xref: /linux-6.15/include/linux/blkdev.h (revision a7863b34)
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 #ifdef CONFIG_BLK_DEV_ZONED
357 
358 #define BLK_ALL_ZONES  ((unsigned int)-1)
359 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
360 			unsigned int nr_zones, report_zones_cb cb, void *data);
361 unsigned int blkdev_nr_zones(struct gendisk *disk);
362 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
363 			    sector_t sectors, sector_t nr_sectors,
364 			    gfp_t gfp_mask);
365 int blk_revalidate_disk_zones(struct gendisk *disk,
366 			      void (*update_driver_data)(struct gendisk *disk));
367 
368 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
369 				     unsigned int cmd, unsigned long arg);
370 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
371 				  unsigned int cmd, unsigned long arg);
372 
373 #else /* CONFIG_BLK_DEV_ZONED */
374 
375 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
376 {
377 	return 0;
378 }
379 
380 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
381 					    fmode_t mode, unsigned int cmd,
382 					    unsigned long arg)
383 {
384 	return -ENOTTY;
385 }
386 
387 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
388 					 fmode_t mode, unsigned int cmd,
389 					 unsigned long arg)
390 {
391 	return -ENOTTY;
392 }
393 
394 #endif /* CONFIG_BLK_DEV_ZONED */
395 
396 struct request_queue {
397 	struct request		*last_merge;
398 	struct elevator_queue	*elevator;
399 
400 	struct blk_queue_stats	*stats;
401 	struct rq_qos		*rq_qos;
402 
403 	const struct blk_mq_ops	*mq_ops;
404 
405 	/* sw queues */
406 	struct blk_mq_ctx __percpu	*queue_ctx;
407 
408 	unsigned int		queue_depth;
409 
410 	/* hw dispatch queues */
411 	struct blk_mq_hw_ctx	**queue_hw_ctx;
412 	unsigned int		nr_hw_queues;
413 
414 	struct backing_dev_info	*backing_dev_info;
415 
416 	/*
417 	 * The queue owner gets to use this for whatever they like.
418 	 * ll_rw_blk doesn't touch it.
419 	 */
420 	void			*queuedata;
421 
422 	/*
423 	 * various queue flags, see QUEUE_* below
424 	 */
425 	unsigned long		queue_flags;
426 	/*
427 	 * Number of contexts that have called blk_set_pm_only(). If this
428 	 * counter is above zero then only RQF_PM and RQF_PREEMPT requests are
429 	 * processed.
430 	 */
431 	atomic_t		pm_only;
432 
433 	/*
434 	 * ida allocated id for this queue.  Used to index queues from
435 	 * ioctx.
436 	 */
437 	int			id;
438 
439 	/*
440 	 * queue needs bounce pages for pages above this limit
441 	 */
442 	gfp_t			bounce_gfp;
443 
444 	spinlock_t		queue_lock;
445 
446 	/*
447 	 * queue kobject
448 	 */
449 	struct kobject kobj;
450 
451 	/*
452 	 * mq queue kobject
453 	 */
454 	struct kobject *mq_kobj;
455 
456 #ifdef  CONFIG_BLK_DEV_INTEGRITY
457 	struct blk_integrity integrity;
458 #endif	/* CONFIG_BLK_DEV_INTEGRITY */
459 
460 #ifdef CONFIG_PM
461 	struct device		*dev;
462 	enum rpm_status		rpm_status;
463 	unsigned int		nr_pending;
464 #endif
465 
466 	/*
467 	 * queue settings
468 	 */
469 	unsigned long		nr_requests;	/* Max # of requests */
470 
471 	unsigned int		dma_pad_mask;
472 	unsigned int		dma_alignment;
473 
474 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
475 	/* Inline crypto capabilities */
476 	struct blk_keyslot_manager *ksm;
477 #endif
478 
479 	unsigned int		rq_timeout;
480 	int			poll_nsec;
481 
482 	struct blk_stat_callback	*poll_cb;
483 	struct blk_rq_stat	poll_stat[BLK_MQ_POLL_STATS_BKTS];
484 
485 	struct timer_list	timeout;
486 	struct work_struct	timeout_work;
487 
488 	struct list_head	icq_list;
489 #ifdef CONFIG_BLK_CGROUP
490 	DECLARE_BITMAP		(blkcg_pols, BLKCG_MAX_POLS);
491 	struct blkcg_gq		*root_blkg;
492 	struct list_head	blkg_list;
493 #endif
494 
495 	struct queue_limits	limits;
496 
497 	unsigned int		required_elevator_features;
498 
499 #ifdef CONFIG_BLK_DEV_ZONED
500 	/*
501 	 * Zoned block device information for request dispatch control.
502 	 * nr_zones is the total number of zones of the device. This is always
503 	 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
504 	 * bits which indicates if a zone is conventional (bit set) or
505 	 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
506 	 * bits which indicates if a zone is write locked, that is, if a write
507 	 * request targeting the zone was dispatched. All three fields are
508 	 * initialized by the low level device driver (e.g. scsi/sd.c).
509 	 * Stacking drivers (device mappers) may or may not initialize
510 	 * these fields.
511 	 *
512 	 * Reads of this information must be protected with blk_queue_enter() /
513 	 * blk_queue_exit(). Modifying this information is only allowed while
514 	 * no requests are being processed. See also blk_mq_freeze_queue() and
515 	 * blk_mq_unfreeze_queue().
516 	 */
517 	unsigned int		nr_zones;
518 	unsigned long		*conv_zones_bitmap;
519 	unsigned long		*seq_zones_wlock;
520 	unsigned int		max_open_zones;
521 	unsigned int		max_active_zones;
522 #endif /* CONFIG_BLK_DEV_ZONED */
523 
524 	/*
525 	 * sg stuff
526 	 */
527 	unsigned int		sg_timeout;
528 	unsigned int		sg_reserved_size;
529 	int			node;
530 	struct mutex		debugfs_mutex;
531 #ifdef CONFIG_BLK_DEV_IO_TRACE
532 	struct blk_trace __rcu	*blk_trace;
533 #endif
534 	/*
535 	 * for flush operations
536 	 */
537 	struct blk_flush_queue	*fq;
538 
539 	struct list_head	requeue_list;
540 	spinlock_t		requeue_lock;
541 	struct delayed_work	requeue_work;
542 
543 	struct mutex		sysfs_lock;
544 	struct mutex		sysfs_dir_lock;
545 
546 	/*
547 	 * for reusing dead hctx instance in case of updating
548 	 * nr_hw_queues
549 	 */
550 	struct list_head	unused_hctx_list;
551 	spinlock_t		unused_hctx_lock;
552 
553 	int			mq_freeze_depth;
554 
555 #if defined(CONFIG_BLK_DEV_BSG)
556 	struct bsg_class_device bsg_dev;
557 #endif
558 
559 #ifdef CONFIG_BLK_DEV_THROTTLING
560 	/* Throttle data */
561 	struct throtl_data *td;
562 #endif
563 	struct rcu_head		rcu_head;
564 	wait_queue_head_t	mq_freeze_wq;
565 	/*
566 	 * Protect concurrent access to q_usage_counter by
567 	 * percpu_ref_kill() and percpu_ref_reinit().
