1 | /* SPDX-License-Identifier: GPL-2.0 */ |
2 | /* |
3 | * Portions Copyright (C) 1992 Drew Eckhardt |
4 | */ |
5 | #ifndef _LINUX_BLKDEV_H |
6 | #define _LINUX_BLKDEV_H |
7 | |
8 | #include <linux/types.h> |
9 | #include <linux/blk_types.h> |
10 | #include <linux/device.h> |
11 | #include <linux/list.h> |
12 | #include <linux/llist.h> |
13 | #include <linux/minmax.h> |
14 | #include <linux/timer.h> |
15 | #include <linux/workqueue.h> |
16 | #include <linux/wait.h> |
17 | #include <linux/bio.h> |
18 | #include <linux/gfp.h> |
19 | #include <linux/kdev_t.h> |
20 | #include <linux/rcupdate.h> |
21 | #include <linux/percpu-refcount.h> |
22 | #include <linux/blkzoned.h> |
23 | #include <linux/sched.h> |
24 | #include <linux/sbitmap.h> |
25 | #include <linux/uuid.h> |
26 | #include <linux/xarray.h> |
27 | |
28 | struct module; |
29 | struct request_queue; |
30 | struct elevator_queue; |
31 | struct blk_trace; |
32 | struct request; |
33 | struct sg_io_hdr; |
34 | struct blkcg_gq; |
35 | struct blk_flush_queue; |
36 | struct kiocb; |
37 | struct pr_ops; |
38 | struct rq_qos; |
39 | struct blk_queue_stats; |
40 | struct blk_stat_callback; |
41 | struct blk_crypto_profile; |
42 | |
43 | extern const struct device_type disk_type; |
44 | extern const struct device_type part_type; |
45 | extern struct class block_class; |
46 | |
47 | /* |
48 | * Maximum number of blkcg policies allowed to be registered concurrently. |
49 | * Defined here to simplify include dependency. |
50 | */ |
51 | #define BLKCG_MAX_POLS 6 |
52 | |
53 | #define DISK_MAX_PARTS 256 |
54 | #define DISK_NAME_LEN 32 |
55 | |
56 | #define PARTITION_META_INFO_VOLNAMELTH 64 |
57 | /* |
58 | * Enough for the string representation of any kind of UUID plus NULL. |
59 | * EFI UUID is 36 characters. MSDOS UUID is 11 characters. |
60 | */ |
61 | #define PARTITION_META_INFO_UUIDLTH (UUID_STRING_LEN + 1) |
62 | |
63 | struct partition_meta_info { |
64 | char uuid[PARTITION_META_INFO_UUIDLTH]; |
65 | u8 volname[PARTITION_META_INFO_VOLNAMELTH]; |
66 | }; |
67 | |
68 | /** |
69 | * DOC: genhd capability flags |
70 | * |
71 | * ``GENHD_FL_REMOVABLE``: indicates that the block device gives access to |
72 | * removable media. When set, the device remains present even when media is not |
73 | * inserted. Shall not be set for devices which are removed entirely when the |
74 | * media is removed. |
75 | * |
76 | * ``GENHD_FL_HIDDEN``: the block device is hidden; it doesn't produce events, |
77 | * doesn't appear in sysfs, and can't be opened from userspace or using |
78 | * blkdev_get*. Used for the underlying components of multipath devices. |
79 | * |
80 | * ``GENHD_FL_NO_PART``: partition support is disabled. The kernel will not |
81 | * scan for partitions from add_disk, and users can't add partitions manually. |
82 | * |
83 | */ |
84 | enum { |
85 | GENHD_FL_REMOVABLE = 1 << 0, |
86 | GENHD_FL_HIDDEN = 1 << 1, |
87 | GENHD_FL_NO_PART = 1 << 2, |
88 | }; |
89 | |
90 | enum { |
91 | DISK_EVENT_MEDIA_CHANGE = 1 << 0, /* media changed */ |
92 | DISK_EVENT_EJECT_REQUEST = 1 << 1, /* eject requested */ |
93 | }; |
94 | |
95 | enum { |
96 | /* Poll even if events_poll_msecs is unset */ |
97 | DISK_EVENT_FLAG_POLL = 1 << 0, |
98 | /* Forward events to udev */ |
99 | DISK_EVENT_FLAG_UEVENT = 1 << 1, |
100 | /* Block event polling when open for exclusive write */ |
101 | DISK_EVENT_FLAG_BLOCK_ON_EXCL_WRITE = 1 << 2, |
102 | }; |
103 | |
104 | struct disk_events; |
105 | struct badblocks; |
106 | |
107 | struct blk_integrity { |
108 | const struct blk_integrity_profile *profile; |
109 | unsigned char flags; |
110 | unsigned char tuple_size; |
111 | unsigned char interval_exp; |
112 | unsigned char tag_size; |
113 | }; |
114 | |
115 | typedef unsigned int __bitwise blk_mode_t; |
116 | |
117 | /* open for reading */ |
118 | #define BLK_OPEN_READ ((__force blk_mode_t)(1 << 0)) |
119 | /* open for writing */ |
120 | #define BLK_OPEN_WRITE ((__force blk_mode_t)(1 << 1)) |
121 | /* open exclusively (vs other exclusive openers */ |
122 | #define BLK_OPEN_EXCL ((__force blk_mode_t)(1 << 2)) |
123 | /* opened with O_NDELAY */ |
124 | #define BLK_OPEN_NDELAY ((__force blk_mode_t)(1 << 3)) |
125 | /* open for "writes" only for ioctls (specialy hack for floppy.c) */ |
126 | #define BLK_OPEN_WRITE_IOCTL ((__force blk_mode_t)(1 << 4)) |
127 | |
128 | struct gendisk { |
129 | /* |
130 | * major/first_minor/minors should not be set by any new driver, the |
131 | * block core will take care of allocating them automatically. |
132 | */ |
133 | int major; |
134 | int first_minor; |
135 | int minors; |
136 | |
137 | char disk_name[DISK_NAME_LEN]; /* name of major driver */ |
138 | |
139 | unsigned short events; /* supported events */ |
140 | unsigned short event_flags; /* flags related to event processing */ |
141 | |
142 | struct xarray part_tbl; |
143 | struct block_device *part0; |
144 | |
145 | const struct block_device_operations *fops; |
146 | struct request_queue *queue; |
147 | void *private_data; |
148 | |
149 | struct bio_set bio_split; |
150 | |
151 | int flags; |
152 | unsigned long state; |
153 | #define GD_NEED_PART_SCAN 0 |
154 | #define GD_READ_ONLY 1 |
155 | #define GD_DEAD 2 |
156 | #define GD_NATIVE_CAPACITY 3 |
157 | #define GD_ADDED 4 |
158 | #define GD_SUPPRESS_PART_SCAN 5 |
159 | #define GD_OWNS_QUEUE 6 |
160 | |
161 | struct mutex open_mutex; /* open/close mutex */ |
162 | unsigned open_partitions; /* number of open partitions */ |
163 | |
164 | struct backing_dev_info *bdi; |
165 | struct kobject queue_kobj; /* the queue/ directory */ |
166 | struct kobject *slave_dir; |
167 | #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED |
168 | struct list_head slave_bdevs; |
169 | #endif |
170 | struct timer_rand_state *random; |
171 | atomic_t sync_io; /* RAID */ |
172 | struct disk_events *ev; |
173 | |
174 | #ifdef CONFIG_BLK_DEV_ZONED |
175 | /* |
176 | * Zoned block device information for request dispatch control. |
177 | * nr_zones is the total number of zones of the device. This is always |
178 | * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones |
179 | * bits which indicates if a zone is conventional (bit set) or |
180 | * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones |
181 | * bits which indicates if a zone is write locked, that is, if a write |
182 | * request targeting the zone was dispatched. |
183 | * |
184 | * Reads of this information must be protected with blk_queue_enter() / |
185 | * blk_queue_exit(). Modifying this information is only allowed while |
186 | * no requests are being processed. See also blk_mq_freeze_queue() and |
187 | * blk_mq_unfreeze_queue(). |
188 | */ |
189 | unsigned int nr_zones; |
190 | unsigned int max_open_zones; |
191 | unsigned int max_active_zones; |
192 | unsigned long *conv_zones_bitmap; |
193 | unsigned long *seq_zones_wlock; |
194 | #endif /* CONFIG_BLK_DEV_ZONED */ |
195 | |
196 | #if IS_ENABLED(CONFIG_CDROM) |
197 | struct cdrom_device_info *cdi; |
198 | #endif |
199 | int node_id; |
200 | struct badblocks *bb; |
201 | struct lockdep_map lockdep_map; |
202 | u64 diskseq; |
203 | blk_mode_t open_mode; |
204 | |
205 | /* |
