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 /* open is exclusive wrt all other BLK_OPEN_WRITE opens to the device */ 128 #define BLK_OPEN_RESTRICT_WRITES ((__force blk_mode_t)(1 << 5)) 129 130 struct gendisk { 131 /* 132 * major/first_minor/minors should not be set by any new driver, the 133 * block core will take care of allocating them automatically. 134 */ 135 int major; 136 int first_minor; 137 int minors; 138 139 char disk_name[DISK_NAME_LEN]; /* name of major driver */ 140 141 unsigned short events; /* supported events */ 142 unsigned short event_flags; /* flags related to event processing */ 143 144 struct xarray part_tbl; 145 struct block_device *part0; 146 147 const struct block_device_operations *fops; 148 struct request_queue *queue; 149 void *private_data; 150 151 struct bio_set bio_split; 152 153 int flags; 154 unsigned long state; 155 #define GD_NEED_PART_SCAN 0 156 #define GD_READ_ONLY 1 157 #define GD_DEAD 2 158 #define GD_NATIVE_CAPACITY 3 159 #define GD_ADDED 4 160 #define GD_SUPPRESS_PART_SCAN 5 161 #define GD_OWNS_QUEUE 6 162 163 struct mutex open_mutex; /* open/close mutex */ 164 unsigned open_partitions; /* number of open partitions */ 165 166 struct backing_dev_info *bdi; 167 struct kobject queue_kobj; /* the queue/ directory */ 168 struct kobject *slave_dir; 169 #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED 170 struct list_head slave_bdevs; 171 #endif 172 struct timer_rand_state *random; 173 atomic_t sync_io; /* RAID */ 174 struct disk_events *ev; 175 176 #ifdef CONFIG_BLK_DEV_ZONED 177 /* 178 * Zoned block device information for request dispatch control. 179 * nr_zones is the total number of zones of the device. This is always 180 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones 181 * bits which indicates if a zone is conventional (bit set) or 182 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones 183 * bits which indicates if a zone is write locked, that is, if a write 184 * request targeting the zone was dispatched. 185 * 186 * Reads of this information must be protected with blk_queue_enter() / 187 * blk_queue_exit(). Modifying this information is only allowed while 188 * no requests are being processed. See also blk_mq_freeze_queue() and 189 * blk_mq_unfreeze_queue(). 190 */ 191 unsigned int nr_zones; 192 unsigned int max_open_zones; 193 unsigned int max_active_zones; 194 unsigned long *conv_zones_bitmap; 195 unsigned long *seq_zones_wlock; 196 #endif /* CONFIG_BLK_DEV_ZONED */ 197 198 #if IS_ENABLED(CONFIG_CDROM) 199 struct cdrom_device_info *cdi; 200 #endif 201 int node_id; 202 struct badblocks *bb; 203 struct lockdep_map lockdep_map; 204 u64 diskseq; 205 blk_mode_t open_mode; 206 207 /* 208 * Independent sector access ranges. This is always NULL for 209 * devices that do not have multiple independent access ranges. 210 */ 211 struct blk_independent_access_ranges *ia_ranges; 212 }; 213 214 static inline bool disk_live(struct gendisk *disk) 215 { 216 return !inode_unhashed(disk->part0->bd_inode); 217 } 218 219 /** 220 * disk_openers - returns how many openers are there for a disk 221 * @disk: disk to check 222 * 223 * This returns the number of openers for a disk. Note that this value is only 224 * stable if disk->open_mutex is held. 225 * 226 * Note: Due to a quirk in the block layer open code, each open partition is 227 * only counted once even if there are multiple openers. 228 */ 229 static inline unsigned int disk_openers(struct gendisk *disk) 230 { 231 return atomic_read(&disk->part0->bd_openers); 232 } 233 234 /* 235 * The gendisk is refcounted by the part0 block_device, and the bd_device 236 * therein is also used for device model presentation in sysfs. 237 */ 238 #define dev_to_disk(device) \ 239 (dev_to_bdev(device)->bd_disk) 240 #define disk_to_dev(disk) \ 241 (&((disk)->part0->bd_device)) 242 243 #if IS_REACHABLE(CONFIG_CDROM) 244 #define disk_to_cdi(disk) ((disk)->cdi) 245 #else 246 #define disk_to_cdi(disk) NULL 247 #endif 248 249 static inline dev_t disk_devt(struct gendisk *disk) 250 { 251 return MKDEV(disk->major, disk->first_minor); 252 } 253 254 static inline int blk_validate_block_size(unsigned long bsize) 255 { 256 if (bsize < 512 || bsize > PAGE_SIZE || !is_power_of_2(bsize)) 257 return -EINVAL; 258 259 return 0; 260 } 261 262 static inline bool blk_op_is_passthrough(blk_opf_t op) 263 { 264 op &= REQ_OP_MASK; 265 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT; 266 } 267 268 /* 269 * Zoned block device models (zoned limit). 270 * 271 * Note: This needs to be ordered from the least to the most severe 272 * restrictions for the inheritance in blk_stack_limits() to work. 273 */ 274 enum blk_zoned_model { 275 BLK_ZONED_NONE = 0, /* Regular block device */ 276 BLK_ZONED_HA, /* Host-aware zoned block device */ 277 BLK_ZONED_HM, /* Host-managed zoned block device */ 278 }; 279 280 /* 281 * BLK_BOUNCE_NONE: never bounce (default) 282 * BLK_BOUNCE_HIGH: bounce all highmem pages 283 */ 284 enum blk_bounce { 285 BLK_BOUNCE_NONE, 286 BLK_BOUNCE_HIGH, 287 }; 288 289 struct queue_limits { 290 enum blk_bounce bounce; 291 unsigned long seg_boundary_mask; 292 unsigned long virt_boundary_mask; 293 294 unsigned int max_hw_sectors; 295 unsigned int max_dev_sectors; 296 unsigned int chunk_sectors; 297 unsigned int max_sectors; 298 unsigned int max_user_sectors; 299 unsigned int max_segment_size; 300 unsigned int physical_block_size; 301 unsigned int logical_block_size; 302 unsigned int alignment_offset; 303 unsigned int io_min; 304 unsigned int io_opt; 305 unsigned int max_discard_sectors; 306 unsigned int max_hw_discard_sectors; 307 unsigned int max_secure_erase_sectors; 308 unsigned int max_write_zeroes_sectors; 309 unsigned int max_zone_append_sectors; 310 unsigned int discard_granularity; 311 unsigned int discard_alignment; 312 unsigned int zone_write_granularity; 313 314 unsigned short max_segments; 315 unsigned short max_integrity_segments; 316 unsigned short max_discard_segments; 317 318 unsigned char misaligned; 319 unsigned char discard_misaligned; 320 unsigned char raid_partial_stripes_expensive; 321 enum blk_zoned_model zoned; 322 323 /* 324 * Drivers that set dma_alignment to less than 511 must be prepared to 325 * handle individual bvec's that are not a multiple of a SECTOR_SIZE 326 * due to possible offsets. 