• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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 
8 #ifdef CONFIG_BLOCK
9 
10 #include <linux/major.h>
11 #include <linux/genhd.h>
12 #include <linux/list.h>
13 #include <linux/llist.h>
14 #include <linux/timer.h>
15 #include <linux/workqueue.h>
16 #include <linux/pagemap.h>
17 #include <linux/backing-dev-defs.h>
18 #include <linux/wait.h>
19 #include <linux/mempool.h>
20 #include <linux/pfn.h>
21 #include <linux/bio.h>
22 #include <linux/stringify.h>
23 #include <linux/gfp.h>
24 #include <linux/bsg.h>
25 #include <linux/smp.h>
26 #include <linux/rcupdate.h>
27 #include <linux/percpu-refcount.h>
28 #include <linux/scatterlist.h>
29 #include <linux/blkzoned.h>
30 
31 struct module;
32 struct scsi_ioctl_command;
33 
34 struct request_queue;
35 struct elevator_queue;
36 struct blk_trace;
37 struct request;
38 struct sg_io_hdr;
39 struct bsg_job;
40 struct blkcg_gq;
41 struct blk_flush_queue;
42 struct pr_ops;
43 struct rq_qos;
44 struct blk_queue_stats;
45 struct blk_stat_callback;
46 struct keyslot_manager;
47 
48 #define BLKDEV_MIN_RQ	4
49 #define BLKDEV_MAX_RQ	128	/* Default maximum */
50 
51 /* Must be consistent with blk_mq_poll_stats_bkt() */
52 #define BLK_MQ_POLL_STATS_BKTS 16
53 
54 /* Doing classic polling */
55 #define BLK_MQ_POLL_CLASSIC -1
56 
57 /*
58  * Maximum number of blkcg policies allowed to be registered concurrently.
59  * Defined here to simplify include dependency.
60  */
61 #define BLKCG_MAX_POLS		5
62 
63 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
64 
65 /*
66  * request flags */
67 typedef __u32 __bitwise req_flags_t;
68 
69 /* elevator knows about this request */
70 #define RQF_SORTED		((__force req_flags_t)(1 << 0))
71 /* drive already may have started this one */
72 #define RQF_STARTED		((__force req_flags_t)(1 << 1))
73 /* may not be passed by ioscheduler */
74 #define RQF_SOFTBARRIER		((__force req_flags_t)(1 << 3))
75 /* request for flush sequence */
76 #define RQF_FLUSH_SEQ		((__force req_flags_t)(1 << 4))
77 /* merge of different types, fail separately */
78 #define RQF_MIXED_MERGE		((__force req_flags_t)(1 << 5))
79 /* track inflight for MQ */
80 #define RQF_MQ_INFLIGHT		((__force req_flags_t)(1 << 6))
81 /* don't call prep for this one */
82 #define RQF_DONTPREP		((__force req_flags_t)(1 << 7))
83 /* set for "ide_preempt" requests and also for requests for which the SCSI
84    "quiesce" state must be ignored. */
85 #define RQF_PREEMPT		((__force req_flags_t)(1 << 8))
86 /* contains copies of user pages */
87 #define RQF_COPY_USER		((__force req_flags_t)(1 << 9))
88 /* vaguely specified driver internal error.  Ignored by the block layer */
89 #define RQF_FAILED		((__force req_flags_t)(1 << 10))
90 /* don't warn about errors */
91 #define RQF_QUIET		((__force req_flags_t)(1 << 11))
92 /* elevator private data attached */
93 #define RQF_ELVPRIV		((__force req_flags_t)(1 << 12))
94 /* account into disk and partition IO statistics */
95 #define RQF_IO_STAT		((__force req_flags_t)(1 << 13))
96 /* request came from our alloc pool */
97 #define RQF_ALLOCED		((__force req_flags_t)(1 << 14))
98 /* runtime pm request */
99 #define RQF_PM			((__force req_flags_t)(1 << 15))
100 /* on IO scheduler merge hash */
101 #define RQF_HASHED		((__force req_flags_t)(1 << 16))
102 /* track IO completion time */
103 #define RQF_STATS		((__force req_flags_t)(1 << 17))
104 /* Look at ->special_vec for the actual data payload instead of the
105    bio chain. */
106 #define RQF_SPECIAL_PAYLOAD	((__force req_flags_t)(1 << 18))
107 /* The per-zone write lock is held for this request */
108 #define RQF_ZONE_WRITE_LOCKED	((__force req_flags_t)(1 << 19))
109 /* already slept for hybrid poll */
110 #define RQF_MQ_POLL_SLEPT	((__force req_flags_t)(1 << 20))
111 /* ->timeout has been called, don't expire again */
112 #define RQF_TIMED_OUT		((__force req_flags_t)(1 << 21))
113 
114 /* flags that prevent us from merging requests: */
115 #define RQF_NOMERGE_FLAGS \
116 	(RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
117 
118 /*
119  * Request state for blk-mq.
120  */
121 enum mq_rq_state {
122 	MQ_RQ_IDLE		= 0,
123 	MQ_RQ_IN_FLIGHT		= 1,
124 	MQ_RQ_COMPLETE		= 2,
125 };
126 
127 /*
128  * Try to put the fields that are referenced together in the same cacheline.
129  *
130  * If you modify this structure, make sure to update blk_rq_init() and
131  * especially blk_mq_rq_ctx_init() to take care of the added fields.
132  */
133 struct request {
134 	struct request_queue *q;
135 	struct blk_mq_ctx *mq_ctx;
136 	struct blk_mq_hw_ctx *mq_hctx;
137 
138 	unsigned int cmd_flags;		/* op and common flags */
139 	req_flags_t rq_flags;
140 
141 	int tag;
142 	int internal_tag;
143 
144 	/* the following two fields are internal, NEVER access directly */
145 	unsigned int __data_len;	/* total data len */
146 	sector_t __sector;		/* sector cursor */
147 
148 	struct bio *bio;
149 	struct bio *biotail;
150 
151 	struct list_head queuelist;
152 
153 	/*
154 	 * The hash is used inside the scheduler, and killed once the
155 	 * request reaches the dispatch list. The ipi_list is only used
156 	 * to queue the request for softirq completion, which is long
157 	 * after the request has been unhashed (and even removed from
158 	 * the dispatch list).
159 	 */
160 	union {
161 		struct hlist_node hash;	/* merge hash */
162 		struct list_head ipi_list;
163 	};
164 
165 	/*
166 	 * The rb_node is only used inside the io scheduler, requests
167 	 * are pruned when moved to the dispatch queue. So let the
168 	 * completion_data share space with the rb_node.
169 	 */
170 	union {
171 		struct rb_node rb_node;	/* sort/lookup */
172 		struct bio_vec special_vec;
173 		void *completion_data;
174 		int error_count; /* for legacy drivers, don't use */
175 	};
176 
177 	/*
178 	 * Three pointers are available for the IO schedulers, if they need
179 	 * more they have to dynamically allocate it.  Flush requests are
180 	 * never put on the IO scheduler. So let the flush fields share
181 	 * space with the elevator data.
182 	 */
183 	union {
184 		struct {
185 			struct io_cq		*icq;
186 			void			*priv[2];
187 		} elv;
188 
189 		struct {
190 			unsigned int		seq;
191 			struct list_head	list;
192 			rq_end_io_fn		*saved_end_io;
193 		} flush;
194 	};
195 
196 	struct gendisk *rq_disk;
197 	struct hd_struct *part;
198 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
199 	/* Time that the first bio started allocating this request. */
200 	u64 alloc_time_ns;
201 #endif
202 	/* Time that this request was allocated for this IO. */
203 	u64 start_time_ns;
204 	/* Time that I/O was submitted to the device. */
205 	u64 io_start_time_ns;
206 
207 #ifdef CONFIG_BLK_WBT
208 	unsigned short wbt_flags;
209 #endif
210 	/*
211 	 * rq sectors used for blk stats. It has the same value
212 	 * with blk_rq_sectors(rq), except that it never be zeroed
213 	 * by completion.