568 	 */
569 	struct mutex		mq_freeze_lock;
570 	struct percpu_ref	q_usage_counter;
571 
572 	struct blk_mq_tag_set	*tag_set;
573 	struct list_head	tag_set_list;
574 	struct bio_set		bio_split;
575 
576 	struct dentry		*debugfs_dir;
577 
578 #ifdef CONFIG_BLK_DEBUG_FS
579 	struct dentry		*sched_debugfs_dir;
580 	struct dentry		*rqos_debugfs_dir;
581 #endif
582 
583 	bool			mq_sysfs_init_done;
584 
585 	size_t			cmd_size;
586 
587 #define BLK_MAX_WRITE_HINTS	5
588 	u64			write_hints[BLK_MAX_WRITE_HINTS];
589 };
590 
591 /* Keep blk_queue_flag_name[] in sync with the definitions below */
592 #define QUEUE_FLAG_STOPPED	0	/* queue is stopped */
593 #define QUEUE_FLAG_DYING	1	/* queue being torn down */
594 #define QUEUE_FLAG_NOMERGES     3	/* disable merge attempts */
595 #define QUEUE_FLAG_SAME_COMP	4	/* complete on same CPU-group */
596 #define QUEUE_FLAG_FAIL_IO	5	/* fake timeout */
597 #define QUEUE_FLAG_NONROT	6	/* non-rotational device (SSD) */
598 #define QUEUE_FLAG_VIRT		QUEUE_FLAG_NONROT /* paravirt device */
599 #define QUEUE_FLAG_IO_STAT	7	/* do disk/partitions IO accounting */
600 #define QUEUE_FLAG_DISCARD	8	/* supports DISCARD */
601 #define QUEUE_FLAG_NOXMERGES	9	/* No extended merges */
602 #define QUEUE_FLAG_ADD_RANDOM	10	/* Contributes to random pool */
603 #define QUEUE_FLAG_SECERASE	11	/* supports secure erase */
604 #define QUEUE_FLAG_SAME_FORCE	12	/* force complete on same CPU */
605 #define QUEUE_FLAG_DEAD		13	/* queue tear-down finished */
606 #define QUEUE_FLAG_INIT_DONE	14	/* queue is initialized */
607 #define QUEUE_FLAG_POLL		16	/* IO polling enabled if set */
608 #define QUEUE_FLAG_WC		17	/* Write back caching */
609 #define QUEUE_FLAG_FUA		18	/* device supports FUA writes */
610 #define QUEUE_FLAG_DAX		19	/* device supports DAX */
611 #define QUEUE_FLAG_STATS	20	/* track IO start and completion times */
612 #define QUEUE_FLAG_POLL_STATS	21	/* collecting stats for hybrid polling */
613 #define QUEUE_FLAG_REGISTERED	22	/* queue has been registered to a disk */
614 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23	/* queue supports SCSI commands */
615 #define QUEUE_FLAG_QUIESCED	24	/* queue has been quiesced */
616 #define QUEUE_FLAG_PCI_P2PDMA	25	/* device supports PCI p2p requests */
617 #define QUEUE_FLAG_ZONE_RESETALL 26	/* supports Zone Reset All */
618 #define QUEUE_FLAG_RQ_ALLOC_TIME 27	/* record rq->alloc_time_ns */
619 
620 #define QUEUE_FLAG_MQ_DEFAULT	((1 << QUEUE_FLAG_IO_STAT) |		\
621 				 (1 << QUEUE_FLAG_SAME_COMP))
622 
623 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
624 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
625 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
626 
627 #define blk_queue_stopped(q)	test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
628 #define blk_queue_dying(q)	test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
629 #define blk_queue_dead(q)	test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
630 #define blk_queue_init_done(q)	test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
631 #define blk_queue_nomerges(q)	test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
632 #define blk_queue_noxmerges(q)	\
633 	test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
634 #define blk_queue_nonrot(q)	test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
635 #define blk_queue_io_stat(q)	test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
636 #define blk_queue_add_random(q)	test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
637 #define blk_queue_discard(q)	test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
638 #define blk_queue_zone_resetall(q)	\
639 	test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
640 #define blk_queue_secure_erase(q) \
641 	(test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
642 #define blk_queue_dax(q)	test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
643 #define blk_queue_scsi_passthrough(q)	\
644 	test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
645 #define blk_queue_pci_p2pdma(q)	\
646 	test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
647 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
648 #define blk_queue_rq_alloc_time(q)	\
649 	test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
650 #else
651 #define blk_queue_rq_alloc_time(q)	false
652 #endif
653 
654 #define blk_noretry_request(rq) \
655 	((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
656 			     REQ_FAILFAST_DRIVER))
657 #define blk_queue_quiesced(q)	test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
658 #define blk_queue_pm_only(q)	atomic_read(&(q)->pm_only)
659 #define blk_queue_fua(q)	test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
660 #define blk_queue_registered(q)	test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
661 
662 extern void blk_set_pm_only(struct request_queue *q);
663 extern void blk_clear_pm_only(struct request_queue *q);
664 
665 static inline bool blk_account_rq(struct request *rq)
666 {
667 	return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
668 }
669 
670 #define list_entry_rq(ptr)	list_entry((ptr), struct request, queuelist)
671 
672 #define rq_data_dir(rq)		(op_is_write(req_op(rq)) ? WRITE : READ)
673 
674 #define rq_dma_dir(rq) \
675 	(op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
676 
677 #define dma_map_bvec(dev, bv, dir, attrs) \
678 	dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
679 	(dir), (attrs))
680 
681 static inline bool queue_is_mq(struct request_queue *q)
682 {
683 	return q->mq_ops;
684 }
685 
686 static inline enum blk_zoned_model
687 blk_queue_zoned_model(struct request_queue *q)
688 {
689 	return q->limits.zoned;
690 }
691 
692 static inline bool blk_queue_is_zoned(struct request_queue *q)
693 {
694 	switch (blk_queue_zoned_model(q)) {
695 	case BLK_ZONED_HA:
696 	case BLK_ZONED_HM:
697 		return true;
698 	default:
699 		return false;
700 	}
701 }
702 
703 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
704 {
705 	return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
706 }
707 
708 #ifdef CONFIG_BLK_DEV_ZONED
709 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
710 {
711 	return blk_queue_is_zoned(q) ? q->nr_zones : 0;
712 }
713 
714 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
715 					     sector_t sector)
716 {
717 	if (!blk_queue_is_zoned(q))
718 		return 0;
719 	return sector >> ilog2(q->limits.chunk_sectors);
720 }
721 
722 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
723 					 sector_t sector)
724 {
725 	if (!blk_queue_is_zoned(q))
726 		return false;
727 	if (!q->conv_zones_bitmap)
728 		return true;
729 	return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
730 }
731 
732 static inline void blk_queue_max_open_zones(struct request_queue *q,
733 		unsigned int max_open_zones)
734 {