327 */ 328 unsigned int dma_alignment; 329 }; 330 331 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx, 332 void *data); 333 334 void disk_set_zoned(struct gendisk *disk, enum blk_zoned_model model); 335 336 #ifdef CONFIG_BLK_DEV_ZONED 337 #define BLK_ALL_ZONES ((unsigned int)-1) 338 int blkdev_report_zones(struct block_device *bdev, sector_t sector, 339 unsigned int nr_zones, report_zones_cb cb, void *data); 340 unsigned int bdev_nr_zones(struct block_device *bdev); 341 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_op op, 342 sector_t sectors, sector_t nr_sectors, 343 gfp_t gfp_mask); 344 int blk_revalidate_disk_zones(struct gendisk *disk, 345 void (*update_driver_data)(struct gendisk *disk)); 346 #else /* CONFIG_BLK_DEV_ZONED */ 347 static inline unsigned int bdev_nr_zones(struct block_device *bdev) 348 { 349 return 0; 350 } 351 #endif /* CONFIG_BLK_DEV_ZONED */ 352 353 /* 354 * Independent access ranges: struct blk_independent_access_range describes 355 * a range of contiguous sectors that can be accessed using device command 356 * execution resources that are independent from the resources used for 357 * other access ranges. This is typically found with single-LUN multi-actuator 358 * HDDs where each access range is served by a different set of heads. 359 * The set of independent ranges supported by the device is defined using 360 * struct blk_independent_access_ranges. The independent ranges must not overlap 361 * and must include all sectors within the disk capacity (no sector holes 362 * allowed). 363 * For a device with multiple ranges, requests targeting sectors in different 364 * ranges can be executed in parallel. A request can straddle an access range 365 * boundary. 366 */ 367 struct blk_independent_access_range { 368 struct kobject kobj; 369 sector_t sector; 370 sector_t nr_sectors; 371 }; 372 373 struct blk_independent_access_ranges { 374 struct kobject kobj; 375 bool sysfs_registered; 376 unsigned int nr_ia_ranges; 377 struct blk_independent_access_range ia_range[]; 378 }; 379 380 struct request_queue { 381 struct request *last_merge; 382 struct elevator_queue *elevator; 383 384 struct percpu_ref q_usage_counter; 385 386 struct blk_queue_stats *stats; 387 struct rq_qos *rq_qos; 388 struct mutex rq_qos_mutex; 389 390 const struct blk_mq_ops *mq_ops; 391 392 /* sw queues */ 393 struct blk_mq_ctx __percpu *queue_ctx; 394 395 unsigned int queue_depth; 396 397 /* hw dispatch queues */ 398 struct xarray hctx_table; 399 unsigned int nr_hw_queues; 400 401 /* 402 * The queue owner gets to use this for whatever they like. 403 * ll_rw_blk doesn't touch it. 404 */ 405 void *queuedata; 406 407 /* 408 * various queue flags, see QUEUE_* below 409 */ 410 unsigned long queue_flags; 411 /* 412 * Number of contexts that have called blk_set_pm_only(). If this 413 * counter is above zero then only RQF_PM requests are processed. 414 */ 415 atomic_t pm_only; 416 417 /* 418 * ida allocated id for this queue. Used to index queues from 419 * ioctx. 420 */ 421 int id; 422 423 spinlock_t queue_lock; 424 425 struct gendisk *disk; 426 427 refcount_t refs; 428 429 /* 430 * mq queue kobject 431 */ 432 struct kobject *mq_kobj; 433 434 #ifdef CONFIG_BLK_DEV_INTEGRITY 435 struct blk_integrity integrity; 436 #endif /* CONFIG_BLK_DEV_INTEGRITY */ 437 438 #ifdef CONFIG_PM 439 struct device *dev; 440 enum rpm_status rpm_status; 441 #endif 442 443 /* 444 * queue settings 445 */ 446 unsigned long nr_requests; /* Max # of requests */ 447 448 unsigned int dma_pad_mask; 449 450 #ifdef CONFIG_BLK_INLINE_ENCRYPTION 451 struct blk_crypto_profile *crypto_profile; 452 struct kobject *crypto_kobject; 453 #endif 454 455 unsigned int rq_timeout; 456 457 struct timer_list timeout; 458 struct work_struct timeout_work; 459 460 atomic_t nr_active_requests_shared_tags; 461 462 struct blk_mq_tags *sched_shared_tags; 463 464 struct list_head icq_list; 465 #ifdef CONFIG_BLK_CGROUP 466 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS); 467 struct blkcg_gq *root_blkg; 468 struct list_head blkg_list; 469 struct mutex blkcg_mutex; 470 #endif 471 472 struct queue_limits limits; 473 474 unsigned int required_elevator_features; 475 476 int node; 477 #ifdef CONFIG_BLK_DEV_IO_TRACE 478 struct blk_trace __rcu *blk_trace; 479 #endif 480 /* 481 * for flush operations 482 */ 483 struct blk_flush_queue *fq; 484 struct list_head flush_list; 485 486 struct list_head requeue_list; 487 spinlock_t requeue_lock; 488 struct delayed_work requeue_work; 489 490 struct mutex sysfs_lock; 491 struct mutex sysfs_dir_lock; 492 493 /* 494 * for reusing dead hctx instance in case of updating 495 * nr_hw_queues 496 */ 497 struct list_head unused_hctx_list; 498 spinlock_t unused_hctx_lock; 499 500 int mq_freeze_depth; 501 502 #ifdef CONFIG_BLK_DEV_THROTTLING 503 /* Throttle data */ 504 struct throtl_data *td; 505 #endif 506 struct rcu_head rcu_head; 507 wait_queue_head_t mq_freeze_wq; 508 /* 509 * Protect concurrent access to q_usage_counter by 510 * percpu_ref_kill() and percpu_ref_reinit(). 511 */ 512 struct mutex mq_freeze_lock; 513 514 int quiesce_depth; 515 516 struct blk_mq_tag_set *tag_set; 517 struct list_head tag_set_list; 518 519 struct dentry *debugfs_dir; 520 struct dentry *sched_debugfs_dir; 521 struct dentry *rqos_debugfs_dir; 522 /* 523 * Serializes all debugfs metadata operations using the above dentries. 