214 	 */
215 	unsigned short stats_sectors;
216 
217 	/*
218 	 * Number of scatter-gather DMA addr+len pairs after
219 	 * physical address coalescing is performed.
220 	 */
221 	unsigned short nr_phys_segments;
222 
223 #if defined(CONFIG_BLK_DEV_INTEGRITY)
224 	unsigned short nr_integrity_segments;
225 #endif
226 
227 	unsigned short write_hint;
228 	unsigned short ioprio;
229 
230 	unsigned int extra_len;	/* length of alignment and padding */
231 
232 	enum mq_rq_state state;
233 	refcount_t ref;
234 
235 	unsigned int timeout;
236 	unsigned long deadline;
237 
238 	union {
239 		struct __call_single_data csd;
240 		u64 fifo_time;
241 	};
242 
243 	/*
244 	 * completion callback.
245 	 */
246 	rq_end_io_fn *end_io;
247 	void *end_io_data;
248 };
249 
blk_op_is_scsi(unsigned int op)250 static inline bool blk_op_is_scsi(unsigned int op)
251 {
252 	return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
253 }
254 
blk_op_is_private(unsigned int op)255 static inline bool blk_op_is_private(unsigned int op)
256 {
257 	return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
258 }
259 
blk_rq_is_scsi(struct request * rq)260 static inline bool blk_rq_is_scsi(struct request *rq)
261 {
262 	return blk_op_is_scsi(req_op(rq));
263 }
264 
blk_rq_is_private(struct request * rq)265 static inline bool blk_rq_is_private(struct request *rq)
266 {
267 	return blk_op_is_private(req_op(rq));
268 }
269 
blk_rq_is_passthrough(struct request * rq)270 static inline bool blk_rq_is_passthrough(struct request *rq)
271 {
272 	return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
273 }
274 
bio_is_passthrough(struct bio * bio)275 static inline bool bio_is_passthrough(struct bio *bio)
276 {
277 	unsigned op = bio_op(bio);
278 
279 	return blk_op_is_scsi(op) || blk_op_is_private(op);
280 }
281 
req_get_ioprio(struct request * req)282 static inline unsigned short req_get_ioprio(struct request *req)
283 {
284 	return req->ioprio;
285 }
286 
287 #include <linux/elevator.h>
288 
289 struct blk_queue_ctx;
290 
291 typedef blk_qc_t (make_request_fn) (struct request_queue *q, struct bio *bio);
292 
293 struct bio_vec;
294 typedef int (dma_drain_needed_fn)(struct request *);
295 
296 enum blk_eh_timer_return {
297 	BLK_EH_DONE,		/* drivers has completed the command */
298 	BLK_EH_RESET_TIMER,	/* reset timer and try again */
299 };
300 
301 enum blk_queue_state {
302 	Queue_down,
303 	Queue_up,
304 };
305 
306 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
307 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
308 
309 #define BLK_SCSI_MAX_CMDS	(256)
310 #define BLK_SCSI_CMD_PER_LONG	(BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
311 
312 /*
313  * Zoned block device models (zoned limit).
314  */
315 enum blk_zoned_model {
316 	BLK_ZONED_NONE,	/* Regular block device */
317 	BLK_ZONED_HA,	/* Host-aware zoned block device */
318 	BLK_ZONED_HM,	/* Host-managed zoned block device */
319 };
320 
321 struct queue_limits {
322 	unsigned long		bounce_pfn;
323 	unsigned long		seg_boundary_mask;
324 	unsigned long		virt_boundary_mask;
325 
326 	unsigned int		max_hw_sectors;
327 	unsigned int		max_dev_sectors;
328 	unsigned int		chunk_sectors;
329 	unsigned int		max_sectors;
330 	unsigned int		max_segment_size;
331 	unsigned int		physical_block_size;
332 	unsigned int		logical_block_size;
333 	unsigned int		alignment_offset;
334 	unsigned int		io_min;
335 	unsigned int		io_opt;
336 	unsigned int		max_discard_sectors;
337 	unsigned int		max_hw_discard_sectors;
338 	unsigned int		max_write_same_sectors;
339 	unsigned int		max_write_zeroes_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 extern int blk_revalidate_disk_zones(struct gendisk *disk);
366 
367 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
368 				     unsigned int cmd, unsigned long arg);
369 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
370 				  unsigned int cmd, unsigned long arg);
371 
372 #else /* CONFIG_BLK_DEV_ZONED */
373 
blkdev_nr_zones(struct gendisk * disk)374 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
375 {
376 	return 0;
377 }
378 
blkdev_report_zones_ioctl(struct block_device * bdev,fmode_t mode,unsigned int cmd,unsigned long arg)379 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
380 					    fmode_t mode, unsigned int cmd,
381 					    unsigned long arg)
382 {
383 	return -ENOTTY;
384 }
385 
blkdev_zone_mgmt_ioctl(struct block_device * bdev,fmode_t mode,unsigned int cmd,unsigned long arg)386 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
387 					 fmode_t mode, unsigned int cmd,
388 					 unsigned long arg)
389 {
390 	return -ENOTTY;
391 }
392 
393 #endif /* CONFIG_BLK_DEV_ZONED */
394 
395 struct request_queue {
396 	struct request		*last_merge;
397 	struct elevator_queue	*elevator;
398 
399 	struct blk_queue_stats	*stats;
400 	struct rq_qos		*rq_qos;
401 
402 	make_request_fn		*make_request_fn;
403 	dma_drain_needed_fn	*dma_drain_needed;
404 
405 	const struct blk_mq_ops	*mq_ops;
406 
407 	/* sw queues */
408 	struct blk_mq_ctx __percpu	*queue_ctx;
409 	unsigned int		nr_queues;
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 and RQF_PREEMPT requests are
432 	 * processed.
433 	 */
434 	atomic_t		pm_only;
435 
436 	/*
437 	 * ida allocated id for this queue.  Used to index queues from
438 	 * ioctx.
439 	 */
440 	int			id;
441 
442 	/*
443 	 * queue needs bounce pages for pages above this limit
444 	 */
445 	gfp_t			bounce_gfp;
446 
447 	spinlock_t		queue_lock;
448 
449 	/*
450 	 * queue kobject
451 	 */
452 	struct kobject kobj;
453 
454 	/*
455 	 * mq queue kobject
456 	 */
457 	struct kobject *mq_kobj;
458 
459 #ifdef  CONFIG_BLK_DEV_INTEGRITY
460 	struct blk_integrity integrity;
461 #endif	/* CONFIG_BLK_DEV_INTEGRITY */
462 
463 #ifdef CONFIG_PM
464 	struct device		*dev;
465 	int			rpm_status;
466 	unsigned int		nr_pending;
467 #endif
468 
469 	/*
470 	 * queue settings
471 	 */
472 	unsigned long		nr_requests;	/* Max # of requests */
473 
474 	unsigned int		dma_drain_size;
475 	void			*dma_drain_buffer;
476 	unsigned int		dma_pad_mask;
477 	unsigned int		dma_alignment;
478 
479 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
480 	/* Inline crypto capabilities */
481 	struct keyslot_manager *ksm;
482 #endif
483 
484 	unsigned int		rq_timeout;
485 	int			poll_nsec;
486 
487 	struct blk_stat_callback	*poll_cb;
488 	struct blk_rq_stat	poll_stat[BLK_MQ_POLL_STATS_BKTS];
489 
490 	struct timer_list	timeout;
491 	struct work_struct	timeout_work;
492 
493 	struct list_head	icq_list;
494 #ifdef CONFIG_BLK_CGROUP
495 	DECLARE_BITMAP		(blkcg_pols, BLKCG_MAX_POLS);
496 	struct blkcg_gq		*root_blkg;
497 	struct list_head	blkg_list;
498 #endif
499 
500 	struct queue_limits	limits;
501 
502 	unsigned int		required_elevator_features;
503 
504 #ifdef CONFIG_BLK_DEV_ZONED
505 	/*
506 	 * Zoned block device information for request dispatch control.