735 	q->max_open_zones = max_open_zones;
736 }
737 
738 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
739 {
740 	return q->max_open_zones;
741 }
742 
743 static inline void blk_queue_max_active_zones(struct request_queue *q,
744 		unsigned int max_active_zones)
745 {
746 	q->max_active_zones = max_active_zones;
747 }
748 
749 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
750 {
751 	return q->max_active_zones;
752 }
753 #else /* CONFIG_BLK_DEV_ZONED */
754 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
755 {
756 	return 0;
757 }
758 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
759 					 sector_t sector)
760 {
761 	return false;
762 }
763 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
764 					     sector_t sector)
765 {
766 	return 0;
767 }
768 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
769 {
770 	return 0;
771 }
772 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
773 {
774 	return 0;
775 }
776 #endif /* CONFIG_BLK_DEV_ZONED */
777 
778 static inline bool rq_is_sync(struct request *rq)
779 {
780 	return op_is_sync(rq->cmd_flags);
781 }
782 
783 static inline bool rq_mergeable(struct request *rq)
784 {
785 	if (blk_rq_is_passthrough(rq))
786 		return false;
787 
788 	if (req_op(rq) == REQ_OP_FLUSH)
789 		return false;
790 
791 	if (req_op(rq) == REQ_OP_WRITE_ZEROES)
792 		return false;
793 
794 	if (req_op(rq) == REQ_OP_ZONE_APPEND)
795 		return false;
796 
797 	if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
798 		return false;
799 	if (rq->rq_flags & RQF_NOMERGE_FLAGS)
800 		return false;
801 
802 	return true;
803 }
804 
805 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
806 {
807 	if (bio_page(a) == bio_page(b) &&
808 	    bio_offset(a) == bio_offset(b))
809 		return true;
810 
811 	return false;
812 }
813 
814 static inline unsigned int blk_queue_depth(struct request_queue *q)
815 {
816 	if (q->queue_depth)
817 		return q->queue_depth;
818 
819 	return q->nr_requests;
820 }
821 
822 extern unsigned long blk_max_low_pfn, blk_max_pfn;
823 
824 /*
825  * standard bounce addresses:
826  *
827  * BLK_BOUNCE_HIGH	: bounce all highmem pages
828  * BLK_BOUNCE_ANY	: don't bounce anything
829  * BLK_BOUNCE_ISA	: bounce pages above ISA DMA boundary
830  */
831 
832 #if BITS_PER_LONG == 32
833 #define BLK_BOUNCE_HIGH		((u64)blk_max_low_pfn << PAGE_SHIFT)
834 #else
835 #define BLK_BOUNCE_HIGH		-1ULL
836 #endif
837 #define BLK_BOUNCE_ANY		(-1ULL)
838 #define BLK_BOUNCE_ISA		(DMA_BIT_MASK(24))
839 
840 /*
841  * default timeout for SG_IO if none specified
842  */
843 #define BLK_DEFAULT_SG_TIMEOUT	(60 * HZ)
844 #define BLK_MIN_SG_TIMEOUT	(7 * HZ)
845 
846 struct rq_map_data {
847 	struct page **pages;
848 	int page_order;
849 	int nr_entries;
850 	unsigned long offset;
851 	int null_mapped;
852 	int from_user;
853 };
854 
855 struct req_iterator {
856 	struct bvec_iter iter;
857 	struct bio *bio;
858 };
859 
860 /* This should not be used directly - use rq_for_each_segment */
861 #define for_each_bio(_bio)		\
862 	for (; _bio; _bio = _bio->bi_next)
863 #define __rq_for_each_bio(_bio, rq)	\
864 	if ((rq->bio))			\
865 		for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
866 
867 #define rq_for_each_segment(bvl, _rq, _iter)			\
868 	__rq_for_each_bio(_iter.bio, _rq)			\
869 		bio_for_each_segment(bvl, _iter.bio, _iter.iter)
870 
871 #define rq_for_each_bvec(bvl, _rq, _iter)			\
872 	__rq_for_each_bio(_iter.bio, _rq)			\
873 		bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
874 
875 #define rq_iter_last(bvec, _iter)				\
876 		(_iter.bio->bi_next == NULL &&			\
877 		 bio_iter_last(bvec, _iter.iter))
878 
879 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
880 # error	"You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
881 #endif
882 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
883 extern void rq_flush_dcache_pages(struct request *rq);
884 #else
885 static inline void rq_flush_dcache_pages(struct request *rq)
886 {
887 }
888 #endif
889 
890 extern int blk_register_queue(struct gendisk *disk);
891 extern void blk_unregister_queue(struct gendisk *disk);
892 blk_qc_t submit_bio_noacct(struct bio *bio);
893 extern void blk_rq_init(struct request_queue *q, struct request *rq);
894 extern void blk_put_request(struct request *);
895 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
896 				       blk_mq_req_flags_t flags);
897 extern int blk_lld_busy(struct request_queue *q);
898 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
899 			     struct bio_set *bs, gfp_t gfp_mask,
900 			     int (*bio_ctr)(struct bio *, struct bio *, void *),
901 			     void *data);
902 extern void blk_rq_unprep_clone(struct request *rq);
903 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
904 				     struct request *rq);
905 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
906 extern void blk_queue_split(struct bio **);
907 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
908 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
909 			      unsigned int, void __user *);
910 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
911 			  unsigned int, void __user *);
912 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
913 			 struct scsi_ioctl_command __user *);
914 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
915 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
916 
917 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
918 extern void blk_queue_exit(struct request_queue *q);
919 extern void blk_sync_queue(struct request_queue *q);
920 extern int blk_rq_map_user(struct request_queue *, struct request *,
921 			   struct rq_map_data *, void __user *, unsigned long,
922 			   gfp_t);
923 extern int blk_rq_unmap_user(struct bio *);
924 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
925 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
926 			       struct rq_map_data *, const struct iov_iter *,
927 			       gfp_t);
928 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
929 			  struct request *, int);
930 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
931 				  struct request *, int, rq_end_io_fn *);
932 
933 /* Helper to convert REQ_OP_XXX to its string format XXX */
934 extern const char *blk_op_str(unsigned int op);
935 
936 int blk_status_to_errno(blk_status_t status);
937 blk_status_t errno_to_blk_status(int errno);
938 
939 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
940 
941 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
942 {
943 	return bdev->bd_disk->queue;	/* this is never NULL */
944 }
945 
946 /*
947  * The basic unit of block I/O is a sector. It is used in a number of contexts
948  * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
949  * bytes. Variables of type sector_t represent an offset or size that is a
950  * multiple of 512 bytes. Hence these two constants.