524 */ 525 struct mutex debugfs_mutex; 526 527 bool mq_sysfs_init_done; 528 }; 529 530 /* Keep blk_queue_flag_name[] in sync with the definitions below */ 531 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */ 532 #define QUEUE_FLAG_DYING 1 /* queue being torn down */ 533 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */ 534 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */ 535 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */ 536 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */ 537 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */ 538 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */ 539 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */ 540 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */ 541 #define QUEUE_FLAG_SYNCHRONOUS 11 /* always completes in submit context */ 542 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */ 543 #define QUEUE_FLAG_HW_WC 13 /* Write back caching supported */ 544 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */ 545 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */ 546 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */ 547 #define QUEUE_FLAG_WC 17 /* Write back caching */ 548 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */ 549 #define QUEUE_FLAG_DAX 19 /* device supports DAX */ 550 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */ 551 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */ 552 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */ 553 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */ 554 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */ 555 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */ 556 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */ 557 #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */ 558 #define QUEUE_FLAG_SQ_SCHED 30 /* single queue style io dispatch */ 559 #define QUEUE_FLAG_SKIP_TAGSET_QUIESCE 31 /* quiesce_tagset skip the queue*/ 560 561 #define QUEUE_FLAG_MQ_DEFAULT ((1UL << QUEUE_FLAG_IO_STAT) | \ 562 (1UL << QUEUE_FLAG_SAME_COMP) | \ 563 (1UL << QUEUE_FLAG_NOWAIT)) 564 565 void blk_queue_flag_set(unsigned int flag, struct request_queue *q); 566 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q); 567 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q); 568 569 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags) 570 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags) 571 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags) 572 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags) 573 #define blk_queue_noxmerges(q) \ 574 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags) 575 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags) 576 #define blk_queue_stable_writes(q) \ 577 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags) 578 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags) 579 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags) 580 #define blk_queue_zone_resetall(q) \ 581 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags) 582 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags) 583 #define blk_queue_pci_p2pdma(q) \ 584 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags) 585 #ifdef CONFIG_BLK_RQ_ALLOC_TIME 586 #define blk_queue_rq_alloc_time(q) \ 587 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags) 588 #else 589 #define blk_queue_rq_alloc_time(q) false 590 #endif 591 592 #define blk_noretry_request(rq) \ 593 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \ 594 REQ_FAILFAST_DRIVER)) 595 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags) 596 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only) 597 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags) 598 #define blk_queue_sq_sched(q) test_bit(QUEUE_FLAG_SQ_SCHED, &(q)->queue_flags) 599 #define blk_queue_skip_tagset_quiesce(q) \ 600 test_bit(QUEUE_FLAG_SKIP_TAGSET_QUIESCE, &(q)->queue_flags) 601 602 extern void blk_set_pm_only(struct request_queue *q); 603 extern void blk_clear_pm_only(struct request_queue *q); 604 605 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist) 606 607 #define dma_map_bvec(dev, bv, dir, attrs) \ 608 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \ 609 (dir), (attrs)) 610 611 static inline bool queue_is_mq(struct request_queue *q) 612 { 613 return q->mq_ops; 614 } 615 616 #ifdef CONFIG_PM 617 static inline enum rpm_status queue_rpm_status(struct request_queue *q) 618 { 619 return q->rpm_status; 620 } 621 #else 622 static inline enum rpm_status queue_rpm_status(struct request_queue *q) 623 { 624 return RPM_ACTIVE; 625 } 626 #endif 627 628 static inline enum blk_zoned_model 629 blk_queue_zoned_model(struct request_queue *q) 630 { 631 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED)) 632 return q->limits.zoned; 633 return BLK_ZONED_NONE; 634 } 635 636 static inline bool blk_queue_is_zoned(struct request_queue *q) 637 { 638 switch (blk_queue_zoned_model(q)) { 639 case BLK_ZONED_HA: 640 case BLK_ZONED_HM: 641 return true; 642 default: 643 return false; 644 } 645 } 646 647 #ifdef CONFIG_BLK_DEV_ZONED 648 static inline unsigned int disk_nr_zones(struct gendisk *disk) 649 { 650 return blk_queue_is_zoned(disk->queue) ? disk->nr_zones : 0; 651 } 652 653 static inline unsigned int disk_zone_no(struct gendisk *disk, sector_t sector) 654 { 655 if (!blk_queue_is_zoned(disk->queue)) 656 return 0; 657 return sector >> ilog2(disk->queue->limits.chunk_sectors); 658 } 659 660 static inline bool disk_zone_is_seq(struct gendisk *disk, sector_t sector) 661 { 662 if (!blk_queue_is_zoned(disk->queue)) 663 return false; 664 if (!disk->conv_zones_bitmap) 665 return true; 666 return !test_bit(disk_zone_no(disk, sector), disk->conv_zones_bitmap); 667 } 668 669 static inline void disk_set_max_open_zones(struct gendisk *disk, 670 unsigned int max_open_zones) 671 { 672 disk->max_open_zones = max_open_zones; 673 } 674 675 static inline void disk_set_max_active_zones(struct gendisk *disk, 676 unsigned int