507 	 * nr_zones is the total number of zones of the device. This is always
508 	 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
509 	 * bits which indicates if a zone is conventional (bit set) or
510 	 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
511 	 * bits which indicates if a zone is write locked, that is, if a write
512 	 * request targeting the zone was dispatched. All three fields are
513 	 * initialized by the low level device driver (e.g. scsi/sd.c).
514 	 * Stacking drivers (device mappers) may or may not initialize
515 	 * these fields.
516 	 *
517 	 * Reads of this information must be protected with blk_queue_enter() /
518 	 * blk_queue_exit(). Modifying this information is only allowed while
519 	 * no requests are being processed. See also blk_mq_freeze_queue() and
520 	 * blk_mq_unfreeze_queue().
521 	 */
522 	unsigned int		nr_zones;
523 	unsigned long		*conv_zones_bitmap;
524 	unsigned long		*seq_zones_wlock;
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 #ifdef CONFIG_BLK_DEV_IO_TRACE
534 	struct blk_trace	*blk_trace;
535 	struct mutex		blk_trace_mutex;
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 	struct percpu_ref	q_usage_counter;
574 
575 	struct blk_mq_tag_set	*tag_set;
576 	struct list_head	tag_set_list;
577 	struct bio_set		bio_split;
578 
579 #ifdef CONFIG_BLK_DEBUG_FS
580 	struct dentry		*debugfs_dir;
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 	struct work_struct	release_work;
590 
591 #define BLK_MAX_WRITE_HINTS	5
592 	u64			write_hints[BLK_MAX_WRITE_HINTS];
593 };
594 
595 #define QUEUE_FLAG_STOPPED	0	/* queue is stopped */
596 #define QUEUE_FLAG_DYING	1	/* queue being torn down */
597 #define QUEUE_FLAG_NOMERGES     3	/* disable merge attempts */
598 #define QUEUE_FLAG_SAME_COMP	4	/* complete on same CPU-group */
599 #define QUEUE_FLAG_FAIL_IO	5	/* fake timeout */
600 #define QUEUE_FLAG_NONROT	6	/* non-rotational device (SSD) */
601 #define QUEUE_FLAG_VIRT		QUEUE_FLAG_NONROT /* paravirt device */
602 #define QUEUE_FLAG_IO_STAT	7	/* do disk/partitions IO accounting */
603 #define QUEUE_FLAG_DISCARD	8	/* supports DISCARD */
604 #define QUEUE_FLAG_NOXMERGES	9	/* No extended merges */
605 #define QUEUE_FLAG_ADD_RANDOM	10	/* Contributes to random pool */
606 #define QUEUE_FLAG_SECERASE	11	/* supports secure erase */
607 #define QUEUE_FLAG_SAME_FORCE	12	/* force complete on same CPU */
608 #define QUEUE_FLAG_DEAD		13	/* queue tear-down finished */
609 #define QUEUE_FLAG_INIT_DONE	14	/* queue is initialized */
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 
623 #define QUEUE_FLAG_MQ_DEFAULT	((1 << QUEUE_FLAG_IO_STAT) |		\
624 				 (1 << QUEUE_FLAG_SAME_COMP))
625 
626 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
627 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
628 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
629 
630 #define blk_queue_stopped(q)	test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
631 #define blk_queue_dying(q)	test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
632 #define blk_queue_dead(q)	test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
633 #define blk_queue_init_done(q)	test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
634 #define blk_queue_nomerges(q)	test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
635 #define blk_queue_noxmerges(q)	\
636 	test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
637 #define blk_queue_nonrot(q)	test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
638 #define blk_queue_io_stat(q)	test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
639 #define blk_queue_add_random(q)	test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
640 #define blk_queue_discard(q)	test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
641 #define blk_queue_zone_resetall(q)	\
642 	test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
643 #define blk_queue_secure_erase(q) \
644 	(test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
645 #define blk_queue_dax(q)	test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
646 #define blk_queue_scsi_passthrough(q)	\
647 	test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
648 #define blk_queue_pci_p2pdma(q)	\
649 	test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
650 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
651 #define blk_queue_rq_alloc_time(q)	\
652 	test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
653 #else
654 #define blk_queue_rq_alloc_time(q)	false
655 #endif
656 
657 #define blk_noretry_request(rq) \
658 	((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
659 			     REQ_FAILFAST_DRIVER))
660 #define blk_queue_quiesced(q)	test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
661 #define blk_queue_pm_only(q)	atomic_read(&(q)->pm_only)
662 #define blk_queue_fua(q)	test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
663 #define blk_queue_registered(q)	test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
664 
665 extern void blk_set_pm_only(struct request_queue *q);
666 extern void blk_clear_pm_only(struct request_queue *q);
667 
blk_account_rq(struct request * rq)668 static inline bool blk_account_rq(struct request *rq)
669 {
670 	return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
671 }
672 
673 #define list_entry_rq(ptr)	list_entry((ptr), struct request, queuelist)
674 
675 #define rq_data_dir(rq)		(op_is_write(req_op(rq)) ? WRITE : READ)
676 
677 #define rq_dma_dir(rq) \
678 	(op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
679 
680 #define dma_map_bvec(dev, bv, dir, attrs) \
681 	dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
682 	(dir), (attrs))
683 
queue_is_mq(struct request_queue * q)684 static inline bool queue_is_mq(struct request_queue *q)
685 {
686 	return q->mq_ops;
687 }
688 
689 static inline enum blk_zoned_model
blk_queue_zoned_model(struct request_queue * q)690 blk_queue_zoned_model(struct request_queue *q)
691 {
692 	return q->limits.zoned;
693 }
694 
blk_queue_is_zoned(struct request_queue * q)695 static inline bool blk_queue_is_zoned(struct request_queue *q)
696 {
697 	switch (blk_queue_zoned_model(q)) {
698 	case BLK_ZONED_HA:
699 	case BLK_ZONED_HM:
700 		return true;
701 	default:
702 		return false;
703 	}
704 }
705 
blk_queue_zone_sectors(struct request_queue * q)706 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
707 {
708 	return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
709 }
710 
711 #ifdef CONFIG_BLK_DEV_ZONED
blk_queue_nr_zones(struct request_queue * q)712 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
713 {
714 	return blk_queue_is_zoned(q) ? q->nr_zones : 0;
715 }
716 
blk_queue_zone_no(struct request_queue * q,sector_t sector)717 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
718 					     sector_t sector)
719 {
720 	if (!blk_queue_is_zoned(q))
721 		return 0;
722 	return sector >> ilog2(q->limits.chunk_sectors);
723 }
724 
blk_queue_zone_is_seq(struct request_queue * q,sector_t sector)725 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
726 					 sector_t sector)
727 {
728 	if (!blk_queue_is_zoned(q))
729 		return false;
730 	if (!q->conv_zones_bitmap)
731 		return true;
732 	return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
733 }
734 #else /* CONFIG_BLK_DEV_ZONED */
blk_queue_nr_zones(struct request_queue * q)735 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
736 {
737 	return 0;
738 }
739 #endif /* CONFIG_BLK_DEV_ZONED */
740 
rq_is_sync(struct request * rq)741 static inline bool rq_is_sync(struct request *rq)
742 {
743 	return op_is_sync(rq->cmd_flags);
744 }
745 
rq_mergeable(struct request * rq)746 static inline bool rq_mergeable(struct request *rq)
747 {
748 	if (blk_rq_is_passthrough(rq))
749 		return false;
750 
751 	if (req_op(rq) == REQ_OP_FLUSH)
752 		return false;
753 
754 	if (req_op(rq) == REQ_OP_WRITE_ZEROES)
755 		return false;
756 
757 	if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