951  */
952 #ifndef SECTOR_SHIFT
953 #define SECTOR_SHIFT 9
954 #endif
955 #ifndef SECTOR_SIZE
956 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
957 #endif
958 
959 /*
960  * blk_rq_pos()			: the current sector
961  * blk_rq_bytes()		: bytes left in the entire request
962  * blk_rq_cur_bytes()		: bytes left in the current segment
963  * blk_rq_err_bytes()		: bytes left till the next error boundary
964  * blk_rq_sectors()		: sectors left in the entire request
965  * blk_rq_cur_sectors()		: sectors left in the current segment
966  * blk_rq_stats_sectors()	: sectors of the entire request used for stats
967  */
968 static inline sector_t blk_rq_pos(const struct request *rq)
969 {
970 	return rq->__sector;
971 }
972 
973 static inline unsigned int blk_rq_bytes(const struct request *rq)
974 {
975 	return rq->__data_len;
976 }
977 
978 static inline int blk_rq_cur_bytes(const struct request *rq)
979 {
980 	return rq->bio ? bio_cur_bytes(rq->bio) : 0;
981 }
982 
983 extern unsigned int blk_rq_err_bytes(const struct request *rq);
984 
985 static inline unsigned int blk_rq_sectors(const struct request *rq)
986 {
987 	return blk_rq_bytes(rq) >> SECTOR_SHIFT;
988 }
989 
990 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
991 {
992 	return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
993 }
994 
995 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
996 {
997 	return rq->stats_sectors;
998 }
999 
1000 #ifdef CONFIG_BLK_DEV_ZONED
1001 
1002 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
1003 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
1004 
1005 static inline unsigned int blk_rq_zone_no(struct request *rq)
1006 {
1007 	return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1008 }
1009 
1010 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1011 {
1012 	return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1013 }
1014 #endif /* CONFIG_BLK_DEV_ZONED */
1015 
1016 /*
1017  * Some commands like WRITE SAME have a payload or data transfer size which
1018  * is different from the size of the request.  Any driver that supports such
1019  * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1020  * calculate the data transfer size.
1021  */
1022 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1023 {
1024 	if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1025 		return rq->special_vec.bv_len;
1026 	return blk_rq_bytes(rq);
1027 }
1028 
1029 /*
1030  * Return the first full biovec in the request.  The caller needs to check that
1031  * there are any bvecs before calling this helper.
1032  */
1033 static inline struct bio_vec req_bvec(struct request *rq)
1034 {
1035 	if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1036 		return rq->special_vec;
1037 	return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1038 }
1039 
1040 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1041 						     int op)
1042 {
1043 	if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1044 		return min(q->limits.max_discard_sectors,
1045 			   UINT_MAX >> SECTOR_SHIFT);
1046 
1047 	if (unlikely(op == REQ_OP_WRITE_SAME))
1048 		return q->limits.max_write_same_sectors;
1049 
1050 	if (unlikely(op == REQ_OP_WRITE_ZEROES))
1051 		return q->limits.max_write_zeroes_sectors;
1052 
1053 	return q->limits.max_sectors;
1054 }
1055 
1056 /*
1057  * Return maximum size of a request at given offset. Only valid for
1058  * file system requests.
1059  */
1060 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1061 					       sector_t offset)
1062 {
1063 	if (!q->limits.chunk_sectors)
1064 		return q->limits.max_sectors;
1065 
1066 	return min(q->limits.max_sectors, (unsigned int)(q->limits.chunk_sectors -
1067 			(offset & (q->limits.chunk_sectors - 1))));
1068 }
1069 
1070 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1071 						  sector_t offset)
1072 {
1073 	struct request_queue *q = rq->q;
1074 
1075 	if (blk_rq_is_passthrough(rq))
1076 		return q->limits.max_hw_sectors;
1077 
1078 	if (!q->limits.chunk_sectors ||
1079 	    req_op(rq) == REQ_OP_DISCARD ||
1080 	    req_op(rq) == REQ_OP_SECURE_ERASE)
1081 		return blk_queue_get_max_sectors(q, req_op(rq));
1082 
1083 	return min(blk_max_size_offset(q, offset),
1084 			blk_queue_get_max_sectors(q, req_op(rq)));
1085 }
1086 
1087 static inline unsigned int blk_rq_count_bios(struct request *rq)
1088 {
1089 	unsigned int nr_bios = 0;
1090 	struct bio *bio;
1091 
1092 	__rq_for_each_bio(bio, rq)
1093 		nr_bios++;
1094 
1095 	return nr_bios;
1096 }
1097 
1098 void blk_steal_bios(struct bio_list *list, struct request *rq);
1099 
1100 /*
1101  * Request completion related functions.
1102  *
1103  * blk_update_request() completes given number of bytes and updates
1104  * the request without completing it.