max_active_zones) 677 { 678 disk->max_active_zones = max_active_zones; 679 } 680 681 static inline unsigned int bdev_max_open_zones(struct block_device *bdev) 682 { 683 return bdev->bd_disk->max_open_zones; 684 } 685 686 static inline unsigned int bdev_max_active_zones(struct block_device *bdev) 687 { 688 return bdev->bd_disk->max_active_zones; 689 } 690 691 #else /* CONFIG_BLK_DEV_ZONED */ 692 static inline unsigned int disk_nr_zones(struct gendisk *disk) 693 { 694 return 0; 695 } 696 static inline bool disk_zone_is_seq(struct gendisk *disk, sector_t sector) 697 { 698 return false; 699 } 700 static inline unsigned int disk_zone_no(struct gendisk *disk, sector_t sector) 701 { 702 return 0; 703 } 704 static inline unsigned int bdev_max_open_zones(struct block_device *bdev) 705 { 706 return 0; 707 } 708 709 static inline unsigned int bdev_max_active_zones(struct block_device *bdev) 710 { 711 return 0; 712 } 713 #endif /* CONFIG_BLK_DEV_ZONED */ 714 715 static inline unsigned int blk_queue_depth(struct request_queue *q) 716 { 717 if (q->queue_depth) 718 return q->queue_depth; 719 720 return q->nr_requests; 721 } 722 723 /* 724 * default timeout for SG_IO if none specified 725 */ 726 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ) 727 #define BLK_MIN_SG_TIMEOUT (7 * HZ) 728 729 /* This should not be used directly - use rq_for_each_segment */ 730 #define for_each_bio(_bio) \ 731 for (; _bio; _bio = _bio->bi_next) 732 733 int __must_check device_add_disk(struct device *parent, struct gendisk *disk, 734 const struct attribute_group **groups); 735 static inline int __must_check add_disk(struct gendisk *disk) 736 { 737 return device_add_disk(NULL, disk, NULL); 738 } 739 void del_gendisk(struct gendisk *gp); 740 void invalidate_disk(struct gendisk *disk); 741 void set_disk_ro(struct gendisk *disk, bool read_only); 742 void disk_uevent(struct gendisk *disk, enum kobject_action action); 743 744 static inline int get_disk_ro(struct gendisk *disk) 745 { 746 return disk->part0->bd_read_only || 747 test_bit(GD_READ_ONLY, &disk->state); 748 } 749 750 static inline int bdev_read_only(struct block_device *bdev) 751 { 752 return bdev->bd_read_only || get_disk_ro(bdev->bd_disk); 753 } 754 755 bool set_capacity_and_notify(struct gendisk *disk, sector_t size); 756 void disk_force_media_change(struct gendisk *disk); 757 void bdev_mark_dead(struct block_device *bdev, bool surprise); 758 759 void add_disk_randomness(struct gendisk *disk) __latent_entropy; 760 void rand_initialize_disk(struct gendisk *disk); 761 762 static inline sector_t get_start_sect(struct block_device *bdev) 763 { 764 return bdev->bd_start_sect; 765 } 766 767 static inline sector_t bdev_nr_sectors(struct block_device *bdev) 768 { 769 return bdev->bd_nr_sectors; 770 } 771 772 static inline loff_t bdev_nr_bytes(struct block_device *bdev) 773 { 774 return (loff_t)bdev_nr_sectors(bdev) << SECTOR_SHIFT; 775 } 776 777 static inline sector_t get_capacity(struct gendisk *disk) 778 { 779 return bdev_nr_sectors(disk->part0); 780 } 781 782 static inline u64 sb_bdev_nr_blocks(struct super_block *sb) 783 { 784 return bdev_nr_sectors(sb->s_bdev) >> 785 (sb->s_blocksize_bits - SECTOR_SHIFT); 786 } 787 788 int bdev_disk_changed(struct gendisk *disk, bool invalidate); 789 790 void put_disk(struct gendisk *disk); 791 struct gendisk *__blk_alloc_disk(int node, struct lock_class_key *lkclass); 792 793 /** 794 * blk_alloc_disk - allocate a gendisk structure 795 * @node_id: numa node to allocate on 796 * 797 * Allocate and pre-initialize a gendisk structure for use with BIO based 798 * drivers. 799 * 800 * Context: can sleep 801 */ 802 #define blk_alloc_disk(node_id) \ 803 ({ \ 804 static struct lock_class_key __key; \ 805 \ 806 __blk_alloc_disk(node_id, &__key); \ 807 }) 808 809 int __register_blkdev(unsigned int major, const char *name, 810 void (*probe)(dev_t devt)); 811 #define register_blkdev(major, name) \ 812 __register_blkdev(major, name, NULL) 813 void unregister_blkdev(unsigned int major, const char *name); 814 815 bool disk_check_media_change(struct gendisk *disk); 816 void set_capacity(struct gendisk *disk, sector_t size); 817 818 #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED 819 int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk); 820 void bd_unlink_disk_holder(struct block_device *bdev, struct gendisk *disk); 821 #else 822 static inline int bd_link_disk_holder(struct block_device *bdev, 823 struct gendisk *disk) 824 { 825 return 0; 826 } 827 static inline void bd_unlink_disk_holder(struct block_device *bdev, 828 struct gendisk *disk) 829 { 830 } 831 #endif /* CONFIG_BLOCK_HOLDER_DEPRECATED */ 832 833 dev_t part_devt(struct gendisk *disk, u8 partno); 834 void inc_diskseq(struct gendisk *disk); 835 void blk_request_module(dev_t devt); 836 837 extern int blk_register_queue(struct gendisk *disk); 838 extern void blk_unregister_queue(struct gendisk *disk); 839 void submit_bio_noacct(struct bio *bio); 840 struct bio *bio_split_to_limits(struct bio *bio); 841 842 extern int blk_lld_busy(struct request_queue *q); 843 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags); 844 extern void blk_queue_exit(struct request_queue *q); 845 extern void blk_sync_queue(struct request_queue *q); 846 847 /* Helper to convert REQ_OP_XXX to its string format XXX */ 848 extern const char *blk_op_str(enum req_op op); 849 850 int blk_status_to_errno(blk_status_t status); 851 blk_status_t errno_to_blk_status(int errno); 852 const char *blk_status_to_str(blk_status_t status); 853 854 /* only poll the hardware once, don't continue until a completion was found */ 855 #define BLK_POLL_ONESHOT (1 << 0) 856 int bio_poll(struct bio *bio, struct io_comp_batch *iob, unsigned int flags); 857 int iocb_bio_iopoll(struct kiocb *kiocb, struct io_comp_batch *iob, 858 unsigned int flags); 859 860 static inline struct request_queue *bdev_get_queue(struct block_device *bdev) 861 { 862 return bdev->bd_queue; /* this is never NULL */ 863 } 864 865 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */ 866 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond); 867 868 static inline unsigned int bio_zone_no(struct bio *bio) 869 { 870 return disk_zone_no(bio->bi_bdev->bd_disk, bio->bi_iter.bi_sector); 871 } 872 873 static inline unsigned int bio_zone_is_seq(struct bio *bio) 874 { 875 return disk_zone_is_seq(bio->bi_bdev->bd_disk, bio->bi_iter.bi_sector); 876 } 877 878 /* 879 * Return how much of the chunk is left to be used for I/O at a given offset. 