758 		return false;
759 	if (rq->rq_flags & RQF_NOMERGE_FLAGS)
760 		return false;
761 
762 	return true;
763 }
764 
blk_write_same_mergeable(struct bio * a,struct bio * b)765 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
766 {
767 	if (bio_page(a) == bio_page(b) &&
768 	    bio_offset(a) == bio_offset(b))
769 		return true;
770 
771 	return false;
772 }
773 
blk_queue_depth(struct request_queue * q)774 static inline unsigned int blk_queue_depth(struct request_queue *q)
775 {
776 	if (q->queue_depth)
777 		return q->queue_depth;
778 
779 	return q->nr_requests;
780 }
781 
782 extern unsigned long blk_max_low_pfn, blk_max_pfn;
783 
784 /*
785  * standard bounce addresses:
786  *
787  * BLK_BOUNCE_HIGH	: bounce all highmem pages
788  * BLK_BOUNCE_ANY	: don't bounce anything
789  * BLK_BOUNCE_ISA	: bounce pages above ISA DMA boundary
790  */
791 
792 #if BITS_PER_LONG == 32
793 #define BLK_BOUNCE_HIGH		((u64)blk_max_low_pfn << PAGE_SHIFT)
794 #else
795 #define BLK_BOUNCE_HIGH		-1ULL
796 #endif
797 #define BLK_BOUNCE_ANY		(-1ULL)
798 #define BLK_BOUNCE_ISA		(DMA_BIT_MASK(24))
799 
800 /*
801  * default timeout for SG_IO if none specified
802  */
803 #define BLK_DEFAULT_SG_TIMEOUT	(60 * HZ)
804 #define BLK_MIN_SG_TIMEOUT	(7 * HZ)
805 
806 struct rq_map_data {
807 	struct page **pages;
808 	int page_order;
809 	int nr_entries;
810 	unsigned long offset;
811 	int null_mapped;
812 	int from_user;
813 };
814 
815 struct req_iterator {
816 	struct bvec_iter iter;
817 	struct bio *bio;
818 };
819 
820 /* This should not be used directly - use rq_for_each_segment */
821 #define for_each_bio(_bio)		\
822 	for (; _bio; _bio = _bio->bi_next)
823 #define __rq_for_each_bio(_bio, rq)	\
824 	if ((rq->bio))			\
825 		for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
826 
827 #define rq_for_each_segment(bvl, _rq, _iter)			\
828 	__rq_for_each_bio(_iter.bio, _rq)			\
829 		bio_for_each_segment(bvl, _iter.bio, _iter.iter)
830 
831 #define rq_for_each_bvec(bvl, _rq, _iter)			\
832 	__rq_for_each_bio(_iter.bio, _rq)			\
833 		bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
834 
835 #define rq_iter_last(bvec, _iter)				\
836 		(_iter.bio->bi_next == NULL &&			\
837 		 bio_iter_last(bvec, _iter.iter))
838 
839 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
840 # error	"You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
841 #endif
842 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
843 extern void rq_flush_dcache_pages(struct request *rq);
844 #else
rq_flush_dcache_pages(struct request * rq)845 static inline void rq_flush_dcache_pages(struct request *rq)
846 {
847 }
848 #endif
849 
850 extern int blk_register_queue(struct gendisk *disk);
851 extern void blk_unregister_queue(struct gendisk *disk);
852 extern blk_qc_t generic_make_request(struct bio *bio);
853 extern blk_qc_t direct_make_request(struct bio *bio);
854 extern void blk_rq_init(struct request_queue *q, struct request *rq);
855 extern void blk_put_request(struct request *);
856 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
857 				       blk_mq_req_flags_t flags);
858 extern int blk_lld_busy(struct request_queue *q);
859 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
860 			     struct bio_set *bs, gfp_t gfp_mask,
861 			     int (*bio_ctr)(struct bio *, struct bio *, void *),
862 			     void *data);
863 extern void blk_rq_unprep_clone(struct request *rq);
864 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
865 				     struct request *rq);
866 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
867 extern void blk_queue_split(struct request_queue *, struct bio **);
868 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
869 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
870 			      unsigned int, void __user *);
871 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
872 			  unsigned int, void __user *);
873 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
874 			 struct scsi_ioctl_command __user *);
875 
876 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
877 extern void blk_queue_exit(struct request_queue *q);
878 extern void blk_sync_queue(struct request_queue *q);
879 extern int blk_rq_map_user(struct request_queue *, struct request *,
880 			   struct rq_map_data *, void __user *, unsigned long,
881 			   gfp_t);
882 extern int blk_rq_unmap_user(struct bio *);
883 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
884 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
885 			       struct rq_map_data *, const struct iov_iter *,
886 			       gfp_t);
887 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
888 			  struct request *, int);
889 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
890 				  struct request *, int, rq_end_io_fn *);
891 
892 /* Helper to convert REQ_OP_XXX to its string format XXX */
893 extern const char *blk_op_str(unsigned int op);
894 
895 int blk_status_to_errno(blk_status_t status);
896 blk_status_t errno_to_blk_status(int errno);
897 
898 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
899 
bdev_get_queue(struct block_device * bdev)900 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
901 {
902 	return bdev->bd_disk->queue;	/* this is never NULL */
903 }
904 
905 /*
906  * The basic unit of block I/O is a sector. It is used in a number of contexts
907  * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
908  * bytes. Variables of type sector_t represent an offset or size that is a
909  * multiple of 512 bytes. Hence these two constants.
910  */
911 #ifndef SECTOR_SHIFT
912 #define SECTOR_SHIFT 9
913 #endif
914 #ifndef SECTOR_SIZE
915 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
916 #endif
917 
918 /*
919  * blk_rq_pos()			: the current sector
920  * blk_rq_bytes()		: bytes left in the entire request
921  * blk_rq_cur_bytes()		: bytes left in the current segment
922  * blk_rq_err_bytes()		: bytes left till the next error boundary
923  * blk_rq_sectors()		: sectors left in the entire request
924  * blk_rq_cur_sectors()		: sectors left in the current segment
925  * blk_rq_stats_sectors()	: sectors of the entire request used for stats
926  */
blk_rq_pos(const struct request * rq)927 static inline sector_t blk_rq_pos(const struct request *rq)
928 {
929 	return rq->__sector;
930 }
931 
blk_rq_bytes(const struct request * rq)932 static inline unsigned int blk_rq_bytes(const struct request *rq)
933 {
934 	return rq->__data_len;
935 }
936 
blk_rq_cur_bytes(const struct request * rq)937 static inline int blk_rq_cur_bytes(const struct request *rq)
938 {
939 	return rq->bio ? bio_cur_bytes(rq->bio) : 0;
940 }
941 
942 extern unsigned int blk_rq_err_bytes(const struct request *rq);
943 
blk_rq_sectors(const struct request * rq)944 static inline unsigned int blk_rq_sectors(const struct request *rq)
945 {
946 	return blk_rq_bytes(rq) >> SECTOR_SHIFT;
947 }
948 
blk_rq_cur_sectors(const struct request * rq)949 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
950 {
951 	return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
952 }
953 
blk_rq_stats_sectors(const struct request * rq)954 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
955 {
956 	return rq->stats_sectors;
957 }
958 
959 #ifdef CONFIG_BLK_DEV_ZONED
blk_rq_zone_no(struct request * rq)960 static inline unsigned int blk_rq_zone_no(struct request *rq)
961 {
962 	return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
963 }
964 
blk_rq_zone_is_seq(struct request * rq)965 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
966 {
967 	return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
968 }
969 #endif /* CONFIG_BLK_DEV_ZONED */
970 
971 /*
972  * Some commands like WRITE SAME have a payload or data transfer size which
973  * is different from the size of the request.  Any driver that supports such
974  * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
975  * calculate the data transfer size.