1105  */
1106 extern bool blk_update_request(struct request *rq, blk_status_t error,
1107 			       unsigned int nr_bytes);
1108 
1109 extern void blk_abort_request(struct request *);
1110 
1111 /*
1112  * Access functions for manipulating queue properties
1113  */
1114 extern void blk_cleanup_queue(struct request_queue *);
1115 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1116 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1117 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1118 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1119 extern void blk_queue_max_discard_segments(struct request_queue *,
1120 		unsigned short);
1121 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1122 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1123 		unsigned int max_discard_sectors);
1124 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1125 		unsigned int max_write_same_sectors);
1126 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1127 		unsigned int max_write_same_sectors);
1128 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1129 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1130 		unsigned int max_zone_append_sectors);
1131 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1132 extern void blk_queue_alignment_offset(struct request_queue *q,
1133 				       unsigned int alignment);
1134 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1135 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1136 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1137 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1138 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1139 extern void blk_set_default_limits(struct queue_limits *lim);
1140 extern void blk_set_stacking_limits(struct queue_limits *lim);
1141 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1142 			    sector_t offset);
1143 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1144 			      sector_t offset);
1145 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1146 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1147 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1148 extern void blk_queue_dma_alignment(struct request_queue *, int);
1149 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1150 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1151 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1152 extern void blk_queue_required_elevator_features(struct request_queue *q,
1153 						 unsigned int features);
1154 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1155 					      struct device *dev);
1156 
1157 /*
1158  * Number of physical segments as sent to the device.
1159  *
1160  * Normally this is the number of discontiguous data segments sent by the
1161  * submitter.  But for data-less command like discard we might have no
1162  * actual data segments submitted, but the driver might have to add it's
1163  * own special payload.  In that case we still return 1 here so that this
1164  * special payload will be mapped.
1165  */
1166 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1167 {
1168 	if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1169 		return 1;
1170 	return rq->nr_phys_segments;
1171 }
1172 
1173 /*
1174  * Number of discard segments (or ranges) the driver needs to fill in.
1175  * Each discard bio merged into a request is counted as one segment.
1176  */
1177 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1178 {
1179 	return max_t(unsigned short, rq->nr_phys_segments, 1);
1180 }
1181 
1182 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1183 		struct scatterlist *sglist, struct scatterlist **last_sg);
1184 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1185 		struct scatterlist *sglist)
1186 {
1187 	struct scatterlist *last_sg = NULL;
1188 
1189 	return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1190 }
1191 extern void blk_dump_rq_flags(struct request *, char *);
1192 
1193 bool __must_check blk_get_queue(struct request_queue *);
1194 struct request_queue *blk_alloc_queue(int node_id);
1195 extern void blk_put_queue(struct request_queue *);
1196 extern void blk_set_queue_dying(struct request_queue *);
1197 
1198 #ifdef CONFIG_BLOCK
1199 /*
1200  * blk_plug permits building a queue of related requests by holding the I/O
1201  * fragments for a short period. This allows merging of sequential requests
1202  * into single larger request. As the requests are moved from a per-task list to
1203  * the device's request_queue in a batch, this results in improved scalability
1204  * as the lock contention for request_queue lock is reduced.
1205  *
1206  * It is ok not to disable preemption when adding the request to the plug list
1207  * or when attempting a merge, because blk_schedule_flush_list() will only flush
1208  * the plug list when the task sleeps by itself. For details, please see
1209  * schedule() where blk_schedule_flush_plug() is called.
1210  */
1211 struct blk_plug {
1212 	struct list_head mq_list; /* blk-mq requests */
1213 	struct list_head cb_list; /* md requires an unplug callback */
1214 	unsigned short rq_count;
1215 	bool multiple_queues;
1216 	bool nowait;
1217 };
1218 #define BLK_MAX_REQUEST_COUNT 16
1219 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1220 
1221 struct blk_plug_cb;
1222 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1223 struct blk_plug_cb {
1224 	struct list_head list;
1225 	blk_plug_cb_fn callback;
1226 	void *data;
1227 };
1228 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1229 					     void *data, int size);
1230 extern void blk_start_plug(struct blk_plug *);
1231 extern void blk_finish_plug(struct blk_plug *);
1232 extern void blk_flush_plug_list(struct blk_plug *, bool);
1233 
1234 static inline void blk_flush_plug(struct task_struct *tsk)
1235 {
1236 	struct blk_plug *plug = tsk->plug;
1237 
1238 	if (plug)
1239 		blk_flush_plug_list(plug, false);
1240 }
1241 
1242 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1243 {
1244 	struct blk_plug *plug = tsk->plug;
1245 
1246 	if (plug)
1247 		blk_flush_plug_list(plug, true);
1248 }
1249 
1250 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1251 {
1252 	struct blk_plug *plug = tsk->plug;
1253 
1254 	return plug &&
1255 		 (!list_empty(&plug->mq_list) ||
1256 		 !list_empty(&plug->cb_list));
1257 }
1258 
1259 int blkdev_issue_flush(struct block_device *, gfp_t);
1260 long nr_blockdev_pages(void);
1261 #else /* CONFIG_BLOCK */
1262 struct blk_plug {