206 | * Independent sector access ranges. This is always NULL for |
207 | * devices that do not have multiple independent access ranges. |
208 | */ |
209 | struct blk_independent_access_ranges *ia_ranges; |
210 | }; |
211 | |
212 | static inline bool disk_live(struct gendisk *disk) |
213 | { |
214 | return !inode_unhashed(inode: disk->part0->bd_inode); |
215 | } |
216 | |
217 | /** |
218 | * disk_openers - returns how many openers are there for a disk |
219 | * @disk: disk to check |
220 | * |
221 | * This returns the number of openers for a disk. Note that this value is only |
222 | * stable if disk->open_mutex is held. |
223 | * |
224 | * Note: Due to a quirk in the block layer open code, each open partition is |
225 | * only counted once even if there are multiple openers. |
226 | */ |
227 | static inline unsigned int disk_openers(struct gendisk *disk) |
228 | { |
229 | return atomic_read(v: &disk->part0->bd_openers); |
230 | } |
231 | |
232 | /* |
233 | * The gendisk is refcounted by the part0 block_device, and the bd_device |
234 | * therein is also used for device model presentation in sysfs. |
235 | */ |
236 | #define dev_to_disk(device) \ |
237 | (dev_to_bdev(device)->bd_disk) |
238 | #define disk_to_dev(disk) \ |
239 | (&((disk)->part0->bd_device)) |
240 | |
241 | #if IS_REACHABLE(CONFIG_CDROM) |
242 | #define disk_to_cdi(disk) ((disk)->cdi) |
243 | #else |
244 | #define disk_to_cdi(disk) NULL |
245 | #endif |
246 | |
247 | static inline dev_t disk_devt(struct gendisk *disk) |
248 | { |
249 | return MKDEV(disk->major, disk->first_minor); |
250 | } |
251 | |
252 | static inline int blk_validate_block_size(unsigned long bsize) |
253 | { |
254 | if (bsize < 512 || bsize > PAGE_SIZE || !is_power_of_2(n: bsize)) |
255 | return -EINVAL; |
256 | |
257 | return 0; |
258 | } |
259 | |
260 | static inline bool blk_op_is_passthrough(blk_opf_t op) |
261 | { |
262 | op &= REQ_OP_MASK; |
263 | return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT; |
264 | } |
265 | |
266 | /* |
267 | * Zoned block device models (zoned limit). |
268 | * |
269 | * Note: This needs to be ordered from the least to the most severe |
270 | * restrictions for the inheritance in blk_stack_limits() to work. |
271 | */ |
272 | enum blk_zoned_model { |
273 | BLK_ZONED_NONE = 0, /* Regular block device */ |
274 | BLK_ZONED_HA, /* Host-aware zoned block device */ |
275 | BLK_ZONED_HM, /* Host-managed zoned block device */ |
276 | }; |
277 | |
278 | /* |
279 | * BLK_BOUNCE_NONE: never bounce (default) |
280 | * BLK_BOUNCE_HIGH: bounce all highmem pages |
281 | */ |
282 | enum blk_bounce { |
283 | BLK_BOUNCE_NONE, |
284 | BLK_BOUNCE_HIGH, |
285 | }; |
286 | |
287 | struct queue_limits { |
288 | enum blk_bounce bounce; |
289 | unsigned long seg_boundary_mask; |
290 | unsigned long virt_boundary_mask; |
291 | |
292 | unsigned int max_hw_sectors; |
293 | unsigned int max_dev_sectors; |
294 | unsigned int chunk_sectors; |
295 | unsigned int max_sectors; |
296 | unsigned int max_user_sectors; |
297 | unsigned int max_segment_size; |
298 | unsigned int physical_block_size; |
299 | unsigned int logical_block_size; |
300 | unsigned int alignment_offset; |
301 | unsigned int io_min; |
302 | unsigned int io_opt; |
303 | unsigned int max_discard_sectors; |
304 | unsigned int max_hw_discard_sectors; |
305 | unsigned int max_secure_erase_sectors; |
306 | unsigned int max_write_zeroes_sectors; |
307 | unsigned int max_zone_append_sectors; |
308 | unsigned int discard_granularity; |
309 | unsigned int discard_alignment; |
310 | unsigned int zone_write_granularity; |
311 | |
312 | unsigned short max_segments; |
313 | unsigned short max_integrity_segments; |
314 | unsigned short max_discard_segments; |
315 | |
316 | unsigned char misaligned; |
317 | unsigned char discard_misaligned; |
318 | unsigned char raid_partial_stripes_expensive; |
319 | enum blk_zoned_model zoned; |
320 | |
321 | /* |
322 | * Drivers that set dma_alignment to less than 511 must be prepared to |
323 | * handle individual bvec's that are not a multiple of a SECTOR_SIZE |
324 | * due to possible offsets. |
325 | */ |
326 | unsigned int dma_alignment; |
327 | }; |
328 | |
329 | typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx, |
330 | void *data); |
331 | |
332 | void disk_set_zoned(struct gendisk *disk, enum blk_zoned_model model); |
333 | |
334 | #ifdef CONFIG_BLK_DEV_ZONED |
335 | #define BLK_ALL_ZONES ((unsigned int)-1) |
336 | int blkdev_report_zones(struct block_device *bdev, sector_t sector, |
337 | unsigned int nr_zones, report_zones_cb cb, void *data); |
338 | unsigned int bdev_nr_zones(struct block_device *bdev); |
339 | extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_op op, |
340 | sector_t sectors, sector_t nr_sectors, |
341 | gfp_t gfp_mask); |
342 | int blk_revalidate_disk_zones(struct gendisk *disk, |
343 | void (*update_driver_data)(struct gendisk *disk)); |
344 | #else /* CONFIG_BLK_DEV_ZONED */ |
345 | static inline unsigned int bdev_nr_zones(struct block_device *bdev) |
346 | { |
347 | return 0; |
348 | } |
349 | #endif /* CONFIG_BLK_DEV_ZONED */ |
350 | |
351 | /* |
352 | * Independent access ranges: struct blk_independent_access_range describes |
353 | * a range of contiguous sectors that can be accessed using device command |
354 | * execution resources that are independent from the resources used for |
355 | * other access ranges. This is typically found with single-LUN multi-actuator |
356 | * HDDs where each access range is served by a different set of heads. |
357 | * The set of independent ranges supported by the device is defined using |
358 | * struct blk_independent_access_ranges. The independent ranges must not overlap |
359 | * and must include all sectors within the disk capacity (no sector holes |
360 | * allowed). |
361 | * For a device with multiple ranges, requests targeting sectors in different |
362 | * ranges can be executed in parallel. A request can straddle an access range |
363 | * boundary. |
364 | */ |
365 | struct blk_independent_access_range { |
366 | struct kobject kobj; |
367 | sector_t sector; |
368 | sector_t nr_sectors; |
369 | }; |
370 | |
371 | struct blk_independent_access_ranges { |
372 | struct kobject kobj; |
373 | bool sysfs_registered; |
374 | unsigned int nr_ia_ranges; |
375 | struct blk_independent_access_range ia_range[]; |
376 | }; |
377 | |
378 | struct request_queue { |
379 | struct request *last_merge; |
380 | struct elevator_queue *elevator; |
381 | |
382 | struct percpu_ref q_usage_counter; |
383 | |
384 | struct blk_queue_stats *stats; |
385 | struct rq_qos *rq_qos; |
386 | struct mutex rq_qos_mutex; |
387 | |
388 | const struct blk_mq_ops *mq_ops; |
389 | |
390 | /* sw queues */ |
391 | struct blk_mq_ctx __percpu *queue_ctx; |
392 | |
393 | unsigned int queue_depth; |
394 | |
395 | /* hw dispatch queues */ |
396 | struct xarray hctx_table; |
397 | unsigned int nr_hw_queues; |
398 | |
399 | /* |
400 | * The queue owner gets to use this for whatever they like. |
401 | * ll_rw_blk doesn't touch it. |
402 | */ |
403 | void *queuedata; |
404 | |
405 | /* |
406 | * various queue flags, see QUEUE_* below |
407 | */ |