976  */
blk_rq_payload_bytes(struct request * rq)977 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
978 {
979 	if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
980 		return rq->special_vec.bv_len;
981 	return blk_rq_bytes(rq);
982 }
983 
984 /*
985  * Return the first full biovec in the request.  The caller needs to check that
986  * there are any bvecs before calling this helper.
987  */
req_bvec(struct request * rq)988 static inline struct bio_vec req_bvec(struct request *rq)
989 {
990 	if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
991 		return rq->special_vec;
992 	return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
993 }
994 
blk_queue_get_max_sectors(struct request_queue * q,int op)995 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
996 						     int op)
997 {
998 	if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
999 		return min(q->limits.max_discard_sectors,
1000 			   UINT_MAX >> SECTOR_SHIFT);
1001 
1002 	if (unlikely(op == REQ_OP_WRITE_SAME))
1003 		return q->limits.max_write_same_sectors;
1004 
1005 	if (unlikely(op == REQ_OP_WRITE_ZEROES))
1006 		return q->limits.max_write_zeroes_sectors;
1007 
1008 	return q->limits.max_sectors;
1009 }
1010 
1011 /*
1012  * Return maximum size of a request at given offset. Only valid for
1013  * file system requests.
1014  */
blk_max_size_offset(struct request_queue * q,sector_t offset)1015 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1016 					       sector_t offset)
1017 {
1018 	if (!q->limits.chunk_sectors)
1019 		return q->limits.max_sectors;
1020 
1021 	return min(q->limits.max_sectors, (unsigned int)(q->limits.chunk_sectors -
1022 			(offset & (q->limits.chunk_sectors - 1))));
1023 }
1024 
blk_rq_get_max_sectors(struct request * rq,sector_t offset)1025 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1026 						  sector_t offset)
1027 {
1028 	struct request_queue *q = rq->q;
1029 
1030 	if (blk_rq_is_passthrough(rq))
1031 		return q->limits.max_hw_sectors;
1032 
1033 	if (!q->limits.chunk_sectors ||
1034 	    req_op(rq) == REQ_OP_DISCARD ||
1035 	    req_op(rq) == REQ_OP_SECURE_ERASE)
1036 		return blk_queue_get_max_sectors(q, req_op(rq));
1037 
1038 	return min(blk_max_size_offset(q, offset),
1039 			blk_queue_get_max_sectors(q, req_op(rq)));
1040 }
1041 
blk_rq_count_bios(struct request * rq)1042 static inline unsigned int blk_rq_count_bios(struct request *rq)
1043 {
1044 	unsigned int nr_bios = 0;
1045 	struct bio *bio;
1046 
1047 	__rq_for_each_bio(bio, rq)
1048 		nr_bios++;
1049 
1050 	return nr_bios;
1051 }
1052 
1053 void blk_steal_bios(struct bio_list *list, struct request *rq);
1054 
1055 /*
1056  * Request completion related functions.
1057  *
1058  * blk_update_request() completes given number of bytes and updates
1059  * the request without completing it.
1060  */
1061 extern bool blk_update_request(struct request *rq, blk_status_t error,
1062 			       unsigned int nr_bytes);
1063 
1064 extern void __blk_complete_request(struct request *);
1065 extern void blk_abort_request(struct request *);
1066 
1067 /*
1068  * Access functions for manipulating queue properties
1069  */
1070 extern void blk_cleanup_queue(struct request_queue *);
1071 extern void blk_queue_make_request(struct request_queue *, make_request_fn *);
1072 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1073 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1074 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1075 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1076 extern void blk_queue_max_discard_segments(struct request_queue *,
1077 		unsigned short);
1078 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1079 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1080 		unsigned int max_discard_sectors);
1081 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1082 		unsigned int max_write_same_sectors);
1083 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1084 		unsigned int max_write_same_sectors);
1085 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1086 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1087 extern void blk_queue_alignment_offset(struct request_queue *q,
1088 				       unsigned int alignment);
1089 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1090 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1091 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1092 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1093 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1094 extern void blk_set_default_limits(struct queue_limits *lim);
1095 extern void blk_set_stacking_limits(struct queue_limits *lim);
1096 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1097 			    sector_t offset);
1098 extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev,
1099 			    sector_t offset);
1100 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1101 			      sector_t offset);
1102 extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b);
1103 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1104 extern int blk_queue_dma_drain(struct request_queue *q,
1105 			       dma_drain_needed_fn *dma_drain_needed,
1106 			       void *buf, unsigned int size);
1107 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1108 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1109 extern void blk_queue_dma_alignment(struct request_queue *, int);
1110 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1111 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1112 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1113 extern void blk_queue_required_elevator_features(struct request_queue *q,
1114 						 unsigned int features);
1115 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1116 					      struct device *dev);
1117 
1118 /*
1119  * Number of physical segments as sent to the device.
1120  *
1121  * Normally this is the number of discontiguous data segments sent by the
1122  * submitter.  But for data-less command like discard we might have no
1123  * actual data segments submitted, but the driver might have to add it's
1124  * own special payload.  In that case we still return 1 here so that this
1125  * special payload will be mapped.
1126  */
blk_rq_nr_phys_segments(struct request * rq)1127 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1128 {
1129 	if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1130 		return 1;
1131 	return rq->nr_phys_segments;
1132 }
1133 
1134 /*
1135  * Number of discard segments (or ranges) the driver needs to fill in.
1136  * Each discard bio merged into a request is counted as one segment.
1137  */
blk_rq_nr_discard_segments(struct request * rq)1138 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1139 {
1140 	return max_t(unsigned short, rq->nr_phys_segments, 1);
1141 }
1142 
1143 extern int blk_rq_map_sg(struct request_queue *, struct request *, struct scatterlist *);
1144 extern void blk_dump_rq_flags(struct request *, char *);
1145 extern long nr_blockdev_pages(void);
1146 
1147 bool __must_check blk_get_queue(struct request_queue *);
1148 struct request_queue *blk_alloc_queue(gfp_t);
1149 struct request_queue *blk_alloc_queue_node(gfp_t gfp_mask, int node_id);
1150 extern void blk_put_queue(struct request_queue *);
1151 extern void blk_set_queue_dying(struct request_queue *);
1152 
1153 /*
1154  * blk_plug permits building a queue of related requests by holding the I/O
1155  * fragments for a short period. This allows merging of sequential requests
1156  * into single larger request. As the requests are moved from a per-task list to
1157  * the device's request_queue in a batch, this results in improved scalability
1158  * as the lock contention for request_queue lock is reduced.
1159  *
1160  * It is ok not to disable preemption when adding the request to the plug list
1161  * or when attempting a merge, because blk_schedule_flush_list() will only flush
1162  * the plug list when the task sleeps by itself. For details, please see
1163  * schedule() where blk_schedule_flush_plug() is called.
