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