408 | unsigned long queue_flags; |
409 | /* |
410 | * Number of contexts that have called blk_set_pm_only(). If this |
411 | * counter is above zero then only RQF_PM requests are processed. |
412 | */ |
413 | atomic_t pm_only; |
414 | |
415 | /* |
416 | * ida allocated id for this queue. Used to index queues from |
417 | * ioctx. |
418 | */ |
419 | int id; |
420 | |
421 | spinlock_t queue_lock; |
422 | |
423 | struct gendisk *disk; |
424 | |
425 | refcount_t refs; |
426 | |
427 | /* |
428 | * mq queue kobject |
429 | */ |
430 | struct kobject *mq_kobj; |
431 | |
432 | #ifdef CONFIG_BLK_DEV_INTEGRITY |
433 | struct blk_integrity integrity; |
434 | #endif /* CONFIG_BLK_DEV_INTEGRITY */ |
435 | |
436 | #ifdef CONFIG_PM |
437 | struct device *dev; |
438 | enum rpm_status rpm_status; |
439 | #endif |
440 | |
441 | /* |
442 | * queue settings |
443 | */ |
444 | unsigned long nr_requests; /* Max # of requests */ |
445 | |
446 | unsigned int dma_pad_mask; |
447 | |
448 | #ifdef CONFIG_BLK_INLINE_ENCRYPTION |
449 | struct blk_crypto_profile *crypto_profile; |
450 | struct kobject *crypto_kobject; |
451 | #endif |
452 | |
453 | unsigned int rq_timeout; |
454 | |
455 | struct timer_list timeout; |
456 | struct work_struct timeout_work; |
457 | |
458 | atomic_t nr_active_requests_shared_tags; |
459 | |
460 | struct blk_mq_tags *sched_shared_tags; |
461 | |
462 | struct list_head icq_list; |
463 | #ifdef CONFIG_BLK_CGROUP |
464 | DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS); |
465 | struct blkcg_gq *root_blkg; |
466 | struct list_head blkg_list; |
467 | struct mutex blkcg_mutex; |
468 | #endif |
469 | |
470 | struct queue_limits limits; |
471 | |
472 | unsigned int required_elevator_features; |
473 | |
474 | int node; |
475 | #ifdef CONFIG_BLK_DEV_IO_TRACE |
476 | struct blk_trace __rcu *blk_trace; |
477 | #endif |
478 | /* |
479 | * for flush operations |
480 | */ |
481 | struct blk_flush_queue *fq; |
482 | struct list_head flush_list; |
483 | |
484 | struct list_head requeue_list; |
485 | spinlock_t requeue_lock; |
486 | struct delayed_work requeue_work; |
487 | |
488 | struct mutex sysfs_lock; |
489 | struct mutex sysfs_dir_lock; |
490 | |
491 | /* |
492 | * for reusing dead hctx instance in case of updating |
493 | * nr_hw_queues |
494 | */ |
495 | struct list_head unused_hctx_list; |
496 | spinlock_t unused_hctx_lock; |
497 | |
498 | int mq_freeze_depth; |
499 | |
500 | #ifdef CONFIG_BLK_DEV_THROTTLING |
501 | /* Throttle data */ |
502 | struct throtl_data *td; |
503 | #endif |
504 | struct rcu_head rcu_head; |
505 | wait_queue_head_t mq_freeze_wq; |
506 | /* |
507 | * Protect concurrent access to q_usage_counter by |
508 | * percpu_ref_kill() and percpu_ref_reinit(). |
509 | */ |
510 | struct mutex mq_freeze_lock; |
511 | |
512 | int quiesce_depth; |
513 | |
514 | struct blk_mq_tag_set *tag_set; |
515 | struct list_head tag_set_list; |
516 | |
517 | struct dentry *debugfs_dir; |
518 | struct dentry *sched_debugfs_dir; |
519 | struct dentry *rqos_debugfs_dir; |
520 | /* |
521 | * Serializes all debugfs metadata operations using the above dentries. |
522 | */ |
523 | struct mutex debugfs_mutex; |
524 | |
525 | bool mq_sysfs_init_done; |
526 | }; |
527 | |
528 | /* Keep blk_queue_flag_name[] in sync with the definitions below */ |
529 | #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */ |
530 | #define QUEUE_FLAG_DYING 1 /* queue being torn down */ |
531 | #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */ |
532 | #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */ |
533 | #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */ |
534 | #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */ |
535 | #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */ |
536 | #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */ |
537 | #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */ |
538 | #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */ |
539 | #define QUEUE_FLAG_SYNCHRONOUS 11 /* always completes in submit context */ |
540 | #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */ |
541 | #define QUEUE_FLAG_HW_WC 18 /* Write back caching supported */ |
542 | #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */ |
543 | #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */ |
544 | #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */ |
545 | #define QUEUE_FLAG_WC 17 /* Write back caching */ |
546 | #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */ |
547 | #define QUEUE_FLAG_DAX 19 /* device supports DAX */ |
548 | #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */ |
549 | #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */ |
550 | #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */ |
551 | #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */ |
552 | #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */ |
553 | #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */ |
554 | #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */ |
555 | #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */ |
556 | #define QUEUE_FLAG_SQ_SCHED 30 /* single queue style io dispatch */ |
557 | #define QUEUE_FLAG_SKIP_TAGSET_QUIESCE 31 /* quiesce_tagset skip the queue*/ |
558 | |
559 | #define QUEUE_FLAG_MQ_DEFAULT ((1UL << QUEUE_FLAG_IO_STAT) | \ |
560 | (1UL << QUEUE_FLAG_SAME_COMP) | \ |
561 | (1UL << QUEUE_FLAG_NOWAIT)) |
562 | |
563 | void blk_queue_flag_set(unsigned int flag, struct request_queue *q); |
564 | void blk_queue_flag_clear(unsigned int flag, struct request_queue *q); |
565 | bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q); |
566 | |
567 | #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags) |
568 | #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags) |
569 | #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags) |
570 | #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags) |
571 | #define blk_queue_noxmerges(q) \ |
572 | test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags) |
573 | #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags) |
574 | #define blk_queue_stable_writes(q) \ |
575 | test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags) |
576 | #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags) |
577 | #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags) |
578 | #define blk_queue_zone_resetall(q) \ |
579 | test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags) |
580 | #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags) |
581 | #define blk_queue_pci_p2pdma(q) \ |
582 | test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags) |
583 | #ifdef CONFIG_BLK_RQ_ALLOC_TIME |
584 | #define blk_queue_rq_alloc_time(q) \ |
585 | test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags) |
586 | #else |
587 | #define blk_queue_rq_alloc_time(q) false |
588 | #endif |
589 | |
590 | #define blk_noretry_request(rq) \ |
591 | ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \ |
592 | REQ_FAILFAST_DRIVER)) |
593 | #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags) |
594 | #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only) |
595 | #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags) |