1164  */
1165 struct blk_plug {
1166 	struct list_head mq_list; /* blk-mq requests */
1167 	struct list_head cb_list; /* md requires an unplug callback */
1168 	unsigned short rq_count;
1169 	bool multiple_queues;
1170 };
1171 #define BLK_MAX_REQUEST_COUNT 16
1172 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1173 
1174 struct blk_plug_cb;
1175 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1176 struct blk_plug_cb {
1177 	struct list_head list;
1178 	blk_plug_cb_fn callback;
1179 	void *data;
1180 };
1181 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1182 					     void *data, int size);
1183 extern void blk_start_plug(struct blk_plug *);
1184 extern void blk_finish_plug(struct blk_plug *);
1185 extern void blk_flush_plug_list(struct blk_plug *, bool);
1186 
blk_flush_plug(struct task_struct * tsk)1187 static inline void blk_flush_plug(struct task_struct *tsk)
1188 {
1189 	struct blk_plug *plug = tsk->plug;
1190 
1191 	if (plug)
1192 		blk_flush_plug_list(plug, false);
1193 }
1194 
blk_schedule_flush_plug(struct task_struct * tsk)1195 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1196 {
1197 	struct blk_plug *plug = tsk->plug;
1198 
1199 	if (plug)
1200 		blk_flush_plug_list(plug, true);
1201 }
1202 
blk_needs_flush_plug(struct task_struct * tsk)1203 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1204 {
1205 	struct blk_plug *plug = tsk->plug;
1206 
1207 	return plug &&
1208 		 (!list_empty(&plug->mq_list) ||
1209 		 !list_empty(&plug->cb_list));
1210 }
1211 
1212 extern int blkdev_issue_flush(struct block_device *, gfp_t, sector_t *);
1213 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1214 		sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1215 
1216 #define BLKDEV_DISCARD_SECURE	(1 << 0)	/* issue a secure erase */
1217 
1218 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1219 		sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1220 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1221 		sector_t nr_sects, gfp_t gfp_mask, int flags,
1222 		struct bio **biop);
1223 
1224 #define BLKDEV_ZERO_NOUNMAP	(1 << 0)  /* do not free blocks */
1225 #define BLKDEV_ZERO_NOFALLBACK	(1 << 1)  /* don't write explicit zeroes */
1226 
1227 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1228 		sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1229 		unsigned flags);
1230 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1231 		sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1232 
sb_issue_discard(struct super_block * sb,sector_t block,sector_t nr_blocks,gfp_t gfp_mask,unsigned long flags)1233 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1234 		sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1235 {
1236 	return blkdev_issue_discard(sb->s_bdev,
1237 				    block << (sb->s_blocksize_bits -
1238 					      SECTOR_SHIFT),
1239 				    nr_blocks << (sb->s_blocksize_bits -
1240 						  SECTOR_SHIFT),
1241 				    gfp_mask, flags);
1242 }
sb_issue_zeroout(struct super_block * sb,sector_t block,sector_t nr_blocks,gfp_t gfp_mask)1243 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1244 		sector_t nr_blocks, gfp_t gfp_mask)
1245 {
1246 	return blkdev_issue_zeroout(sb->s_bdev,
1247 				    block << (sb->s_blocksize_bits -
1248 					      SECTOR_SHIFT),
1249 				    nr_blocks << (sb->s_blocksize_bits -
1250 						  SECTOR_SHIFT),
1251 				    gfp_mask, 0);
1252 }
1253 
1254 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1255 
1256 enum blk_default_limits {
1257 	BLK_MAX_SEGMENTS	= 128,
1258 	BLK_SAFE_MAX_SECTORS	= 255,
1259 	BLK_DEF_MAX_SECTORS	= 2560,
1260 	BLK_MAX_SEGMENT_SIZE	= 65536,
1261 	BLK_SEG_BOUNDARY_MASK	= 0xFFFFFFFFUL,
1262 };
1263 
queue_segment_boundary(const struct request_queue * q)1264 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1265 {
1266 	return q->limits.seg_boundary_mask;
1267 }
1268 
queue_virt_boundary(const struct request_queue * q)1269 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1270 {
1271 	return q->limits.virt_boundary_mask;
1272 }
1273 
queue_max_sectors(const struct request_queue * q)1274 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1275 {
1276 	return q->limits.max_sectors;
1277 }
1278 
queue_max_hw_sectors(const struct request_queue * q)1279 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1280 {
1281 	return q->limits.max_hw_sectors;
1282 }
1283 
queue_max_segments(const struct request_queue * q)1284 static inline unsigned short queue_max_segments(const struct request_queue *q)
1285 {
1286 	return q->limits.max_segments;
1287 }
1288 
queue_max_discard_segments(const struct request_queue * q)1289 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1290 {
1291 	return q->limits.max_discard_segments;
1292 }
1293 
queue_max_segment_size(const struct request_queue * q)1294 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1295 {
1296 	return q->limits.max_segment_size;
1297 }
1298 
queue_logical_block_size(const struct request_queue * q)1299 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1300 {
1301 	int retval = 512;
1302 
1303 	if (q && q->limits.logical_block_size)
1304 		retval = q->limits.logical_block_size;
1305 
1306 	return retval;
1307 }
1308 
bdev_logical_block_size(struct block_device * bdev)1309 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1310 {
1311 	return queue_logical_block_size(bdev_get_queue(bdev));
1312 }
1313 
queue_physical_block_size(const struct request_queue * q)1314 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1315 {
1316 	return q->limits.physical_block_size;
1317 }
1318 
bdev_physical_block_size(struct block_device * bdev)1319 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1320 {
1321 	return queue_physical_block_size(bdev_get_queue(bdev));
1322 }
1323 
queue_io_min(const struct request_queue * q)1324 static inline unsigned int queue_io_min(const struct request_queue *q)
1325 {
1326 	return q->limits.io_min;
1327 }
1328 
bdev_io_min(struct block_device * bdev)1329 static inline int bdev_io_min(struct block_device *bdev)
1330 {
1331 	return queue_io_min(bdev_get_queue(bdev));
1332 }
1333 
queue_io_opt(const struct request_queue * q)1334 static inline unsigned int queue_io_opt(const struct request_queue *q)
1335 {
1336 	return q->limits.io_opt;
1337 }
1338 
bdev_io_opt(struct block_device * bdev)1339 static inline int bdev_io_opt(struct block_device *bdev)
1340 {
1341 	return queue_io_opt(bdev_get_queue(bdev));
1342 }
1343 
queue_alignment_offset(const struct request_queue * q)1344 static inline int queue_alignment_offset(const struct request_queue *q)
1345 {
1346 	if (q->limits.misaligned)
1347 		return -1;
1348 
1349 	return q->limits.alignment_offset;
1350 }
1351 
queue_limit_alignment_offset(struct queue_limits * lim,sector_t sector)1352 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1353 {
1354 	unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1355 	unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1356 		<< SECTOR_SHIFT;
1357 
1358 	return (granularity + lim->alignment_offset - alignment) % granularity;
1359 }
1360 
bdev_alignment_offset(struct block_device * bdev)1361 static inline int bdev_alignment_offset(struct block_device *bdev)
1362 {
1363 	struct request_queue *q = bdev_get_queue(bdev);
1364 
1365 	if (q->limits.misaligned)
1366 		return -1;
1367 
1368 	if (bdev != bdev->bd_contains)
1369 		return bdev->bd_part->alignment_offset;
1370 
1371 	return q->limits.alignment_offset;
1372 }
1373 
queue_discard_alignment(const struct request_queue * q)1374 static inline int queue_discard_alignment(const struct request_queue *q)
1375 {
1376 	if (q->limits.discard_misaligned)
1377 		return -1;
1378 
1379 	return q->limits.discard_alignment;
1380 }
1381 