1263 };
1264 
1265 static inline void blk_start_plug(struct blk_plug *plug)
1266 {
1267 }
1268 
1269 static inline void blk_finish_plug(struct blk_plug *plug)
1270 {
1271 }
1272 
1273 static inline void blk_flush_plug(struct task_struct *task)
1274 {
1275 }
1276 
1277 static inline void blk_schedule_flush_plug(struct task_struct *task)
1278 {
1279 }
1280 
1281 
1282 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1283 {
1284 	return false;
1285 }
1286 
1287 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask)
1288 {
1289 	return 0;
1290 }
1291 
1292 static inline long nr_blockdev_pages(void)
1293 {
1294 	return 0;
1295 }
1296 #endif /* CONFIG_BLOCK */
1297 
1298 extern void blk_io_schedule(void);
1299 
1300 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1301 		sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1302 
1303 #define BLKDEV_DISCARD_SECURE	(1 << 0)	/* issue a secure erase */
1304 
1305 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1306 		sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1307 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1308 		sector_t nr_sects, gfp_t gfp_mask, int flags,
1309 		struct bio **biop);
1310 
1311 #define BLKDEV_ZERO_NOUNMAP	(1 << 0)  /* do not free blocks */
1312 #define BLKDEV_ZERO_NOFALLBACK	(1 << 1)  /* don't write explicit zeroes */
1313 
1314 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1315 		sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1316 		unsigned flags);
1317 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1318 		sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1319 
1320 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1321 		sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1322 {
1323 	return blkdev_issue_discard(sb->s_bdev,
1324 				    block << (sb->s_blocksize_bits -
1325 					      SECTOR_SHIFT),
1326 				    nr_blocks << (sb->s_blocksize_bits -
1327 						  SECTOR_SHIFT),
1328 				    gfp_mask, flags);
1329 }
1330 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1331 		sector_t nr_blocks, gfp_t gfp_mask)
1332 {
1333 	return blkdev_issue_zeroout(sb->s_bdev,
1334 				    block << (sb->s_blocksize_bits -
1335 					      SECTOR_SHIFT),
1336 				    nr_blocks << (sb->s_blocksize_bits -
1337 						  SECTOR_SHIFT),
1338 				    gfp_mask, 0);
1339 }
1340 
1341 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1342 
1343 enum blk_default_limits {
1344 	BLK_MAX_SEGMENTS	= 128,
1345 	BLK_SAFE_MAX_SECTORS	= 255,
1346 	BLK_DEF_MAX_SECTORS	= 2560,
1347 	BLK_MAX_SEGMENT_SIZE	= 65536,
1348 	BLK_SEG_BOUNDARY_MASK	= 0xFFFFFFFFUL,
1349 };
1350 
1351 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1352 {
1353 	return q->limits.seg_boundary_mask;
1354 }
1355 
1356 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1357 {
1358 	return q->limits.virt_boundary_mask;
1359 }
1360 
1361 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1362 {
1363 	return q->limits.max_sectors;
1364 }
1365 
1366 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1367 {
1368 	return q->limits.max_hw_sectors;
1369 }
1370 
1371 static inline unsigned short queue_max_segments(const struct request_queue *q)
1372 {
1373 	return q->limits.max_segments;
1374 }
1375 
1376 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1377 {
1378 	return q->limits.max_discard_segments;
1379 }
1380 
1381 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1382 {
1383 	return q->limits.max_segment_size;
1384 }
1385 
1386 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1387 {
1388 	return q->limits.max_zone_append_sectors;
1389 }
1390 
1391 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1392 {
1393 	int retval = 512;
1394 
1395 	if (q && q->limits.logical_block_size)
1396 		retval = q->limits.logical_block_size;
1397 
1398 	return retval;
1399 }
1400 
1401 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1402 {
1403 	return queue_logical_block_size(bdev_get_queue(bdev));
1404 }
1405 
1406 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1407 {
1408 	return q->limits.physical_block_size;
1409 }
1410 
1411 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1412 {
1413 	return queue_physical_block_size(bdev_get_queue(bdev));
1414 }
1415 
1416 static inline unsigned int queue_io_min(const struct request_queue *q)
1417 {
1418 	return q->limits.io_min;
1419 }
1420 
1421 static inline int bdev_io_min(struct block_device *bdev)
1422 {
1423 	return queue_io_min(bdev_get_queue(bdev));
1424 }
1425 
1426 static inline unsigned int queue_io_opt(const struct request_queue *q)
1427 {
1428 	return q->limits.io_opt;
1429 }
1430 
1431 static inline int bdev_io_opt(struct block_device *bdev)
1432 {
1433 	return queue_io_opt(bdev_get_queue(bdev));
1434 }
1435 
1436 static inline int queue_alignment_offset(const struct request_queue *q)
1437 {
1438 	if (q->limits.misaligned)
1439 		return -1;
1440 
1441 	return q->limits.alignment_offset;
1442 }
1443 
1444 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1445 {
1446 	unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1447 	unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1448 		<< SECTOR_SHIFT;
1449 
1450 	return (granularity + lim->alignment_offset - alignment) % granularity;
1451 }
1452 
1453 static inline int bdev_alignment_offset(struct block_device *bdev)
1454 {
1455 	struct request_queue *q = bdev_get_queue(bdev);
1456 
1457 	if (q->limits.misaligned)
1458 		return -1;
1459 	if (bdev != bdev->bd_contains)
1460 		return queue_limit_alignment_offset(&q->limits,
1461 				bdev->bd_part->start_sect);
1462 	return q->limits.alignment_offset;
1463 }
1464 
1465 static inline int queue_discard_alignment(const struct request_queue *q)
1466 {
1467 	if (q->limits.discard_misaligned)
1468 		return -1;
1469 
1470 	return q->limits.discard_alignment;
1471 }
1472 
1473 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1474 {
1475 	unsigned int alignment, granularity, offset;
1476 
1477 	if (!lim->max_discard_sectors)
1478 		return 0;
1479 
1480 	/* Why are these in bytes, not sectors? */
1481 	alignment = lim->discard_alignment >> SECTOR_SHIFT;
1482 	granularity = lim->discard_granularity >> SECTOR_SHIFT;
1483 	if (!granularity)
1484 		return 0;
1485 
1486 	/* Offset of the partition start in 'granularity' sectors */
1487 	offset = sector_div(sector, granularity);
1488 
1489 	/* And why do we do this modulus *again* in blkdev_issue_discard()? */
1490 	offset = (granularity + alignment - offset) % granularity;