596 | #define blk_queue_sq_sched(q) test_bit(QUEUE_FLAG_SQ_SCHED, &(q)->queue_flags) |
597 | #define blk_queue_skip_tagset_quiesce(q) \ |
598 | test_bit(QUEUE_FLAG_SKIP_TAGSET_QUIESCE, &(q)->queue_flags) |
599 | |
600 | extern void blk_set_pm_only(struct request_queue *q); |
601 | extern void blk_clear_pm_only(struct request_queue *q); |
602 | |
603 | #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist) |
604 | |
605 | #define dma_map_bvec(dev, bv, dir, attrs) \ |
606 | dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \ |
607 | (dir), (attrs)) |
608 | |
609 | static inline bool queue_is_mq(struct request_queue *q) |
610 | { |
611 | return q->mq_ops; |
612 | } |
613 | |
614 | #ifdef CONFIG_PM |
615 | static inline enum rpm_status queue_rpm_status(struct request_queue *q) |
616 | { |
617 | return q->rpm_status; |
618 | } |
619 | #else |
620 | static inline enum rpm_status queue_rpm_status(struct request_queue *q) |
621 | { |
622 | return RPM_ACTIVE; |
623 | } |
624 | #endif |
625 | |
626 | static inline enum blk_zoned_model |
627 | blk_queue_zoned_model(struct request_queue *q) |
628 | { |
629 | if (IS_ENABLED(CONFIG_BLK_DEV_ZONED)) |
630 | return q->limits.zoned; |
631 | return BLK_ZONED_NONE; |
632 | } |
633 | |
634 | static inline bool blk_queue_is_zoned(struct request_queue *q) |
635 | { |
636 | switch (blk_queue_zoned_model(q)) { |
637 | case BLK_ZONED_HA: |
638 | case BLK_ZONED_HM: |
639 | return true; |
640 | default: |
641 | return false; |
642 | } |
643 | } |
644 | |
645 | #ifdef CONFIG_BLK_DEV_ZONED |
646 | static inline unsigned int disk_nr_zones(struct gendisk *disk) |
647 | { |
648 | return blk_queue_is_zoned(q: disk->queue) ? disk->nr_zones : 0; |
649 | } |
650 | |
651 | static inline unsigned int disk_zone_no(struct gendisk *disk, sector_t sector) |
652 | { |
653 | if (!blk_queue_is_zoned(q: disk->queue)) |
654 | return 0; |
655 | return sector >> ilog2(disk->queue->limits.chunk_sectors); |
656 | } |
657 | |
658 | static inline bool disk_zone_is_seq(struct gendisk *disk, sector_t sector) |
659 | { |
660 | if (!blk_queue_is_zoned(q: disk->queue)) |
661 | return false; |
662 | if (!disk->conv_zones_bitmap) |
663 | return true; |
664 | return !test_bit(disk_zone_no(disk, sector), disk->conv_zones_bitmap); |
665 | } |
666 | |
667 | static inline void disk_set_max_open_zones(struct gendisk *disk, |
668 | unsigned int max_open_zones) |
669 | { |
670 | disk->max_open_zones = max_open_zones; |
671 | } |
672 | |
673 | static inline void disk_set_max_active_zones(struct gendisk *disk, |
674 | unsigned int max_active_zones) |
675 | { |
676 | disk->max_active_zones = max_active_zones; |
677 | } |
678 | |
679 | static inline unsigned int bdev_max_open_zones(struct block_device *bdev) |
680 | { |
681 | return bdev->bd_disk->max_open_zones; |
682 | } |
683 | |
684 | static inline unsigned int bdev_max_active_zones(struct block_device *bdev) |
685 | { |
686 | return bdev->bd_disk->max_active_zones; |
687 | } |
688 | |
689 | #else /* CONFIG_BLK_DEV_ZONED */ |
690 | static inline unsigned int disk_nr_zones(struct gendisk *disk) |
691 | { |
692 | return 0; |
693 | } |
694 | static inline bool disk_zone_is_seq(struct gendisk *disk, sector_t sector) |
695 | { |
696 | return false; |
697 | } |
698 | static inline unsigned int disk_zone_no(struct gendisk *disk, sector_t sector) |
699 | { |
700 | return 0; |
701 | } |
702 | static inline unsigned int bdev_max_open_zones(struct block_device *bdev) |
703 | { |
704 | return 0; |
705 | } |
706 | |
707 | static inline unsigned int bdev_max_active_zones(struct block_device *bdev) |
708 | { |
709 | return 0; |
710 | } |
711 | #endif /* CONFIG_BLK_DEV_ZONED */ |
712 | |
713 | static inline unsigned int blk_queue_depth(struct request_queue *q) |
714 | { |
715 | if (q->queue_depth) |
716 | return q->queue_depth; |
717 | |
718 | return q->nr_requests; |
719 | } |
720 | |
721 | /* |
722 | * default timeout for SG_IO if none specified |
723 | */ |
724 | #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ) |
725 | #define BLK_MIN_SG_TIMEOUT (7 * HZ) |
726 | |
727 | /* This should not be used directly - use rq_for_each_segment */ |
728 | #define for_each_bio(_bio) \ |
729 | for (; _bio; _bio = _bio->bi_next) |
730 | |
731 | int __must_check device_add_disk(struct device *parent, struct gendisk *disk, |
732 | const struct attribute_group **groups); |
733 | static inline int __must_check add_disk(struct gendisk *disk) |
734 | { |
735 | return device_add_disk(NULL, disk, NULL); |
736 | } |
737 | void del_gendisk(struct gendisk *gp); |
738 | void invalidate_disk(struct gendisk *disk); |
739 | void set_disk_ro(struct gendisk *disk, bool read_only); |
740 | void disk_uevent(struct gendisk *disk, enum kobject_action action); |
741 | |
742 | static inline int get_disk_ro(struct gendisk *disk) |
743 | { |
744 | return disk->part0->bd_read_only || |
745 | test_bit(GD_READ_ONLY, &disk->state); |
746 | } |
747 | |
748 | static inline int bdev_read_only(struct block_device *bdev) |
749 | { |
750 | return bdev->bd_read_only || get_disk_ro(disk: bdev->bd_disk); |
751 | } |
752 | |
753 | bool set_capacity_and_notify(struct gendisk *disk, sector_t size); |
754 | void disk_force_media_change(struct gendisk *disk); |
755 | void bdev_mark_dead(struct block_device *bdev, bool surprise); |
756 | |
757 | void add_disk_randomness(struct gendisk *disk) __latent_entropy; |
758 | void rand_initialize_disk(struct gendisk *disk); |
759 | |
760 | static inline sector_t get_start_sect(struct block_device *bdev) |
761 | { |
762 | return bdev->bd_start_sect; |
763 | } |
764 | |
765 | static inline sector_t bdev_nr_sectors(struct block_device *bdev) |
766 | { |
767 | return bdev->bd_nr_sectors; |
768 | } |
769 | |
770 | static inline loff_t bdev_nr_bytes(struct block_device *bdev) |
771 | { |
772 | return (loff_t)bdev_nr_sectors(bdev) << SECTOR_SHIFT; |
773 | } |
774 | |
775 | static inline sector_t get_capacity(struct gendisk *disk) |
776 | { |
777 | return bdev_nr_sectors(bdev: disk->part0); |
778 | } |
779 | |
780 | static inline u64 sb_bdev_nr_blocks(struct super_block *sb) |
781 | { |
782 | return bdev_nr_sectors(bdev: sb->s_bdev) >> |
783 | (sb->s_blocksize_bits - SECTOR_SHIFT); |
784 | } |
785 | |
786 | int bdev_disk_changed(struct gendisk *disk, bool invalidate); |
787 | |
788 | void put_disk(struct gendisk *disk); |
789 | struct gendisk *__blk_alloc_disk(int node, struct lock_class_key *lkclass); |
790 | |
791 | /** |
792 | * blk_alloc_disk - allocate a gendisk structure |
793 | * @node_id: numa node to allocate on |
794 | * |
795 | * Allocate and pre-initialize a gendisk structure for use with BIO based |
796 | * drivers. |
797 | * |
798 | * Context: can sleep |
799 | */ |
800 | #define blk_alloc_disk(node_id) \ |
801 | ({ \ |
802 | static struct lock_class_key __key; \ |
803 | \ |
804 | __blk_alloc_disk(node_id, &__key); \ |
805 | }) |
806 | |
807 | int __register_blkdev(unsigned int major, const char *name, |
808 | void (*probe)(dev_t devt)); |
809 | #define register_blkdev(major, name) \ |
810 | __register_blkdev(major, name, NULL) |
811 | void unregister_blkdev(unsigned int major, const char *name); |