queue_limit_discard_alignment(struct queue_limits * lim,sector_t sector)1382 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1383 {
1384 	unsigned int alignment, granularity, offset;
1385 
1386 	if (!lim->max_discard_sectors)
1387 		return 0;
1388 
1389 	/* Why are these in bytes, not sectors? */
1390 	alignment = lim->discard_alignment >> SECTOR_SHIFT;
1391 	granularity = lim->discard_granularity >> SECTOR_SHIFT;
1392 	if (!granularity)
1393 		return 0;
1394 
1395 	/* Offset of the partition start in 'granularity' sectors */
1396 	offset = sector_div(sector, granularity);
1397 
1398 	/* And why do we do this modulus *again* in blkdev_issue_discard()? */
1399 	offset = (granularity + alignment - offset) % granularity;
1400 
1401 	/* Turn it back into bytes, gaah */
1402 	return offset << SECTOR_SHIFT;
1403 }
1404 
bdev_discard_alignment(struct block_device * bdev)1405 static inline int bdev_discard_alignment(struct block_device *bdev)
1406 {
1407 	struct request_queue *q = bdev_get_queue(bdev);
1408 
1409 	if (bdev != bdev->bd_contains)
1410 		return bdev->bd_part->discard_alignment;
1411 
1412 	return q->limits.discard_alignment;
1413 }
1414 
bdev_write_same(struct block_device * bdev)1415 static inline unsigned int bdev_write_same(struct block_device *bdev)
1416 {
1417 	struct request_queue *q = bdev_get_queue(bdev);
1418 
1419 	if (q)
1420 		return q->limits.max_write_same_sectors;
1421 
1422 	return 0;
1423 }
1424 
bdev_write_zeroes_sectors(struct block_device * bdev)1425 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1426 {
1427 	struct request_queue *q = bdev_get_queue(bdev);
1428 
1429 	if (q)
1430 		return q->limits.max_write_zeroes_sectors;
1431 
1432 	return 0;
1433 }
1434 
bdev_zoned_model(struct block_device * bdev)1435 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1436 {
1437 	struct request_queue *q = bdev_get_queue(bdev);
1438 
1439 	if (q)
1440 		return blk_queue_zoned_model(q);
1441 
1442 	return BLK_ZONED_NONE;
1443 }
1444 
bdev_is_zoned(struct block_device * bdev)1445 static inline bool bdev_is_zoned(struct block_device *bdev)
1446 {
1447 	struct request_queue *q = bdev_get_queue(bdev);
1448 
1449 	if (q)
1450 		return blk_queue_is_zoned(q);
1451 
1452 	return false;
1453 }
1454 
bdev_zone_sectors(struct block_device * bdev)1455 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1456 {
1457 	struct request_queue *q = bdev_get_queue(bdev);
1458 
1459 	if (q)
1460 		return blk_queue_zone_sectors(q);
1461 	return 0;
1462 }
1463 
queue_dma_alignment(const struct request_queue * q)1464 static inline int queue_dma_alignment(const struct request_queue *q)
1465 {
1466 	return q ? q->dma_alignment : 511;
1467 }
1468 
blk_rq_aligned(struct request_queue * q,unsigned long addr,unsigned int len)1469 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1470 				 unsigned int len)
1471 {
1472 	unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1473 	return !(addr & alignment) && !(len & alignment);
1474 }
1475 
1476 /* assumes size > 256 */
blksize_bits(unsigned int size)1477 static inline unsigned int blksize_bits(unsigned int size)
1478 {
1479 	unsigned int bits = 8;
1480 	do {
1481 		bits++;
1482 		size >>= 1;
1483 	} while (size > 256);
1484 	return bits;
1485 }
1486 
block_size(struct block_device * bdev)1487 static inline unsigned int block_size(struct block_device *bdev)
1488 {
1489 	return bdev->bd_block_size;
1490 }
1491 
1492 typedef struct {struct page *v;} Sector;
1493 
1494 unsigned char *read_dev_sector(struct block_device *, sector_t, Sector *);
1495 
put_dev_sector(Sector p)1496 static inline void put_dev_sector(Sector p)
1497 {
1498 	put_page(p.v);
1499 }
1500 
1501 int kblockd_schedule_work(struct work_struct *work);
1502 int kblockd_schedule_work_on(int cpu, struct work_struct *work);
1503 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1504 
1505 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1506 	MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1507 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1508 	MODULE_ALIAS("block-major-" __stringify(major) "-*")
1509 
1510 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1511 
1512 enum blk_integrity_flags {
1513 	BLK_INTEGRITY_VERIFY		= 1 << 0,
1514 	BLK_INTEGRITY_GENERATE		= 1 << 1,
1515 	BLK_INTEGRITY_DEVICE_CAPABLE	= 1 << 2,
1516 	BLK_INTEGRITY_IP_CHECKSUM	= 1 << 3,
1517 };
1518 
1519 struct blk_integrity_iter {
1520 	void			*prot_buf;
1521 	void			*data_buf;
1522 	sector_t		seed;
1523 	unsigned int		data_size;
1524 	unsigned short		interval;
1525 	const char		*disk_name;
1526 };
1527 
1528 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1529 typedef void (integrity_prepare_fn) (struct request *);
1530 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1531 
1532 struct blk_integrity_profile {
1533 	integrity_processing_fn		*generate_fn;
1534 	integrity_processing_fn		*verify_fn;
1535 	integrity_prepare_fn		*prepare_fn;
1536 	integrity_complete_fn		*complete_fn;
1537 	const char			*name;
1538 };
1539 
1540 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1541 extern void blk_integrity_unregister(struct gendisk *);
1542 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1543 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1544 				   struct scatterlist *);
1545 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1546 extern bool blk_integrity_merge_rq(struct request_queue *, struct request *,
1547 				   struct request *);
1548 extern bool blk_integrity_merge_bio(struct request_queue *, struct request *,
1549 				    struct bio *);
1550 
blk_get_integrity(struct gendisk * disk)1551 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1552 {
1553 	struct blk_integrity *bi = &disk->queue->integrity;
1554 
1555 	if (!bi->profile)
1556 		return NULL;
1557 
1558 	return bi;
1559 }
1560 
1561 static inline
bdev_get_integrity(struct block_device * bdev)1562 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1563 {
1564 	return blk_get_integrity(bdev->bd_disk);
1565 }
1566 
blk_integrity_rq(struct request * rq)1567 static inline bool blk_integrity_rq(struct request *rq)
1568 {
1569 	return rq->cmd_flags & REQ_INTEGRITY;
1570 }
1571 
blk_queue_max_integrity_segments(struct request_queue * q,unsigned int segs)1572 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1573 						    unsigned int segs)
1574 {
1575 	q->limits.max_integrity_segments = segs;
1576 }
1577 
1578 static inline unsigned short
queue_max_integrity_segments(const struct request_queue * q)1579 queue_max_integrity_segments(const struct request_queue *q)
1580 {
1581 	return q->limits.max_integrity_segments;
1582 }
1583 
1584 /**
1585  * bio_integrity_intervals - Return number of integrity intervals for a bio
1586  * @bi:		blk_integrity profile for device
1587  * @sectors:	Size of the bio in 512-byte sectors
1588  *
1589  * Description: The block layer calculates everything in 512 byte
1590  * sectors but integrity metadata is done in terms of the data integrity
1591  * interval size of the storage device.  Convert the block layer sectors
1592  * to the appropriate number of integrity intervals.
1593  */
bio_integrity_intervals(struct blk_integrity * bi,unsigned int sectors)1594 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1595 						   unsigned int sectors)
1596 {
1597 	return sectors >> (bi->interval_exp - 9);
1598 }
1599 
bio_integrity_bytes(struct blk_integrity * bi,unsigned int sectors)1600 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1601 					       unsigned int sectors)
1602 {
1603 	return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1604 }
1605 
1606 /*
1607  * Return the first bvec that contains integrity data.  Only drivers that are
1608  * limited to a single integrity segment should use this helper.