880 */ 881 static inline unsigned int blk_chunk_sectors_left(sector_t offset, 882 unsigned int chunk_sectors) 883 { 884 if (unlikely(!is_power_of_2(chunk_sectors))) 885 return chunk_sectors - sector_div(offset, chunk_sectors); 886 return chunk_sectors - (offset & (chunk_sectors - 1)); 887 } 888 889 /* 890 * Access functions for manipulating queue properties 891 */ 892 void blk_queue_bounce_limit(struct request_queue *q, enum blk_bounce limit); 893 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int); 894 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int); 895 extern void blk_queue_max_segments(struct request_queue *, unsigned short); 896 extern void blk_queue_max_discard_segments(struct request_queue *, 897 unsigned short); 898 void blk_queue_max_secure_erase_sectors(struct request_queue *q, 899 unsigned int max_sectors); 900 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int); 901 extern void blk_queue_max_discard_sectors(struct request_queue *q, 902 unsigned int max_discard_sectors); 903 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q, 904 unsigned int max_write_same_sectors); 905 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int); 906 extern void blk_queue_max_zone_append_sectors(struct request_queue *q, 907 unsigned int max_zone_append_sectors); 908 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int); 909 void blk_queue_zone_write_granularity(struct request_queue *q, 910 unsigned int size); 911 extern void blk_queue_alignment_offset(struct request_queue *q, 912 unsigned int alignment); 913 void disk_update_readahead(struct gendisk *disk); 914 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min); 915 extern void blk_queue_io_min(struct request_queue *q, unsigned int min); 916 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt); 917 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt); 918 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth); 919 extern void blk_set_stacking_limits(struct queue_limits *lim); 920 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b, 921 sector_t offset); 922 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev, 923 sector_t offset); 924 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int); 925 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long); 926 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long); 927 extern void blk_queue_dma_alignment(struct request_queue *, int); 928 extern void blk_queue_update_dma_alignment(struct request_queue *, int); 929 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int); 930 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua); 931 932 struct blk_independent_access_ranges * 933 disk_alloc_independent_access_ranges(struct gendisk *disk, int nr_ia_ranges); 934 void disk_set_independent_access_ranges(struct gendisk *disk, 935 struct blk_independent_access_ranges *iars); 936 937 /* 938 * Elevator features for blk_queue_required_elevator_features: 939 */ 940 /* Supports zoned block devices sequential write constraint */ 941 #define ELEVATOR_F_ZBD_SEQ_WRITE (1U << 0) 942 943 extern void blk_queue_required_elevator_features(struct request_queue *q, 944 unsigned int features); 945 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q, 946 struct device *dev); 947 948 bool __must_check blk_get_queue(struct request_queue *); 949 extern void blk_put_queue(struct request_queue *); 950 951 void blk_mark_disk_dead(struct gendisk *disk); 952 953 #ifdef CONFIG_BLOCK 954 /* 955 * blk_plug permits building a queue of related requests by holding the I/O 956 * fragments for a short period. This allows merging of sequential requests 957 * into single larger request. As the requests are moved from a per-task list to 958 * the device's request_queue in a batch, this results in improved scalability 959 * as the lock contention for request_queue lock is reduced. 960 * 961 * It is ok not to disable preemption when adding the request to the plug list 962 * or when attempting a merge. For details, please see schedule() where 963 * blk_flush_plug() is called. 964 */ 965 struct blk_plug { 966 struct request *mq_list; /* blk-mq requests */ 967 968 /* if ios_left is > 1, we can batch tag/rq allocations */ 969 struct request *cached_rq; 970 unsigned short nr_ios; 971 972 unsigned short rq_count; 973 974 bool multiple_queues; 975 bool has_elevator; 976 977 struct list_head cb_list; /* md requires an unplug callback */ 978 }; 979 980 struct blk_plug_cb; 981 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool); 982 struct blk_plug_cb { 983 struct list_head list; 984 blk_plug_cb_fn callback; 985 void *data; 986 }; 987 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug, 988 void *data, int size); 989 extern void blk_start_plug(struct blk_plug *); 990 extern void blk_start_plug_nr_ios(struct blk_plug *, unsigned short); 991 extern void blk_finish_plug(struct blk_plug *); 992 993 void __blk_flush_plug(struct blk_plug *plug, bool from_schedule); 994 static inline void blk_flush_plug(struct blk_plug *plug, bool async) 995 { 996 if (plug) 997 __blk_flush_plug(plug, async); 998 } 999 1000 int blkdev_issue_flush(struct block_device *bdev); 1001 long nr_blockdev_pages(void); 1002 #else /* CONFIG_BLOCK */ 1003 struct blk_plug { 1004 }; 1005 1006 static