812 | |
813 | bool disk_check_media_change(struct gendisk *disk); |
814 | void set_capacity(struct gendisk *disk, sector_t size); |
815 | |
816 | #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED |
817 | int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk); |
818 | void bd_unlink_disk_holder(struct block_device *bdev, struct gendisk *disk); |
819 | #else |
820 | static inline int bd_link_disk_holder(struct block_device *bdev, |
821 | struct gendisk *disk) |
822 | { |
823 | return 0; |
824 | } |
825 | static inline void bd_unlink_disk_holder(struct block_device *bdev, |
826 | struct gendisk *disk) |
827 | { |
828 | } |
829 | #endif /* CONFIG_BLOCK_HOLDER_DEPRECATED */ |
830 | |
831 | dev_t part_devt(struct gendisk *disk, u8 partno); |
832 | void inc_diskseq(struct gendisk *disk); |
833 | void blk_request_module(dev_t devt); |
834 | |
835 | extern int blk_register_queue(struct gendisk *disk); |
836 | extern void blk_unregister_queue(struct gendisk *disk); |
837 | void submit_bio_noacct(struct bio *bio); |
838 | struct bio *bio_split_to_limits(struct bio *bio); |
839 | |
840 | extern int blk_lld_busy(struct request_queue *q); |
841 | extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags); |
842 | extern void blk_queue_exit(struct request_queue *q); |
843 | extern void blk_sync_queue(struct request_queue *q); |
844 | |
845 | /* Helper to convert REQ_OP_XXX to its string format XXX */ |
846 | extern const char *blk_op_str(enum req_op op); |
847 | |
848 | int blk_status_to_errno(blk_status_t status); |
849 | blk_status_t errno_to_blk_status(int errno); |
850 | const char *blk_status_to_str(blk_status_t status); |
851 | |
852 | /* only poll the hardware once, don't continue until a completion was found */ |
853 | #define BLK_POLL_ONESHOT (1 << 0) |
854 | int bio_poll(struct bio *bio, struct io_comp_batch *iob, unsigned int flags); |
855 | int iocb_bio_iopoll(struct kiocb *kiocb, struct io_comp_batch *iob, |
856 | unsigned int flags); |
857 | |
858 | static inline struct request_queue *bdev_get_queue(struct block_device *bdev) |
859 | { |
860 | return bdev->bd_queue; /* this is never NULL */ |
861 | } |
862 | |
863 | /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */ |
864 | const char *blk_zone_cond_str(enum blk_zone_cond zone_cond); |
865 | |
866 | static inline unsigned int bio_zone_no(struct bio *bio) |
867 | { |
868 | return disk_zone_no(disk: bio->bi_bdev->bd_disk, sector: bio->bi_iter.bi_sector); |
869 | } |
870 | |
871 | static inline unsigned int bio_zone_is_seq(struct bio *bio) |
872 | { |
873 | return disk_zone_is_seq(disk: bio->bi_bdev->bd_disk, sector: bio->bi_iter.bi_sector); |
874 | } |
875 | |
876 | /* |
877 | * Return how much of the chunk is left to be used for I/O at a given offset. |
878 | */ |
879 | static inline unsigned int blk_chunk_sectors_left(sector_t offset, |
880 | unsigned int chunk_sectors) |
881 | { |
882 | if (unlikely(!is_power_of_2(chunk_sectors))) |
883 | return chunk_sectors - sector_div(offset, chunk_sectors); |
884 | return chunk_sectors - (offset & (chunk_sectors - 1)); |
885 | } |
886 | |
887 | /* |
888 | * Access functions for manipulating queue properties |
889 | */ |
890 | void blk_queue_bounce_limit(struct request_queue *q, enum blk_bounce limit); |
891 | extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int); |
892 | extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int); |
893 | extern void blk_queue_max_segments(struct request_queue *, unsigned short); |
894 | extern void blk_queue_max_discard_segments(struct request_queue *, |
895 | unsigned short); |
896 | void blk_queue_max_secure_erase_sectors(struct request_queue *q, |
897 | unsigned int max_sectors); |
898 | extern void blk_queue_max_segment_size(struct request_queue *, unsigned int); |
899 | extern void blk_queue_max_discard_sectors(struct request_queue *q, |
900 | unsigned int max_discard_sectors); |
901 | extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q, |
902 | unsigned int max_write_same_sectors); |
903 | extern void blk_queue_logical_block_size(struct request_queue *, unsigned int); |
904 | extern void blk_queue_max_zone_append_sectors(struct request_queue *q, |
905 | unsigned int max_zone_append_sectors); |
906 | extern void blk_queue_physical_block_size(struct request_queue *, unsigned int); |
907 | void blk_queue_zone_write_granularity(struct request_queue *q, |
908 | unsigned int size); |
909 | extern void blk_queue_alignment_offset(struct request_queue *q, |
910 | unsigned int alignment); |
911 | void disk_update_readahead(struct gendisk *disk); |
912 | extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min); |
913 | extern void blk_queue_io_min(struct request_queue *q, unsigned int min); |
914 | extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt); |
915 | extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt); |
916 | extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth); |
917 | extern void blk_set_stacking_limits(struct queue_limits *lim); |
918 | extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b, |
919 | sector_t offset); |
920 | extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev, |
921 | sector_t offset); |
922 | extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int); |
923 | extern void blk_queue_segment_boundary(struct request_queue *, unsigned long); |
924 | extern void blk_queue_virt_boundary(struct request_queue *, unsigned long); |
925 | extern void blk_queue_dma_alignment(struct request_queue *, int); |
926 | extern void blk_queue_update_dma_alignment(struct request_queue *, int); |
927 | extern void blk_queue_rq_timeout(struct request_queue *, unsigned int); |
928 | extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua); |
929 | |
930 | struct blk_independent_access_ranges * |
931 | disk_alloc_independent_access_ranges(struct gendisk *disk, int nr_ia_ranges); |
932 | void disk_set_independent_access_ranges(struct gendisk *disk, |
933 | struct blk_independent_access_ranges *iars); |
934 | |
935 | /* |
936 | * Elevator features for blk_queue_required_elevator_features: |
937 | */ |
938 | /* Supports zoned block devices sequential write constraint */ |
939 | #define ELEVATOR_F_ZBD_SEQ_WRITE (1U << 0) |
940 | |
941 | extern void blk_queue_required_elevator_features(struct request_queue *q, |
942 | unsigned int features); |
943 | extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q, |
944 | struct device *dev); |
945 | |
946 | bool __must_check blk_get_queue(struct request_queue *); |
947 | extern void blk_put_queue(struct request_queue *); |
948 | |
949 | void blk_mark_disk_dead(struct gendisk *disk); |
950 | |
951 | #ifdef CONFIG_BLOCK |
952 | /* |
953 | * blk_plug permits building a queue of related requests by holding the I/O |
954 | * fragments for a short period. This allows merging of sequential requests |
955 | * into single larger request. As the requests are moved from a per-task list to |
956 | * the device's request_queue in a batch, this results in improved scalability |
957 | * as the lock contention for request_queue lock is reduced. |
958 | * |