1491 
1492 	/* Turn it back into bytes, gaah */
1493 	return offset << SECTOR_SHIFT;
1494 }
1495 
1496 static inline int bdev_discard_alignment(struct block_device *bdev)
1497 {
1498 	struct request_queue *q = bdev_get_queue(bdev);
1499 
1500 	if (bdev != bdev->bd_contains)
1501 		return queue_limit_discard_alignment(&q->limits,
1502 				bdev->bd_part->start_sect);
1503 	return q->limits.discard_alignment;
1504 }
1505 
1506 static inline unsigned int bdev_write_same(struct block_device *bdev)
1507 {
1508 	struct request_queue *q = bdev_get_queue(bdev);
1509 
1510 	if (q)
1511 		return q->limits.max_write_same_sectors;
1512 
1513 	return 0;
1514 }
1515 
1516 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1517 {
1518 	struct request_queue *q = bdev_get_queue(bdev);
1519 
1520 	if (q)
1521 		return q->limits.max_write_zeroes_sectors;
1522 
1523 	return 0;
1524 }
1525 
1526 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1527 {
1528 	struct request_queue *q = bdev_get_queue(bdev);
1529 
1530 	if (q)
1531 		return blk_queue_zoned_model(q);
1532 
1533 	return BLK_ZONED_NONE;
1534 }
1535 
1536 static inline bool bdev_is_zoned(struct block_device *bdev)
1537 {
1538 	struct request_queue *q = bdev_get_queue(bdev);
1539 
1540 	if (q)
1541 		return blk_queue_is_zoned(q);
1542 
1543 	return false;
1544 }
1545 
1546 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1547 {
1548 	struct request_queue *q = bdev_get_queue(bdev);
1549 
1550 	if (q)
1551 		return blk_queue_zone_sectors(q);
1552 	return 0;
1553 }
1554 
1555 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1556 {
1557 	struct request_queue *q = bdev_get_queue(bdev);
1558 
1559 	if (q)
1560 		return queue_max_open_zones(q);
1561 	return 0;
1562 }
1563 
1564 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1565 {
1566 	struct request_queue *q = bdev_get_queue(bdev);
1567 
1568 	if (q)
1569 		return queue_max_active_zones(q);
1570 	return 0;
1571 }
1572 
1573 static inline int queue_dma_alignment(const struct request_queue *q)
1574 {
1575 	return q ? q->dma_alignment : 511;
1576 }
1577 
1578 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1579 				 unsigned int len)
1580 {
1581 	unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1582 	return !(addr & alignment) && !(len & alignment);
1583 }
1584 
1585 /* assumes size > 256 */
1586 static inline unsigned int blksize_bits(unsigned int size)
1587 {
1588 	unsigned int bits = 8;
1589 	do {
1590 		bits++;
1591 		size >>= 1;
1592 	} while (size > 256);
1593 	return bits;
1594 }
1595 
1596 static inline unsigned int block_size(struct block_device *bdev)
1597 {
1598 	return 1 << bdev->bd_inode->i_blkbits;
1599 }
1600 
1601 int kblockd_schedule_work(struct work_struct *work);
1602 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1603 
1604 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1605 	MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1606 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1607 	MODULE_ALIAS("block-major-" __stringify(major) "-*")
1608 
1609 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1610 
1611 enum blk_integrity_flags {
1612 	BLK_INTEGRITY_VERIFY		= 1 << 0,
1613 	BLK_INTEGRITY_GENERATE		= 1 << 1,
1614 	BLK_INTEGRITY_DEVICE_CAPABLE	= 1 << 2,
1615 	BLK_INTEGRITY_IP_CHECKSUM	= 1 << 3,
1616 };
1617 
1618 struct blk_integrity_iter {
1619 	void			*prot_buf;
1620 	void			*data_buf;
1621 	sector_t		seed;
1622 	unsigned int		data_size;
1623 	unsigned short		interval;
1624 	const char		*disk_name;
1625 };
1626 
1627 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1628 typedef void (integrity_prepare_fn) (struct request *);
1629 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1630 
1631 struct blk_integrity_profile {
1632 	integrity_processing_fn		*generate_fn;
1633 	integrity_processing_fn		*verify_fn;
1634 	integrity_prepare_fn		*prepare_fn;
1635 	integrity_complete_fn		*complete_fn;
1636 	const char			*name;
1637 };
1638 
1639 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1640 extern void blk_integrity_unregister(struct gendisk *);
1641 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1642 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1643 				   struct scatterlist *);
1644 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1645 extern bool blk_integrity_merge_rq(struct request_queue *, struct request *,
1646 				   struct request *);
1647 extern bool blk_integrity_merge_bio(struct request_queue *, struct request *,
1648 				    struct bio *);
1649 
1650 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1651 {
1652 	struct blk_integrity *bi = &disk->queue->integrity;
1653 
1654 	if (!bi->profile)
1655 		return NULL;
1656 
1657 	return bi;
1658 }
1659 
1660 static inline
1661 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1662 {
1663 	return blk_get_integrity(bdev->bd_disk);
1664 }
1665 
1666 static inline bool
1667 blk_integrity_queue_supports_integrity(struct request_queue *q)
1668 {
1669 	return q->integrity.profile;
1670 }
1671 
1672 static inline bool blk_integrity_rq(struct request *rq)
1673 {
1674 	return rq->cmd_flags & REQ_INTEGRITY;
1675 }
1676 
1677 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1678 						    unsigned int segs)
1679 {
1680 	q->limits.max_integrity_segments = segs;
1681 }
1682 
1683 static inline unsigned short
1684 queue_max_integrity_segments(const struct request_queue *q)
1685 {
1686 	return q->limits.max_integrity_segments;
1687 }
1688 
1689 /**
1690  * bio_integrity_intervals - Return number of integrity intervals for a bio
1691  * @bi:		blk_integrity profile for device
1692  * @sectors:	Size of the bio in 512-byte sectors
1693  *
1694  * Description: The block layer calculates everything in 512 byte
1695  * sectors but integrity metadata is done in terms of the data integrity
1696  * interval size of the storage device.  Convert the block layer sectors
1697  * to the appropriate number of integrity intervals.
1698  */
1699 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1700 						   unsigned int sectors)
1701 {
1702 	return sectors >> (bi->interval_exp - 9);
1703 }
1704 
1705 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1706 					       unsigned int sectors)
1707 {
1708 	return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1709 }
1710 
1711 /*
1712  * Return the first bvec that contains integrity data.  Only drivers that are
1713  * limited to a single integrity segment should use this helper.