inline void blk_start_plug_nr_ios(struct blk_plug *plug, 1007 unsigned short nr_ios) 1008 { 1009 } 1010 1011 static inline void blk_start_plug(struct blk_plug *plug) 1012 { 1013 } 1014 1015 static inline void blk_finish_plug(struct blk_plug *plug) 1016 { 1017 } 1018 1019 static inline void blk_flush_plug(struct blk_plug *plug, bool async) 1020 { 1021 } 1022 1023 static inline int blkdev_issue_flush(struct block_device *bdev) 1024 { 1025 return 0; 1026 } 1027 1028 static inline long nr_blockdev_pages(void) 1029 { 1030 return 0; 1031 } 1032 #endif /* CONFIG_BLOCK */ 1033 1034 extern void blk_io_schedule(void); 1035 1036 int blkdev_issue_discard(struct block_device *bdev, sector_t sector, 1037 sector_t nr_sects, gfp_t gfp_mask); 1038 int __blkdev_issue_discard(struct block_device *bdev, sector_t sector, 1039 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop); 1040 int blkdev_issue_secure_erase(struct block_device *bdev, sector_t sector, 1041 sector_t nr_sects, gfp_t gfp); 1042 1043 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */ 1044 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */ 1045 1046 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector, 1047 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop, 1048 unsigned flags); 1049 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector, 1050 sector_t nr_sects, gfp_t gfp_mask, unsigned flags); 1051 1052 static inline int sb_issue_discard(struct super_block *sb, sector_t block, 1053 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags) 1054 { 1055 return blkdev_issue_discard(sb->s_bdev, 1056 block << (sb->s_blocksize_bits - 1057 SECTOR_SHIFT), 1058 nr_blocks << (sb->s_blocksize_bits - 1059 SECTOR_SHIFT), 1060 gfp_mask); 1061 } 1062 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block, 1063 sector_t nr_blocks, gfp_t gfp_mask) 1064 { 1065 return blkdev_issue_zeroout(sb->s_bdev, 1066 block << (sb->s_blocksize_bits - 1067 SECTOR_SHIFT), 1068 nr_blocks << (sb->s_blocksize_bits - 1069 SECTOR_SHIFT), 1070 gfp_mask, 0); 1071 } 1072 1073 static inline bool bdev_is_partition(struct block_device *bdev) 1074 { 1075 return bdev->bd_partno; 1076 } 1077 1078 enum blk_default_limits { 1079 BLK_MAX_SEGMENTS = 128, 1080 BLK_SAFE_MAX_SECTORS = 255, 1081 BLK_MAX_SEGMENT_SIZE = 65536, 1082 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL, 1083 }; 1084 1085 #define BLK_DEF_MAX_SECTORS 2560u 1086 1087 static inline unsigned long queue_segment_boundary(const struct request_queue *q) 1088 { 1089 return q->limits.seg_boundary_mask; 1090 } 1091 1092 static inline unsigned long queue_virt_boundary(const struct request_queue *q) 1093 { 1094 return q->limits.virt_boundary_mask; 1095 } 1096 1097 static inline unsigned int queue_max_sectors(const struct request_queue *q) 1098 { 1099 return q->limits.max_sectors; 1100 } 1101 1102 static inline unsigned int queue_max_bytes(struct request_queue *q) 1103 { 1104 return min_t(unsigned int, queue_max_sectors(q), INT_MAX >> 9) << 9; 1105 } 1106 1107 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q) 1108 { 1109 return q->limits.max_hw_sectors; 1110 } 1111 1112 static inline unsigned short queue_max_segments(const struct request_queue *q) 1113 { 1114 return q->limits.max_segments; 1115 } 1116 1117 static inline unsigned short queue_max_discard_segments(const struct request_queue *q) 1118 { 1119 return q->limits.max_discard_segments; 1120 } 1121 1122 static inline unsigned int queue_max_segment_size(const struct request_queue *q) 1123 { 1124 return q->limits.max_segment_size; 1125 } 1126 1127 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q) 1128 { 1129 1130 const struct queue_limits *l = &q->limits; 1131 1132 return min(l->max_zone_append_sectors, l->max_sectors); 1133 } 1134 1135 static inline unsigned int 1136 bdev_max_zone_append_sectors(struct block_device *bdev) 1137 { 1138 return queue_max_zone_append_sectors(bdev_get_queue(bdev)); 1139 } 1140 1141 static inline unsigned int bdev_max_segments(struct block_device *bdev) 1142 { 1143 return queue_max_segments(bdev_get_queue(bdev)); 1144 } 1145 1146 static inline unsigned queue_logical_block_size(const struct request_queue *q) 1147 { 1148 int retval = 512; 1149 1150 if (q && q->limits.logical_block_size) 1151 retval = q->limits.logical_block_size; 1152 1153 return retval; 1154 } 1155 1156 static inline unsigned int bdev_logical_block_size(struct block_device *bdev) 1157 { 1158 return queue_logical_block_size(bdev_get_queue(bdev)); 1159 } 1160 1161 static inline unsigned int queue_physical_block_size(const struct request_queue *q) 1162 { 1163 return q->limits.physical_block_size; 1164 } 1165 1166 static inline unsigned int bdev_physical_block_size(struct block_device *bdev) 1167 { 1168 return queue_physical_block_size(bdev_get_queue(bdev)); 1169 } 1170 1171 static inline unsigned int queue_io_min(const struct request_queue *q) 1172 { 1173 return q->limits.io_min; 1174 } 1175 1176 static inline int bdev_io_min(struct block_device *bdev) 1177 { 1178 return queue_io_min(bdev_get_queue(bdev)); 1179 } 1180 1181 static inline unsigned int queue_io_opt(const struct request_queue *q) 1182 { 1183 return q->limits.io_opt; 1184 } 1185 1186 static inline int bdev_io_opt(struct block_device *bdev) 1187 { 1188 return queue_io_opt(bdev_get_queue(bdev)); 1189 } 1190 1191 static inline unsigned int 1192 queue_zone_write_granularity(const struct request_queue *q) 1193 { 1194 return q->limits.zone_write_granularity; 1195 } 1196 1197 static inline unsigned int 1198 bdev_zone_write_granularity(struct block_device *bdev) 1199 { 1200 return queue_zone_write_granularity(bdev_get_queue(bdev)); 1201 } 1202 1203 int bdev_alignment_offset(struct block_device *bdev); 1204 unsigned int bdev_discard_alignment(struct block_device *bdev); 1205 1206 static inline unsigned int bdev_max_discard_sectors(struct block_device *bdev) 1207 { 1208 return