959 | * It is ok not to disable preemption when adding the request to the plug list |
960 | * or when attempting a merge. For details, please see schedule() where |
961 | * blk_flush_plug() is called. |
962 | */ |
963 | struct blk_plug { |
964 | struct request *mq_list; /* blk-mq requests */ |
965 | |
966 | /* if ios_left is > 1, we can batch tag/rq allocations */ |
967 | struct request *cached_rq; |
968 | unsigned short nr_ios; |
969 | |
970 | unsigned short rq_count; |
971 | |
972 | bool multiple_queues; |
973 | bool has_elevator; |
974 | |
975 | struct list_head cb_list; /* md requires an unplug callback */ |
976 | }; |
977 | |
978 | struct blk_plug_cb; |
979 | typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool); |
980 | struct blk_plug_cb { |
981 | struct list_head list; |
982 | blk_plug_cb_fn callback; |
983 | void *data; |
984 | }; |
985 | extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug, |
986 | void *data, int size); |
987 | extern void blk_start_plug(struct blk_plug *); |
988 | extern void blk_start_plug_nr_ios(struct blk_plug *, unsigned short); |
989 | extern void blk_finish_plug(struct blk_plug *); |
990 | |
991 | void __blk_flush_plug(struct blk_plug *plug, bool from_schedule); |
992 | static inline void blk_flush_plug(struct blk_plug *plug, bool async) |
993 | { |
994 | if (plug) |
995 | __blk_flush_plug(plug, from_schedule: async); |
996 | } |
997 | |
998 | int blkdev_issue_flush(struct block_device *bdev); |
999 | long nr_blockdev_pages(void); |
1000 | #else /* CONFIG_BLOCK */ |
1001 | struct blk_plug { |
1002 | }; |
1003 | |
1004 | static inline void blk_start_plug_nr_ios(struct blk_plug *plug, |
1005 | unsigned short nr_ios) |
1006 | { |
1007 | } |
1008 | |
1009 | static inline void blk_start_plug(struct blk_plug *plug) |
1010 | { |
1011 | } |
1012 | |
1013 | static inline void blk_finish_plug(struct blk_plug *plug) |
1014 | { |
1015 | } |
1016 | |
1017 | static inline void blk_flush_plug(struct blk_plug *plug, bool async) |
1018 | { |
1019 | } |
1020 | |
1021 | static inline int blkdev_issue_flush(struct block_device *bdev) |
1022 | { |
1023 | return 0; |
1024 | } |
1025 | |
1026 | static inline long nr_blockdev_pages(void) |
1027 | { |
1028 | return 0; |
1029 | } |
1030 | #endif /* CONFIG_BLOCK */ |
1031 | |
1032 | extern void blk_io_schedule(void); |
1033 | |
1034 | int blkdev_issue_discard(struct block_device *bdev, sector_t sector, |
1035 | sector_t nr_sects, gfp_t gfp_mask); |
1036 | int __blkdev_issue_discard(struct block_device *bdev, sector_t sector, |
1037 | sector_t nr_sects, gfp_t gfp_mask, struct bio **biop); |
1038 | int blkdev_issue_secure_erase(struct block_device *bdev, sector_t sector, |
1039 | sector_t nr_sects, gfp_t gfp); |
1040 | |
1041 | #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */ |
1042 | #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */ |
1043 | |
1044 | extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector, |
1045 | sector_t nr_sects, gfp_t gfp_mask, struct bio **biop, |
1046 | unsigned flags); |
1047 | extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector, |
1048 | sector_t nr_sects, gfp_t gfp_mask, unsigned flags); |
1049 | |
1050 | static inline int sb_issue_discard(struct super_block *sb, sector_t block, |
1051 | sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags) |
1052 | { |
1053 | return blkdev_issue_discard(bdev: sb->s_bdev, |
1054 | sector: block << (sb->s_blocksize_bits - |
1055 | SECTOR_SHIFT), |
1056 | nr_sects: nr_blocks << (sb->s_blocksize_bits - |
1057 | SECTOR_SHIFT), |
1058 | gfp_mask); |
1059 | } |
1060 | static inline int sb_issue_zeroout(struct super_block *sb, sector_t block, |
1061 | sector_t nr_blocks, gfp_t gfp_mask) |
1062 | { |
1063 | return blkdev_issue_zeroout(bdev: sb->s_bdev, |
1064 | sector: block << (sb->s_blocksize_bits - |
1065 | SECTOR_SHIFT), |
1066 | nr_sects: nr_blocks << (sb->s_blocksize_bits - |
1067 | SECTOR_SHIFT), |
1068 | gfp_mask, flags: 0); |
1069 | } |
1070 | |
1071 | static inline bool bdev_is_partition(struct block_device *bdev) |
1072 | { |
1073 | return bdev->bd_partno; |
1074 | } |
1075 | |
1076 | enum blk_default_limits { |
1077 | BLK_MAX_SEGMENTS = 128, |
1078 | BLK_SAFE_MAX_SECTORS = 255, |
1079 | BLK_MAX_SEGMENT_SIZE = 65536, |
1080 | BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL, |
1081 | }; |
1082 | |
1083 | #define BLK_DEF_MAX_SECTORS 2560u |
1084 | |
1085 | static inline unsigned long queue_segment_boundary(const struct request_queue *q) |
1086 | { |
1087 | return q->limits.seg_boundary_mask; |
1088 | } |
1089 | |
1090 | static inline unsigned long queue_virt_boundary(const struct request_queue *q) |
1091 | { |
1092 | return q->limits.virt_boundary_mask; |
1093 | } |
1094 | |
1095 | static inline unsigned int queue_max_sectors(const struct request_queue *q) |
1096 | { |
1097 | return q->limits.max_sectors; |
1098 | } |
1099 | |
1100 | static inline unsigned int queue_max_bytes(struct request_queue *q) |
1101 | { |
1102 | return min_t(unsigned int, queue_max_sectors(q), INT_MAX >> 9) << 9; |
1103 | } |
1104 | |
1105 | static inline unsigned int queue_max_hw_sectors(const struct request_queue *q) |
1106 | { |
1107 | return q->limits.max_hw_sectors; |
1108 | } |
1109 | |
1110 | static inline unsigned short queue_max_segments(const struct request_queue *q) |
1111 | { |
1112 | return q->limits.max_segments; |
1113 | } |
1114 | |
1115 | static inline unsigned short queue_max_discard_segments(const struct request_queue *q) |
1116 | { |
1117 | return q->limits.max_discard_segments; |
1118 | } |
1119 | |
1120 | static inline unsigned int queue_max_segment_size(const struct request_queue *q) |
1121 | { |
1122 | return q->limits.max_segment_size; |
1123 | } |
1124 | |
1125 | static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q) |
1126 | { |
1127 | |
1128 | const struct queue_limits *l = &q->limits; |
1129 | |
1130 | return min(l->max_zone_append_sectors, l->max_sectors); |
1131 | } |
1132 | |
1133 | static inline unsigned int |
1134 | bdev_max_zone_append_sectors(struct block_device *bdev) |
1135 | { |
1136 | return queue_max_zone_append_sectors(q: bdev_get_queue(bdev)); |
1137 | } |
1138 | |
1139 | static inline unsigned int bdev_max_segments(struct block_device *bdev) |
1140 | { |
1141 | return queue_max_segments(q: bdev_get_queue(bdev)); |
1142 | } |
1143 | |
1144 | static inline unsigned queue_logical_block_size(const struct request_queue *q) |
1145 | { |
1146 | int retval = 512; |
1147 | |
1148 | if (q && q->limits.logical_block_size) |
1149 | retval = q->limits.logical_block_size; |
1150 | |
1151 | return retval; |
1152 | } |
1153 | |
1154 | static inline unsigned int bdev_logical_block_size(struct block_device *bdev) |
1155 | { |
1156 | return queue_logical_block_size(q: bdev_get_queue(bdev)); |
1157 | } |
1158 | |
1159 | static inline unsigned int queue_physical_block_size(const struct request_queue *q) |
1160 | { |
1161 | return q->limits.physical_block_size; |
1162 | } |
1163 | |
1164 | static inline unsigned int bdev_physical_block_size(struct block_device *bdev) |
1165 | { |
1166 | return queue_physical_block_size(q: bdev_get_queue(bdev)); |
1167 | } |
1168 | |
