1722 static inline bool blk_integrity_rq(struct request *rq)
1723 {
1724 	return rq->cmd_flags & REQ_INTEGRITY;
1725 }
1726 
1727 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1728 						    unsigned int segs)
1729 {
1730 	q->limits.max_integrity_segments = segs;
1731 }
1732 
1733 static inline unsigned short
1734 queue_max_integrity_segments(const struct request_queue *q)
1735 {
1736 	return q->limits.max_integrity_segments;
1737 }
1738 
1739 /**
1740  * bio_integrity_intervals - Return number of integrity intervals for a bio
1741  * @bi:		blk_integrity profile for device
1742  * @sectors:	Size of the bio in 512-byte sectors
1743  *
1744  * Description: The block layer calculates everything in 512 byte
1745  * sectors but integrity metadata is done in terms of the data integrity
1746  * interval size of the storage device.  Convert the block layer sectors
1747  * to the appropriate number of integrity intervals.
1748  */
1749 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1750 						   unsigned int sectors)
1751 {
1752 	return sectors >> (bi->interval_exp - 9);
1753 }
1754 
1755 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1756 					       unsigned int sectors)
1757 {
1758 	return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1759 }
1760 
1761 /*
1762  * Return the first bvec that contains integrity data.  Only drivers that are
1763  * limited to a single integrity segment should use this helper.
1764  */
1765 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1766 {
1767 	if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1768 		return NULL;
1769 	return rq->bio->bi_integrity->bip_vec;
1770 }
1771 
1772 #else /* CONFIG_BLK_DEV_INTEGRITY */
1773 
1774 struct bio;
1775 struct block_device;
1776 struct gendisk;
1777 struct blk_integrity;
1778 
1779 static inline int blk_integrity_rq(struct request *rq)
1780 {
1781 	return 0;
1782 }
1783 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1784 					    struct bio *b)
1785 {
1786 	return 0;
1787 }
1788 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1789 					  struct bio *b,
1790 					  struct scatterlist *s)
1791 {
1792 	return 0;
1793 }
1794 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1795 {
1796 	return NULL;
1797 }
1798 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1799 {
1800 	return NULL;
1801 }
1802 static inline bool
1803 blk_integrity_queue_supports_integrity(struct request_queue *q)
1804 {
1805 	return false;
1806 }
1807 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1808 {
1809 	return 0;
1810 }
1811 static inline void blk_integrity_register(struct gendisk *d,
1812 					 struct blk_integrity *b)
1813 {
1814 }
1815 static inline void blk_integrity_unregister(struct gendisk *d)
1816 {
1817 }
1818 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1819 						    unsigned int segs)
1820 {
1821 }
1822 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1823 {
1824 	return 0;
1825 }
1826 
1827 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1828 						   unsigned int sectors)
1829 {
1830 	return 0;
1831 }
1832 
1833 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1834 					       unsigned int sectors)
1835 {
1836 	return 0;
1837 }
1838 
1839 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1840 {
1841 	return NULL;
1842 }
1843 
1844 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1845 
1846 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1847 
1848 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1849 
1850 void blk_ksm_unregister(struct request_queue *q);
1851 
1852 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1853 
1854 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1855 				    struct request_queue *q)
1856 {
1857 	return true;
1858 }
1859 
1860 static inline void blk_ksm_unregister(struct request_queue *q) { }
1861 
1862 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1863 
1864 
1865 struct block_device_operations {
1866 	blk_qc_t (*submit_bio) (struct bio *bio);
1867 	int (*open) (struct block_device *, fmode_t);
1868 	void (*release) (struct gendisk *, fmode_t);
1869 	int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1870 	int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1871 	int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1872 	unsigned int (*check_events) (struct gendisk *disk,
1873 				      unsigned int clearing);
1874 	void (*unlock_native_capacity) (struct gendisk *);
1875 	int (*revalidate_disk) (struct gendisk *);
1876 	int (*getgeo)(struct block_device *, struct hd_geometry *);
1877 	int (*set_read_only)(struct block_device *bdev, bool ro);
1878 	/* this callback is with swap_lock and sometimes page table lock held */
1879 	void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1880 	int (*report_zones)(struct gendisk *, sector_t sector,
1881 			unsigned int nr_zones, report_zones_cb cb, void *data);
1882 	char *(*devnode)(struct gendisk *disk, umode_t *mode);
1883 	struct module *owner;