1609  */
rq_integrity_vec(struct request * rq)1610 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1611 {
1612 	if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1613 		return NULL;
1614 	return rq->bio->bi_integrity->bip_vec;
1615 }
1616 
1617 #else /* CONFIG_BLK_DEV_INTEGRITY */
1618 
1619 struct bio;
1620 struct block_device;
1621 struct gendisk;
1622 struct blk_integrity;
1623 
blk_integrity_rq(struct request * rq)1624 static inline int blk_integrity_rq(struct request *rq)
1625 {
1626 	return 0;
1627 }
blk_rq_count_integrity_sg(struct request_queue * q,struct bio * b)1628 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1629 					    struct bio *b)
1630 {
1631 	return 0;
1632 }
blk_rq_map_integrity_sg(struct request_queue * q,struct bio * b,struct scatterlist * s)1633 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1634 					  struct bio *b,
1635 					  struct scatterlist *s)
1636 {
1637 	return 0;
1638 }
bdev_get_integrity(struct block_device * b)1639 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1640 {
1641 	return NULL;
1642 }
blk_get_integrity(struct gendisk * disk)1643 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1644 {
1645 	return NULL;
1646 }
blk_integrity_compare(struct gendisk * a,struct gendisk * b)1647 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1648 {
1649 	return 0;
1650 }
blk_integrity_register(struct gendisk * d,struct blk_integrity * b)1651 static inline void blk_integrity_register(struct gendisk *d,
1652 					 struct blk_integrity *b)
1653 {
1654 }
blk_integrity_unregister(struct gendisk * d)1655 static inline void blk_integrity_unregister(struct gendisk *d)
1656 {
1657 }
blk_queue_max_integrity_segments(struct request_queue * q,unsigned int segs)1658 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1659 						    unsigned int segs)
1660 {
1661 }
queue_max_integrity_segments(const struct request_queue * q)1662 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1663 {
1664 	return 0;
1665 }
blk_integrity_merge_rq(struct request_queue * rq,struct request * r1,struct request * r2)1666 static inline bool blk_integrity_merge_rq(struct request_queue *rq,
1667 					  struct request *r1,
1668 					  struct request *r2)
1669 {
1670 	return true;
1671 }
blk_integrity_merge_bio(struct request_queue * rq,struct request * r,struct bio * b)1672 static inline bool blk_integrity_merge_bio(struct request_queue *rq,
1673 					   struct request *r,
1674 					   struct bio *b)
1675 {
1676 	return true;
1677 }
1678 
bio_integrity_intervals(struct blk_integrity * bi,unsigned int sectors)1679 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1680 						   unsigned int sectors)
1681 {
1682 	return 0;
1683 }
1684 
bio_integrity_bytes(struct blk_integrity * bi,unsigned int sectors)1685 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1686 					       unsigned int sectors)
1687 {
1688 	return 0;
1689 }
1690 
rq_integrity_vec(struct request * rq)1691 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1692 {
1693 	return NULL;
1694 }
1695 
1696 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1697 
1698 struct block_device_operations {
1699 	int (*open) (struct block_device *, fmode_t);
1700 	void (*release) (struct gendisk *, fmode_t);
1701 	int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1702 	int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1703 	int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1704 	unsigned int (*check_events) (struct gendisk *disk,
1705 				      unsigned int clearing);
1706 	/* ->media_changed() is DEPRECATED, use ->check_events() instead */
1707 	int (*media_changed) (struct gendisk *);
1708 	void (*unlock_native_capacity) (struct gendisk *);
1709 	int (*revalidate_disk) (struct gendisk *);
1710 	int (*getgeo)(struct block_device *, struct hd_geometry *);
1711 	/* this callback is with swap_lock and sometimes page table lock held */
1712 	void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1713 	int (*report_zones)(struct gendisk *, sector_t sector,
1714 			unsigned int nr_zones, report_zones_cb cb, void *data);
1715 	struct module *owner;
1716 	const struct pr_ops *pr_ops;
1717 };
1718 
1719 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1720 				 unsigned long);
1721 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1722 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1723 						struct writeback_control *);
1724 
1725 #ifdef CONFIG_BLK_DEV_ZONED
1726 bool blk_req_needs_zone_write_lock(struct request *rq);
1727 void __blk_req_zone_write_lock(struct request *rq);
1728 void __blk_req_zone_write_unlock(struct request *rq);
1729 
blk_req_zone_write_lock(struct request * rq)1730 static inline void blk_req_zone_write_lock(struct request *rq)
1731 {
1732 	if (blk_req_needs_zone_write_lock(rq))
1733 		__blk_req_zone_write_lock(rq);
1734 }
1735 
blk_req_zone_write_unlock(struct request * rq)1736 static inline void blk_req_zone_write_unlock(struct request *rq)
1737 {
1738 	if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1739 		__blk_req_zone_write_unlock(rq);
1740 }
1741 
blk_req_zone_is_write_locked(struct request * rq)1742 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1743 {
1744 	return rq->q->seq_zones_wlock &&
1745 		test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1746 }
1747 
blk_req_can_dispatch_to_zone(struct request * rq)1748 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1749 {
1750 	if (!blk_req_needs_zone_write_lock(rq))
1751 		return true;
1752 	return !blk_req_zone_is_write_locked(rq);
1753 }
1754 #else
blk_req_needs_zone_write_lock(struct request * rq)1755 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1756 {
1757 	return false;
1758 }
1759 
blk_req_zone_write_lock(struct request * rq)1760 static inline void blk_req_zone_write_lock(struct request *rq)
1761 {
1762 }
1763 
blk_req_zone_write_unlock(struct request * rq)1764 static inline void blk_req_zone_write_unlock(struct request *rq)
1765 {
1766 }
blk_req_zone_is_write_locked(struct request * rq)1767 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1768 {
1769 	return false;
1770 }
1771 
blk_req_can_dispatch_to_zone(struct request * rq)1772 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1773 {
1774 	return true;
1775 }
1776 #endif /* CONFIG_BLK_DEV_ZONED */
1777 
1778 #else /* CONFIG_BLOCK */
1779 
1780 struct block_device;
1781 
1782 /*
1783  * stubs for when the block layer is configured out
1784  */
1785 #define buffer_heads_over_limit 0
1786 
nr_blockdev_pages(void)1787 static inline long nr_blockdev_pages(void)
1788 {
1789 	return 0;
1790 }
1791 
1792 struct blk_plug {
1793 };
1794 
blk_start_plug(struct blk_plug * plug)1795 static inline void blk_start_plug(struct blk_plug *plug)
1796 {
1797 }
1798 
blk_finish_plug(struct blk_plug * plug)1799 static inline void blk_finish_plug(struct blk_plug *plug)
1800 {
1801 }
1802 
blk_flush_plug(struct task_struct * task)1803 static inline void blk_flush_plug(struct task_struct *task)
1804 {
1805 }
1806 
blk_schedule_flush_plug(struct task_struct * task)1807 static inline void blk_schedule_flush_plug(struct task_struct *task)
1808 {
1809 }
1810 
1811 
blk_needs_flush_plug(struct task_struct * tsk)1812 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1813 {
1814 	return false;
1815 }
1816 
blkdev_issue_flush(struct block_device * bdev,gfp_t gfp_mask,sector_t * error_sector)1817 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask,
1818 				     sector_t *error_sector)
1819 {
1820 	return 0;
1821 }
1822 
1823 #endif /* CONFIG_BLOCK */
1824 
blk_wake_io_task(struct task_struct * waiter)1825 static inline void blk_wake_io_task(struct task_struct *waiter)
1826 {
1827 	/*
1828 	 * If we're polling, the task itself is doing the completions. For
1829 	 * that case, we don't need to signal a wakeup, it's enough to just
1830 	 * mark us as RUNNING.
1831 	 */
1832 	if (waiter == current)
1833 		__set_current_state(TASK_RUNNING);
1834 	else
1835 		wake_up_process(waiter);
1836 }
1837 
1838 #endif
1839