1169 | static inline unsigned int queue_io_min(const struct request_queue *q) |
1170 | { |
1171 | return q->limits.io_min; |
1172 | } |
1173 | |
1174 | static inline int bdev_io_min(struct block_device *bdev) |
1175 | { |
1176 | return queue_io_min(q: bdev_get_queue(bdev)); |
1177 | } |
1178 | |
1179 | static inline unsigned int queue_io_opt(const struct request_queue *q) |
1180 | { |
1181 | return q->limits.io_opt; |
1182 | } |
1183 | |
1184 | static inline int bdev_io_opt(struct block_device *bdev) |
1185 | { |
1186 | return queue_io_opt(q: bdev_get_queue(bdev)); |
1187 | } |
1188 | |
1189 | static inline unsigned int |
1190 | queue_zone_write_granularity(const struct request_queue *q) |
1191 | { |
1192 | return q->limits.zone_write_granularity; |
1193 | } |
1194 | |
1195 | static inline unsigned int |
1196 | bdev_zone_write_granularity(struct block_device *bdev) |
1197 | { |
1198 | return queue_zone_write_granularity(q: bdev_get_queue(bdev)); |
1199 | } |
1200 | |
1201 | int bdev_alignment_offset(struct block_device *bdev); |
1202 | unsigned int bdev_discard_alignment(struct block_device *bdev); |
1203 | |
1204 | static inline unsigned int bdev_max_discard_sectors(struct block_device *bdev) |
1205 | { |
1206 | return bdev_get_queue(bdev)->limits.max_discard_sectors; |
1207 | } |
1208 | |
1209 | static inline unsigned int bdev_discard_granularity(struct block_device *bdev) |
1210 | { |
1211 | return bdev_get_queue(bdev)->limits.discard_granularity; |
1212 | } |
1213 | |
1214 | static inline unsigned int |
1215 | bdev_max_secure_erase_sectors(struct block_device *bdev) |
1216 | { |
1217 | return bdev_get_queue(bdev)->limits.max_secure_erase_sectors; |
1218 | } |
1219 | |
1220 | static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev) |
1221 | { |
1222 | struct request_queue *q = bdev_get_queue(bdev); |
1223 | |
1224 | if (q) |
1225 | return q->limits.max_write_zeroes_sectors; |
1226 | |
1227 | return 0; |
1228 | } |
1229 | |
1230 | static inline bool bdev_nonrot(struct block_device *bdev) |
1231 | { |
1232 | return blk_queue_nonrot(bdev_get_queue(bdev)); |
1233 | } |
1234 | |
1235 | static inline bool bdev_synchronous(struct block_device *bdev) |
1236 | { |
1237 | return test_bit(QUEUE_FLAG_SYNCHRONOUS, |
1238 | &bdev_get_queue(bdev)->queue_flags); |
1239 | } |
1240 | |
1241 | static inline bool bdev_stable_writes(struct block_device *bdev) |
1242 | { |
1243 | return test_bit(QUEUE_FLAG_STABLE_WRITES, |
1244 | &bdev_get_queue(bdev)->queue_flags); |
1245 | } |
1246 | |
1247 | static inline bool bdev_write_cache(struct block_device *bdev) |
1248 | { |
1249 | return test_bit(QUEUE_FLAG_WC, &bdev_get_queue(bdev)->queue_flags); |
1250 | } |
1251 | |
1252 | static inline bool bdev_fua(struct block_device *bdev) |
1253 | { |
1254 | return test_bit(QUEUE_FLAG_FUA, &bdev_get_queue(bdev)->queue_flags); |
1255 | } |
1256 | |
1257 | static inline bool bdev_nowait(struct block_device *bdev) |
1258 | { |
1259 | return test_bit(QUEUE_FLAG_NOWAIT, &bdev_get_queue(bdev)->queue_flags); |
1260 | } |
1261 | |
1262 | static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev) |
1263 | { |
1264 | return blk_queue_zoned_model(q: bdev_get_queue(bdev)); |
1265 | } |
1266 | |
1267 | static inline bool bdev_is_zoned(struct block_device *bdev) |
1268 | { |
1269 | return blk_queue_is_zoned(q: bdev_get_queue(bdev)); |
1270 | } |
1271 | |
1272 | static inline unsigned int bdev_zone_no(struct block_device *bdev, sector_t sec) |
1273 | { |
1274 | return disk_zone_no(disk: bdev->bd_disk, sector: sec); |
1275 | } |
1276 | |
1277 | /* Whether write serialization is required for @op on zoned devices. */ |
1278 | static inline bool op_needs_zoned_write_locking(enum req_op op) |
1279 | { |
1280 | return op == REQ_OP_WRITE || op == REQ_OP_WRITE_ZEROES; |
1281 | } |
1282 | |
1283 | static inline bool bdev_op_is_zoned_write(struct block_device *bdev, |
1284 | enum req_op op) |
1285 | { |
1286 | return bdev_is_zoned(bdev) && op_needs_zoned_write_locking(op); |
1287 | } |
1288 | |
1289 | static inline sector_t bdev_zone_sectors(struct block_device *bdev) |
1290 | { |
1291 | struct request_queue *q = bdev_get_queue(bdev); |
1292 | |
1293 | if (!blk_queue_is_zoned(q)) |
1294 | return 0; |
1295 | return q->limits.chunk_sectors; |
1296 | } |
1297 | |
1298 | static inline sector_t bdev_offset_from_zone_start(struct block_device *bdev, |
1299 | sector_t sector) |
1300 | { |
1301 | return sector & (bdev_zone_sectors(bdev) - 1); |
1302 | } |
1303 | |
1304 | static inline bool bdev_is_zone_start(struct block_device *bdev, |
1305 | sector_t sector) |
1306 | { |
1307 | return bdev_offset_from_zone_start(bdev, sector) == 0; |
1308 | } |
1309 | |
1310 | static inline int queue_dma_alignment(const struct request_queue *q) |
1311 | { |
1312 | return q ? q->limits.dma_alignment : 511; |
1313 | } |
1314 | |
1315 | static inline unsigned int bdev_dma_alignment(struct block_device *bdev) |
1316 | { |
1317 | return queue_dma_alignment(q: bdev_get_queue(bdev)); |
1318 | } |
1319 | |
1320 | static inline bool bdev_iter_is_aligned(struct block_device *bdev, |
1321 | struct iov_iter *iter) |
1322 | { |
1323 | return iov_iter_is_aligned(i: iter, addr_mask: bdev_dma_alignment(bdev), |
1324 | len_mask: bdev_logical_block_size(bdev) - 1); |
1325 | } |
1326 | |
1327 | static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr, |
1328 | unsigned int len) |
1329 | { |
1330 | unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask; |
1331 | return !(addr & alignment) && !(len & alignment); |
1332 | } |
1333 | |
1334 | /* assumes size > 256 */ |
1335 | static inline unsigned int blksize_bits(unsigned int size) |
1336 | { |
1337 | return order_base_2(size >> SECTOR_SHIFT) + SECTOR_SHIFT; |
1338 | } |
1339 | |
1340 | static inline unsigned int block_size(struct block_device *bdev) |
1341 | { |
1342 | return 1 << bdev->bd_inode->i_blkbits; |
1343 | } |
1344 | |
1345 | int kblockd_schedule_work(struct work_struct *work); |
1346 | int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay); |
1347 | |
1348 | #define MODULE_ALIAS_BLOCKDEV(major,minor) \ |
1349 | MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor)) |
1350 | #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \ |
1351 | MODULE_ALIAS("block-major-" __stringify(major) "-*") |
1352 | |
1353 | #ifdef CONFIG_BLK_INLINE_ENCRYPTION |
1354 | |
1355 | bool blk_crypto_register(struct blk_crypto_profile *profile, |
1356 | struct request_queue *q); |
1357 | |
1358 | #else /* CONFIG_BLK_INLINE_ENCRYPTION */ |
1359 | |
1360 | static inline bool blk_crypto_register(struct blk_crypto_profile *profile, |
1361 | struct request_queue *q) |
1362 | { |
1363 | return true; |
1364 | } |
1365 | |
1366 | #endif /* CONFIG_BLK_INLINE_ENCRYPTION */ |
1367 | |
1368 | enum blk_unique_id { |
1369 | /* these match the Designator Types specified in SPC */ |
1370 | BLK_UID_T10 = 1, |
1371 | BLK_UID_EUI64 = 2, |
1372 | BLK_UID_NAA = 3, |
1373 | }; |
1374 | |
1375 | struct block_device_operations { |
1376 | void (*submit_bio)(struct bio *bio); |
1377 | int (*poll_bio)(struct bio *bio, struct io_comp_batch *iob, |
1378 | unsigned int flags); |