1884 	const struct pr_ops *pr_ops;
1885 };
1886 
1887 #ifdef CONFIG_COMPAT
1888 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1889 				      unsigned int, unsigned long);
1890 #else
1891 #define blkdev_compat_ptr_ioctl NULL
1892 #endif
1893 
1894 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1895 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1896 						struct writeback_control *);
1897 
1898 #ifdef CONFIG_BLK_DEV_ZONED
1899 bool blk_req_needs_zone_write_lock(struct request *rq);
1900 bool blk_req_zone_write_trylock(struct request *rq);
1901 void __blk_req_zone_write_lock(struct request *rq);
1902 void __blk_req_zone_write_unlock(struct request *rq);
1903 
1904 static inline void blk_req_zone_write_lock(struct request *rq)
1905 {
1906 	if (blk_req_needs_zone_write_lock(rq))
1907 		__blk_req_zone_write_lock(rq);
1908 }
1909 
1910 static inline void blk_req_zone_write_unlock(struct request *rq)
1911 {
1912 	if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1913 		__blk_req_zone_write_unlock(rq);
1914 }
1915 
1916 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1917 {
1918 	return rq->q->seq_zones_wlock &&
1919 		test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1920 }
1921 
1922 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1923 {
1924 	if (!blk_req_needs_zone_write_lock(rq))
1925 		return true;
1926 	return !blk_req_zone_is_write_locked(rq);
1927 }
1928 #else
1929 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1930 {
1931 	return false;
1932 }
1933 
1934 static inline void blk_req_zone_write_lock(struct request *rq)
1935 {
1936 }
1937 
1938 static inline void blk_req_zone_write_unlock(struct request *rq)
1939 {
1940 }
1941 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1942 {
1943 	return false;
1944 }
1945 
1946 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1947 {
1948 	return true;
1949 }
1950 #endif /* CONFIG_BLK_DEV_ZONED */
1951 
1952 static inline void blk_wake_io_task(struct task_struct *waiter)
1953 {
1954 	/*
1955 	 * If we're polling, the task itself is doing the completions. For
1956 	 * that case, we don't need to signal a wakeup, it's enough to just
1957 	 * mark us as RUNNING.
1958 	 */
1959 	if (waiter == current)
1960 		__set_current_state(TASK_RUNNING);
1961 	else
1962 		wake_up_process(waiter);
1963 }
1964 
1965 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1966 		unsigned int op);
1967 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1968 		unsigned long start_time);
1969 
1970 unsigned long bio_start_io_acct(struct bio *bio);
1971 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1972 		struct block_device *orig_bdev);
1973 
1974 /**
1975  * bio_end_io_acct - end I/O accounting for bio based drivers
1976  * @bio:	bio to end account for
1977  * @start:	start time returned by bio_start_io_acct()
1978  */
1979 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1980 {
1981 	return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1982 }
1983 
1984 int bdev_read_only(struct block_device *bdev);
1985 int set_blocksize(struct block_device *bdev, int size);
1986 
1987 const char *bdevname(struct block_device *bdev, char *buffer);
1988 int lookup_bdev(const char *pathname, dev_t *dev);
1989 
1990 void blkdev_show(struct seq_file *seqf, off_t offset);
1991 
1992 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
1993 #define BDEVT_SIZE	10	/* Largest string for MAJ:MIN for blkdev */
1994 #ifdef CONFIG_BLOCK
1995 #define BLKDEV_MAJOR_MAX	512
1996 #else
1997 #define BLKDEV_MAJOR_MAX	0
1998 #endif
1999 
2000 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2001 		void *holder);
2002 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
2003 int bd_prepare_to_claim(struct block_device *bdev, void *holder);
2004 void bd_abort_claiming(struct block_device *bdev, void *holder);
2005 void blkdev_put(struct block_device *bdev, fmode_t mode);
2006 
2007 /* just for blk-cgroup, don't use elsewhere */
2008 struct block_device *blkdev_get_no_open(dev_t dev);
2009 void blkdev_put_no_open(struct block_device *bdev);
2010 
2011 struct block_device *bdev_alloc(struct gendisk *disk, u8 partno);
2012 void bdev_add(struct block_device *bdev, dev_t dev);
2013 struct block_device *I_BDEV(struct inode *inode);
2014 struct block_device *bdgrab(struct block_device *bdev);
2015 void bdput(struct block_device *);
2016 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
2017 		loff_t lend);
2018 
2019 #ifdef CONFIG_BLOCK
2020 void invalidate_bdev(struct block_device *bdev);
2021 int sync_blockdev(struct block_device *bdev);
2022 #else
2023 static inline void invalidate_bdev(struct block_device *bdev)
2024 {
2025 }
2026 static inline int sync_blockdev(struct block_device *bdev)
2027 {
2028 	return 0;
2029 }
2030 #endif
2031 int fsync_bdev(struct block_device *bdev);
2032 
2033 int freeze_bdev(struct block_device *bdev);
2034 int thaw_bdev(struct block_device *bdev);
2035 
2036 #endif /* _LINUX_BLKDEV_H */
2037