bdev_get_queue(bdev)->limits.max_discard_sectors; 1209 } 1210 1211 static inline unsigned int bdev_discard_granularity(struct block_device *bdev) 1212 { 1213 return bdev_get_queue(bdev)->limits.discard_granularity; 1214 } 1215 1216 static inline unsigned int 1217 bdev_max_secure_erase_sectors(struct block_device *bdev) 1218 { 1219 return bdev_get_queue(bdev)->limits.max_secure_erase_sectors; 1220 } 1221 1222 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev) 1223 { 1224 struct request_queue *q = bdev_get_queue(bdev); 1225 1226 if (q) 1227 return q->limits.max_write_zeroes_sectors; 1228 1229 return 0; 1230 } 1231 1232 static inline bool bdev_nonrot(struct block_device *bdev) 1233 { 1234 return blk_queue_nonrot(bdev_get_queue(bdev)); 1235 } 1236 1237 static inline bool bdev_synchronous(struct block_device *bdev) 1238 { 1239 return test_bit(QUEUE_FLAG_SYNCHRONOUS, 1240 &bdev_get_queue(bdev)->queue_flags); 1241 } 1242 1243 static inline bool bdev_stable_writes(struct block_device *bdev) 1244 { 1245 return test_bit(QUEUE_FLAG_STABLE_WRITES, 1246 &bdev_get_queue(bdev)->queue_flags); 1247 } 1248 1249 static inline bool bdev_write_cache(struct block_device *bdev) 1250 { 1251 return test_bit(QUEUE_FLAG_WC, &bdev_get_queue(bdev)->queue_flags); 1252 } 1253 1254 static inline bool bdev_fua(struct block_device *bdev) 1255 { 1256 return test_bit(QUEUE_FLAG_FUA, &bdev_get_queue(bdev)->queue_flags); 1257 } 1258 1259 static inline bool bdev_nowait(struct block_device *bdev) 1260 { 1261 return test_bit(QUEUE_FLAG_NOWAIT, &bdev_get_queue(bdev)->queue_flags); 1262 } 1263 1264 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev) 1265 { 1266 return blk_queue_zoned_model(bdev_get_queue(bdev)); 1267 } 1268 1269 static inline bool bdev_is_zoned(struct block_device *bdev) 1270 { 1271 return blk_queue_is_zoned(bdev_get_queue(bdev)); 1272 } 1273 1274 static inline unsigned int bdev_zone_no(struct block_device *bdev, sector_t sec) 1275 { 1276 return disk_zone_no(bdev->bd_disk, sec); 1277 } 1278 1279 /* Whether write serialization is required for @op on zoned devices. */ 1280 static inline bool op_needs_zoned_write_locking(enum req_op op) 1281 { 1282 return op == REQ_OP_WRITE || op == REQ_OP_WRITE_ZEROES; 1283 } 1284 1285 static inline bool bdev_op_is_zoned_write(struct block_device *bdev, 1286 enum req_op op) 1287 { 1288 return bdev_is_zoned(bdev) && op_needs_zoned_write_locking(op); 1289 } 1290 1291 static inline sector_t bdev_zone_sectors(struct block_device *bdev) 1292 { 1293 struct request_queue *q = bdev_get_queue(bdev); 1294 1295 if (!blk_queue_is_zoned(q)) 1296 return 0; 1297 return q->limits.chunk_sectors; 1298 } 1299 1300 static inline sector_t bdev_offset_from_zone_start(struct block_device *bdev, 1301 sector_t sector) 1302 { 1303 return sector & (bdev_zone_sectors(bdev) - 1); 1304 } 1305 1306 static inline bool bdev_is_zone_start(struct block_device *bdev, 1307 sector_t sector) 1308 { 1309 return bdev_offset_from_zone_start(bdev, sector) == 0; 1310 } 1311 1312 static inline int queue_dma_alignment(const struct request_queue *q) 1313 { 1314 return q ? q->limits.dma_alignment : 511; 1315 } 1316 1317 static inline unsigned int bdev_dma_alignment(struct block_device *bdev) 1318 { 1319 return queue_dma_alignment(bdev_get_queue(bdev)); 1320 } 1321 1322 static inline bool bdev_iter_is_aligned(struct block_device *bdev, 1323 struct iov_iter *iter) 1324 { 1325 return iov_iter_is_aligned(iter, bdev_dma_alignment(bdev), 1326 bdev_logical_block_size(bdev) - 1); 1327 } 1328 1329 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr, 1330 unsigned int len) 1331 { 1332 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask; 1333 return !(addr & alignment) && !(len & alignment); 1334 } 1335 1336 /* assumes size > 256 */ 1337 static inline unsigned int blksize_bits(unsigned int size) 1338 { 1339 return order_base_2(size >> SECTOR_SHIFT) + SECTOR_SHIFT; 1340 } 1341 1342 static inline unsigned int block_size(struct block_device *bdev) 1343 { 1344 return 1 << bdev->bd_inode->i_blkbits; 1345 } 1346 1347 int kblockd_schedule_work(struct work_struct *work); 1348 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay); 1349 1350 #define MODULE_ALIAS_BLOCKDEV(major,minor) \ 1351 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor)) 1352 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \ 1353 MODULE_ALIAS("block-major-" __stringify(major) "-*") 1354 1355 #ifdef CONFIG_BLK_INLINE_ENCRYPTION 1356 1357 bool blk_crypto_register(struct blk_crypto_profile *profile, 1358 struct request_queue *q); 1359 1360 #else /* CONFIG_BLK_INLINE_ENCRYPTION */ 1361 1362 static inline bool blk_crypto_register(struct blk_crypto_profile *profile, 1363 struct request_queue *q) 1364 { 1365 return true; 1366 } 1367 1368 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */ 1369 1370 enum blk_unique_id { 1371 /* these match the Designator Types specified in SPC */ 1372 BLK_UID_T10 = 1, 1373 BLK_UID_EUI64 = 2, 1374 BLK_UID_NAA = 3, 1375 }; 1376 1377 struct block_device_operations { 1378 void (*submit_bio)(struct bio *bio); 1379 int (*poll_bio)(struct bio *bio, struct io_comp_batch *iob, 1380 unsigned int flags); 1381 int (*open)(struct gendisk *disk, blk_mode_t mode); 1382 void (*release)(struct gendisk *disk); 1383 int (*ioctl)(struct block_device *bdev, blk_mode_t mode, 1384 unsigned cmd, unsigned long arg); 1385 int (*compat_ioctl)(struct block_device *bdev, blk_mode_t mode, 1386 unsigned cmd, unsigned long arg); 1387 unsigned int (*check_events) (struct gendisk *disk, 1388 unsigned int clearing); 1389 void (*unlock_native_capacity) (struct gendisk *); 1390 int (*getgeo)(struct block_device *, struct hd_geometry *); 1391 int (*set_read_only)(struct block_device *bdev, bool ro); 1392 void (*free_disk)(struct gendisk *disk); 1393 /* this callback is with swap_lock and sometimes page table lock held */ 1394 void (*swap_slot_free_notify) (struct block_device *, unsigned long); 1395 int (*report_zones)(struct gendisk *, sector_t sector, 1396 unsigned int nr_zones, report_zones_cb cb, void *data); 1397 char *(*devnode)(struct gendisk *disk, umode_t *mode); 1398 /* returns the length of the identifier or a negative errno: */ 1399 int (*get_unique_id)(struct gendisk *disk, u8 id[16], 1400 enum blk_unique_id id_type); 1401 struct module *owner; 1402 const struct pr_ops *pr_ops; 1403 1404 /* 1405 * Special callback for probing GPT entry at a given sector. 