1379 | int (*open)(struct gendisk *disk, blk_mode_t mode); |
1380 | void (*release)(struct gendisk *disk); |
1381 | int (*ioctl)(struct block_device *bdev, blk_mode_t mode, |
1382 | unsigned cmd, unsigned long arg); |
1383 | int (*compat_ioctl)(struct block_device *bdev, blk_mode_t mode, |
1384 | unsigned cmd, unsigned long arg); |
1385 | unsigned int (*check_events) (struct gendisk *disk, |
1386 | unsigned int clearing); |
1387 | void (*unlock_native_capacity) (struct gendisk *); |
1388 | int (*getgeo)(struct block_device *, struct hd_geometry *); |
1389 | int (*set_read_only)(struct block_device *bdev, bool ro); |
1390 | void (*free_disk)(struct gendisk *disk); |
1391 | /* this callback is with swap_lock and sometimes page table lock held */ |
1392 | void (*swap_slot_free_notify) (struct block_device *, unsigned long); |
1393 | int (*report_zones)(struct gendisk *, sector_t sector, |
1394 | unsigned int nr_zones, report_zones_cb cb, void *data); |
1395 | char *(*devnode)(struct gendisk *disk, umode_t *mode); |
1396 | /* returns the length of the identifier or a negative errno: */ |
1397 | int (*get_unique_id)(struct gendisk *disk, u8 id[16], |
1398 | enum blk_unique_id id_type); |
1399 | struct module *owner; |
1400 | const struct pr_ops *pr_ops; |
1401 | |
1402 | /* |
1403 | * Special callback for probing GPT entry at a given sector. |
1404 | * Needed by Android devices, used by GPT scanner and MMC blk |
1405 | * driver. |
1406 | */ |
1407 | int (*alternative_gpt_sector)(struct gendisk *disk, sector_t *sector); |
1408 | }; |
1409 | |
1410 | #ifdef CONFIG_COMPAT |
1411 | extern int blkdev_compat_ptr_ioctl(struct block_device *, blk_mode_t, |
1412 | unsigned int, unsigned long); |
1413 | #else |
1414 | #define blkdev_compat_ptr_ioctl NULL |
1415 | #endif |
1416 | |
1417 | static inline void blk_wake_io_task(struct task_struct *waiter) |
1418 | { |
1419 | /* |
1420 | * If we're polling, the task itself is doing the completions. For |
1421 | * that case, we don't need to signal a wakeup, it's enough to just |
1422 | * mark us as RUNNING. |
1423 | */ |
1424 | if (waiter == current) |
1425 | __set_current_state(TASK_RUNNING); |
1426 | else |
1427 | wake_up_process(tsk: waiter); |
1428 | } |
1429 | |
1430 | unsigned long bdev_start_io_acct(struct block_device *bdev, enum req_op op, |
1431 | unsigned long start_time); |
1432 | void bdev_end_io_acct(struct block_device *bdev, enum req_op op, |
1433 | unsigned int sectors, unsigned long start_time); |
1434 | |
1435 | unsigned long bio_start_io_acct(struct bio *bio); |
1436 | void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time, |
1437 | struct block_device *orig_bdev); |
1438 | |
1439 | /** |
1440 | * bio_end_io_acct - end I/O accounting for bio based drivers |
1441 | * @bio: bio to end account for |
1442 | * @start_time: start time returned by bio_start_io_acct() |
1443 | */ |
1444 | static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time) |
1445 | { |
1446 | return bio_end_io_acct_remapped(bio, start_time, orig_bdev: bio->bi_bdev); |
1447 | } |
1448 | |
1449 | int bdev_read_only(struct block_device *bdev); |
1450 | int set_blocksize(struct block_device *bdev, int size); |
1451 | |
1452 | int lookup_bdev(const char *pathname, dev_t *dev); |
1453 | |
1454 | void blkdev_show(struct seq_file *seqf, off_t offset); |
1455 | |
1456 | #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */ |
1457 | #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */ |
1458 | #ifdef CONFIG_BLOCK |
1459 | #define BLKDEV_MAJOR_MAX 512 |
1460 | #else |
1461 | #define BLKDEV_MAJOR_MAX 0 |
1462 | #endif |
1463 | |
1464 | struct blk_holder_ops { |
1465 | void (*mark_dead)(struct block_device *bdev, bool surprise); |
1466 | |
1467 | /* |
1468 | * Sync the file system mounted on the block device. |
1469 | */ |
1470 | void (*sync)(struct block_device *bdev); |
1471 | }; |
1472 | |
1473 | extern const struct blk_holder_ops fs_holder_ops; |
1474 | |
1475 | /* |
1476 | * Return the correct open flags for blkdev_get_by_* for super block flags |
1477 | * as stored in sb->s_flags. |
1478 | */ |
1479 | #define sb_open_mode(flags) \ |
1480 | (BLK_OPEN_READ | (((flags) & SB_RDONLY) ? 0 : BLK_OPEN_WRITE)) |
1481 | |
1482 | struct bdev_handle { |
1483 | struct block_device *bdev; |
1484 | void *holder; |
1485 | blk_mode_t mode; |
1486 | }; |
1487 | |
1488 | struct block_device *blkdev_get_by_dev(dev_t dev, blk_mode_t mode, void *holder, |
1489 | const struct blk_holder_ops *hops); |
1490 | struct block_device *blkdev_get_by_path(const char *path, blk_mode_t mode, |
1491 | void *holder, const struct blk_holder_ops *hops); |
1492 | struct bdev_handle *bdev_open_by_dev(dev_t dev, blk_mode_t mode, void *holder, |
1493 | const struct blk_holder_ops *hops); |
1494 | struct bdev_handle *bdev_open_by_path(const char *path, blk_mode_t mode, |
1495 | void *holder, const struct blk_holder_ops *hops); |
1496 | int bd_prepare_to_claim(struct block_device *bdev, void *holder, |
1497 | const struct blk_holder_ops *hops); |
1498 | void bd_abort_claiming(struct block_device *bdev, void *holder); |
1499 | void blkdev_put(struct block_device *bdev, void *holder); |
1500 | void bdev_release(struct bdev_handle *handle); |
1501 | |
1502 | /* just for blk-cgroup, don't use elsewhere */ |
1503 | struct block_device *blkdev_get_no_open(dev_t dev); |
1504 | void blkdev_put_no_open(struct block_device *bdev); |
1505 | |
1506 | struct block_device *I_BDEV(struct inode *inode); |
1507 | |
1508 | #ifdef CONFIG_BLOCK |
1509 | void invalidate_bdev(struct block_device *bdev); |
1510 | int sync_blockdev(struct block_device *bdev); |
1511 | int sync_blockdev_range(struct block_device *bdev, loff_t lstart, loff_t lend); |
1512 | int sync_blockdev_nowait(struct block_device *bdev); |
1513 | void sync_bdevs(bool wait); |
1514 | void bdev_statx_dioalign(struct inode *inode, struct kstat *stat); |
1515 | void printk_all_partitions(void); |
1516 | int __init early_lookup_bdev(const char *pathname, dev_t *dev); |
1517 | #else |
1518 | static inline void invalidate_bdev(struct block_device *bdev) |
1519 | { |
1520 | } |
1521 | static inline int sync_blockdev(struct block_device *bdev) |
1522 | { |
1523 | return 0; |
1524 | } |
1525 | static inline int sync_blockdev_nowait(struct block_device *bdev) |
1526 | { |
1527 | return 0; |
1528 | } |
1529 | static inline void sync_bdevs(bool wait) |
1530 | { |
1531 | } |
1532 | static inline void bdev_statx_dioalign(struct inode *inode, struct kstat *stat) |
1533 | { |
1534 | } |
1535 | static inline void printk_all_partitions(void) |
1536 | { |
1537 | } |
1538 | static inline int early_lookup_bdev(const char *pathname, dev_t *dev) |
1539 | { |
1540 | return -EINVAL; |
1541 | } |
1542 | #endif /* CONFIG_BLOCK */ |
1543 | |
1544 | int freeze_bdev(struct block_device *bdev); |
1545 | int thaw_bdev(struct block_device *bdev); |
1546 | |
1547 | struct io_comp_batch { |
1548 | struct request *req_list; |
1549 | bool need_ts; |
1550 | void (*complete)(struct io_comp_batch *); |
1551 | }; |
1552 | |
1553 | #define DEFINE_IO_COMP_BATCH(name) struct io_comp_batch name = { } |
1554 | |
1555 | #endif /* _LINUX_BLKDEV_H */ |
1556 | |