1714  */
1715 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1716 {
1717 	if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1718 		return NULL;
1719 	return rq->bio->bi_integrity->bip_vec;
1720 }
1721 
1722 #else /* CONFIG_BLK_DEV_INTEGRITY */
1723 
1724 struct bio;
1725 struct block_device;
1726 struct gendisk;
1727 struct blk_integrity;
1728 
1729 static inline int blk_integrity_rq(struct request *rq)
1730 {
1731 	return 0;
1732 }
1733 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1734 					    struct bio *b)
1735 {
1736 	return 0;
1737 }
1738 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1739 					  struct bio *b,
1740 					  struct scatterlist *s)
1741 {
1742 	return 0;
1743 }
1744 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1745 {
1746 	return NULL;
1747 }
1748 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1749 {
1750 	return NULL;
1751 }
1752 static inline bool
1753 blk_integrity_queue_supports_integrity(struct request_queue *q)
1754 {
1755 	return false;
1756 }
1757 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1758 {
1759 	return 0;
1760 }
1761 static inline void blk_integrity_register(struct gendisk *d,
1762 					 struct blk_integrity *b)
1763 {
1764 }
1765 static inline void blk_integrity_unregister(struct gendisk *d)
1766 {
1767 }
1768 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1769 						    unsigned int segs)
1770 {
1771 }
1772 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1773 {
1774 	return 0;
1775 }
1776 static inline bool blk_integrity_merge_rq(struct request_queue *rq,
1777 					  struct request *r1,
1778 					  struct request *r2)
1779 {
1780 	return true;
1781 }
1782 static inline bool blk_integrity_merge_bio(struct request_queue *rq,
1783 					   struct request *r,
1784 					   struct bio *b)
1785 {
1786 	return true;
1787 }
1788 
1789 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1790 						   unsigned int sectors)
1791 {
1792 	return 0;
1793 }
1794 
1795 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1796 					       unsigned int sectors)
1797 {
1798 	return 0;
1799 }
1800 
1801 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1802 {
1803 	return NULL;
1804 }
1805 
1806 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1807 
1808 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1809 
1810 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1811 
1812 void blk_ksm_unregister(struct request_queue *q);
1813 
1814 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1815 
1816 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1817 				    struct request_queue *q)
1818 {
1819 	return true;
1820 }
1821 
1822 static inline void blk_ksm_unregister(struct request_queue *q) { }
1823 
1824 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1825 
1826 
1827 struct block_device_operations {
1828 	blk_qc_t (*submit_bio) (struct bio *bio);
1829 	int (*open) (struct block_device *, fmode_t);
1830 	void (*release) (struct gendisk *, fmode_t);
1831 	int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1832 	int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1833 	int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1834 	unsigned int (*check_events) (struct gendisk *disk,
1835 				      unsigned int clearing);
1836 	void (*unlock_native_capacity) (struct gendisk *);
1837 	int (*revalidate_disk) (struct gendisk *);
1838 	int (*getgeo)(struct block_device *, struct hd_geometry *);
1839 	/* this callback is with swap_lock and sometimes page table lock held */
1840 	void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1841 	int (*report_zones)(struct gendisk *, sector_t sector,
1842 			unsigned int nr_zones, report_zones_cb cb, void *data);
1843 	char *(*devnode)(struct gendisk *disk, umode_t *mode);
1844 	struct module *owner;
1845 	const struct pr_ops *pr_ops;
1846 };
1847 
1848 #ifdef CONFIG_COMPAT
1849 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1850 				      unsigned int, unsigned long);
1851 #else
1852 #define blkdev_compat_ptr_ioctl NULL
1853 #endif
1854 
1855 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1856 				 unsigned long);
1857 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1858 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1859 						struct writeback_control *);
1860 
1861 #ifdef CONFIG_BLK_DEV_ZONED
1862 bool blk_req_needs_zone_write_lock(struct request *rq);
1863 bool blk_req_zone_write_trylock(struct request *rq);
1864 void __blk_req_zone_write_lock(struct request *rq);
1865 void __blk_req_zone_write_unlock(struct request *rq);
1866 
1867 static inline void blk_req_zone_write_lock(struct request *rq)
1868 {
1869 	if (blk_req_needs_zone_write_lock(rq))
1870 		__blk_req_zone_write_lock(rq);
1871 }
1872 
1873 static inline void blk_req_zone_write_unlock(struct request *rq)
1874 {
1875 	if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1876 		__blk_req_zone_write_unlock(rq);
1877 }
1878 
1879 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1880 {
1881 	return rq->q->seq_zones_wlock &&
1882 		test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1883 }
1884 
1885 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1886 {
1887 	if (!blk_req_needs_zone_write_lock(rq))
1888 		return true;
1889 	return !blk_req_zone_is_write_locked(rq);
1890 }
1891 #else
1892 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1893 {
1894 	return false;
1895 }
1896 
1897 static inline void blk_req_zone_write_lock(struct request *rq)
1898 {
1899 }
1900 
1901 static inline void blk_req_zone_write_unlock(struct request *rq)
1902 {
1903 }
1904 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1905 {
1906 	return false;
1907 }
1908 
1909 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1910 {
1911 	return true;
1912 }
1913 #endif /* CONFIG_BLK_DEV_ZONED */
1914 
1915 static inline void blk_wake_io_task(struct task_struct *waiter)
1916 {
1917 	/*
1918 	 * If we're polling, the task itself is doing the completions. For
1919 	 * that case, we don't need to signal a wakeup, it's enough to just
1920 	 * mark us as RUNNING.
1921 	 */
1922 	if (waiter == current)
1923 		__set_current_state(TASK_RUNNING);
1924 	else
1925 		wake_up_process(waiter);
1926 }
1927 
1928 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1929 		unsigned int op);
1930 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1931 		unsigned long start_time);
1932 
1933 /**
1934  * bio_start_io_acct - start I/O accounting for bio based drivers
1935  * @bio:	bio to start account for
1936  *
1937  * Returns the start time that should be passed back to bio_end_io_acct().
1938  */
1939 static inline unsigned long bio_start_io_acct(struct bio *bio)
1940 {
1941 	return disk_start_io_acct(bio->bi_disk, bio_sectors(bio), bio_op(bio));
1942 }
1943 
1944 /**
1945  * bio_end_io_acct - end I/O accounting for bio based drivers
1946  * @bio:	bio to end account for
1947  * @start:	start time returned by bio_start_io_acct()
1948  */
1949 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1950 {
1951 	return disk_end_io_acct(bio->bi_disk, bio_op(bio), start_time);
1952 }
1953 
1954 int bdev_read_only(struct block_device *bdev);
1955 int set_blocksize(struct block_device *bdev, int size);
1956 
1957 const char *bdevname(struct block_device *bdev, char *buffer);
1958 struct block_device *lookup_bdev(const char *);
1959 
1960 void blkdev_show(struct seq_file *seqf, off_t offset);
1961 
1962 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
1963 #define BDEVT_SIZE	10	/* Largest string for MAJ:MIN for blkdev */
1964 #ifdef CONFIG_BLOCK
1965 #define BLKDEV_MAJOR_MAX	512
1966 #else
1967 #define BLKDEV_MAJOR_MAX	0
1968 #endif
1969 
1970 int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
1971 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1972 		void *holder);
1973 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1974 int bd_prepare_to_claim(struct block_device *bdev, struct block_device *whole,
1975 		void *holder);
1976 void bd_abort_claiming(struct block_device *bdev, struct block_device *whole,
1977 		void *holder);
1978 void blkdev_put(struct block_device *bdev, fmode_t mode);
1979 
1980 struct block_device *I_BDEV(struct inode *inode);
1981 struct block_device *bdget(dev_t);
1982 struct block_device *bdgrab(struct block_device *bdev);
1983 void bdput(struct block_device *);
1984 
1985 #ifdef CONFIG_BLOCK
1986 void invalidate_bdev(struct block_device *bdev);
1987 int sync_blockdev(struct block_device *bdev);
1988 #else
1989 static inline void invalidate_bdev(struct block_device *bdev)
1990 {
1991 }
1992 static inline int sync_blockdev(struct block_device *bdev)
1993 {
1994 	return 0;
1995 }
1996 #endif
1997 int fsync_bdev(struct block_device *bdev);
1998 
1999 struct super_block *freeze_bdev(struct block_device *bdev);
2000 int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2001 
2002 #endif /* _LINUX_BLKDEV_H */
2003