1406 * Needed by Android devices, used by GPT scanner and MMC blk 1407 * driver. 1408 */ 1409 int (*alternative_gpt_sector)(struct gendisk *disk, sector_t *sector); 1410 }; 1411 1412 #ifdef CONFIG_COMPAT 1413 extern int blkdev_compat_ptr_ioctl(struct block_device *, blk_mode_t, 1414 unsigned int, unsigned long); 1415 #else 1416 #define blkdev_compat_ptr_ioctl NULL 1417 #endif 1418 1419 static inline void blk_wake_io_task(struct task_struct *waiter) 1420 { 1421 /* 1422 * If we're polling, the task itself is doing the completions. For 1423 * that case, we don't need to signal a wakeup, it's enough to just 1424 * mark us as RUNNING. 1425 */ 1426 if (waiter == current) 1427 __set_current_state(TASK_RUNNING); 1428 else 1429 wake_up_process(waiter); 1430 } 1431 1432 unsigned long bdev_start_io_acct(struct block_device *bdev, enum req_op op, 1433 unsigned long start_time); 1434 void bdev_end_io_acct(struct block_device *bdev, enum req_op op, 1435 unsigned int sectors, unsigned long start_time); 1436 1437 unsigned long bio_start_io_acct(struct bio *bio); 1438 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time, 1439 struct block_device *orig_bdev); 1440 1441 /** 1442 * bio_end_io_acct - end I/O accounting for bio based drivers 1443 * @bio: bio to end account for 1444 * @start_time: start time returned by bio_start_io_acct() 1445 */ 1446 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time) 1447 { 1448 return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev); 1449 } 1450 1451 int bdev_read_only(struct block_device *bdev); 1452 int set_blocksize(struct block_device *bdev, int size); 1453 1454 int lookup_bdev(const char *pathname, dev_t *dev); 1455 1456 void blkdev_show(struct seq_file *seqf, off_t offset); 1457 1458 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */ 1459 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */ 1460 #ifdef CONFIG_BLOCK 1461 #define BLKDEV_MAJOR_MAX 512 1462 #else 1463 #define BLKDEV_MAJOR_MAX 0 1464 #endif 1465 1466 struct blk_holder_ops { 1467 void (*mark_dead)(struct block_device *bdev, bool surprise); 1468 1469 /* 1470 * Sync the file system mounted on the block device. 1471 */ 1472 void (*sync)(struct block_device *bdev); 1473 1474 /* 1475 * Freeze the file system mounted on the block device. 1476 */ 1477 int (*freeze)(struct block_device *bdev); 1478 1479 /* 1480 * Thaw the file system mounted on the block device. 1481 */ 1482 int (*thaw)(struct block_device *bdev); 1483 }; 1484 1485 /* 1486 * For filesystems using @fs_holder_ops, the @holder argument passed to 1487 * helpers used to open and claim block devices via 1488 * bd_prepare_to_claim() must point to a superblock. 1489 */ 1490 extern const struct blk_holder_ops fs_holder_ops; 1491 1492 /* 1493 * Return the correct open flags for blkdev_get_by_* for super block flags 1494 * as stored in sb->s_flags. 1495 */ 1496 #define sb_open_mode(flags) \ 1497 (BLK_OPEN_READ | BLK_OPEN_RESTRICT_WRITES | \ 1498 (((flags) & SB_RDONLY) ? 0 : BLK_OPEN_WRITE)) 1499 1500 struct bdev_handle { 1501 struct block_device *bdev; 1502 void *holder; 1503 blk_mode_t mode; 1504 }; 1505 1506 struct bdev_handle *bdev_open_by_dev(dev_t dev, blk_mode_t mode, void *holder, 1507 const struct blk_holder_ops *hops); 1508 struct bdev_handle *bdev_open_by_path(const char *path, blk_mode_t mode, 1509 void *holder, const struct blk_holder_ops *hops); 1510 int bd_prepare_to_claim(struct block_device *bdev, void *holder, 1511 const struct blk_holder_ops *hops); 1512 void bd_abort_claiming(struct block_device *bdev, void *holder); 1513 void bdev_release(struct bdev_handle *handle); 1514 1515 /* just for blk-cgroup, don't use elsewhere */ 1516 struct block_device *blkdev_get_no_open(dev_t dev); 1517 void blkdev_put_no_open(struct block_device *bdev); 1518 1519 struct block_device *I_BDEV(struct inode *inode); 1520 1521 #ifdef CONFIG_BLOCK 1522 void invalidate_bdev(struct block_device *bdev); 1523 int sync_blockdev(struct block_device *bdev); 1524 int sync_blockdev_range(struct block_device *bdev, loff_t lstart, loff_t lend); 1525 int sync_blockdev_nowait(struct block_device *bdev); 1526 void sync_bdevs(bool wait); 1527 void bdev_statx_dioalign(struct inode *inode, struct kstat *stat); 1528 void printk_all_partitions(void); 1529 int __init early_lookup_bdev(const char *pathname, dev_t *dev); 1530 #else 1531 static inline void invalidate_bdev(struct block_device *bdev) 1532 { 1533 } 1534 static inline int sync_blockdev(struct block_device *bdev) 1535 { 1536 return 0; 1537 } 1538 static inline int sync_blockdev_nowait(struct block_device *bdev) 1539 { 1540 return 0; 1541 } 1542 static inline void sync_bdevs(bool wait) 1543 { 1544 } 1545 static inline void bdev_statx_dioalign(struct inode *inode, struct kstat *stat) 1546 { 1547 } 1548 static inline void printk_all_partitions(void) 1549 { 1550 } 1551 static inline int early_lookup_bdev(const char *pathname, dev_t *dev) 1552 { 1553 return -EINVAL; 1554 } 1555 #endif /* CONFIG_BLOCK */ 1556 1557 int bdev_freeze(struct block_device *bdev); 1558 int bdev_thaw(struct block_device *bdev); 1559 1560 struct io_comp_batch { 1561 struct request *req_list; 1562 bool need_ts; 1563 void (*complete)(struct io_comp_batch *); 1564 }; 1565 1566 #define DEFINE_IO_COMP_BATCH(name) struct io_comp_batch name = { } 1567 1568 #endif /* _LINUX_BLKDEV_H */ 1569