1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef _LINUX_BLKDEV_H
3 #define _LINUX_BLKDEV_H
5 #include <linux/sched.h>
6 #include <linux/genhd.h>
7 #include <linux/list.h>
8 #include <linux/llist.h>
9 #include <linux/minmax.h>
10 #include <linux/timer.h>
11 #include <linux/workqueue.h>
12 #include <linux/wait.h>
13 #include <linux/bio.h>
14 #include <linux/gfp.h>
15 #include <linux/rcupdate.h>
16 #include <linux/percpu-refcount.h>
17 #include <linux/blkzoned.h>
18 #include <linux/sbitmap.h>
19 #include <linux/srcu.h>
23 struct elevator_queue;
28 struct blk_flush_queue;
32 struct blk_queue_stats;
33 struct blk_stat_callback;
34 struct blk_crypto_profile;
36 /* Must be consistent with blk_mq_poll_stats_bkt() */
37 #define BLK_MQ_POLL_STATS_BKTS 16
39 /* Doing classic polling */
40 #define BLK_MQ_POLL_CLASSIC -1
43 * Maximum number of blkcg policies allowed to be registered concurrently.
44 * Defined here to simplify include dependency.
46 #define BLKCG_MAX_POLS 6
48 static inline int blk_validate_block_size(unsigned long bsize)
50 if (bsize < 512 || bsize > PAGE_SIZE || !is_power_of_2(bsize))
56 static inline bool blk_op_is_passthrough(unsigned int op)
59 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
63 * Zoned block device models (zoned limit).
65 * Note: This needs to be ordered from the least to the most severe
66 * restrictions for the inheritance in blk_stack_limits() to work.
68 enum blk_zoned_model {
69 BLK_ZONED_NONE = 0, /* Regular block device */
70 BLK_ZONED_HA, /* Host-aware zoned block device */
71 BLK_ZONED_HM, /* Host-managed zoned block device */
75 * BLK_BOUNCE_NONE: never bounce (default)
76 * BLK_BOUNCE_HIGH: bounce all highmem pages
84 enum blk_bounce bounce;
85 unsigned long seg_boundary_mask;
86 unsigned long virt_boundary_mask;
88 unsigned int max_hw_sectors;
89 unsigned int max_dev_sectors;
90 unsigned int chunk_sectors;
91 unsigned int max_sectors;
92 unsigned int max_segment_size;
93 unsigned int physical_block_size;
94 unsigned int logical_block_size;
95 unsigned int alignment_offset;
98 unsigned int max_discard_sectors;
99 unsigned int max_hw_discard_sectors;
100 unsigned int max_write_same_sectors;
101 unsigned int max_write_zeroes_sectors;
102 unsigned int max_zone_append_sectors;
103 unsigned int discard_granularity;
104 unsigned int discard_alignment;
105 unsigned int zone_write_granularity;
107 unsigned short max_segments;
108 unsigned short max_integrity_segments;
109 unsigned short max_discard_segments;
111 unsigned char misaligned;
112 unsigned char discard_misaligned;
113 unsigned char raid_partial_stripes_expensive;
114 enum blk_zoned_model zoned;
117 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
120 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
122 #ifdef CONFIG_BLK_DEV_ZONED
124 #define BLK_ALL_ZONES ((unsigned int)-1)
125 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
126 unsigned int nr_zones, report_zones_cb cb, void *data);
127 unsigned int blkdev_nr_zones(struct gendisk *disk);
128 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
129 sector_t sectors, sector_t nr_sectors,
131 int blk_revalidate_disk_zones(struct gendisk *disk,
132 void (*update_driver_data)(struct gendisk *disk));
134 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
135 unsigned int cmd, unsigned long arg);
136 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
137 unsigned int cmd, unsigned long arg);
139 #else /* CONFIG_BLK_DEV_ZONED */
141 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
146 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
147 fmode_t mode, unsigned int cmd,
153 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
154 fmode_t mode, unsigned int cmd,
160 #endif /* CONFIG_BLK_DEV_ZONED */
163 * Independent access ranges: struct blk_independent_access_range describes
164 * a range of contiguous sectors that can be accessed using device command
165 * execution resources that are independent from the resources used for
166 * other access ranges. This is typically found with single-LUN multi-actuator
167 * HDDs where each access range is served by a different set of heads.
168 * The set of independent ranges supported by the device is defined using
169 * struct blk_independent_access_ranges. The independent ranges must not overlap
170 * and must include all sectors within the disk capacity (no sector holes
172 * For a device with multiple ranges, requests targeting sectors in different
173 * ranges can be executed in parallel. A request can straddle an access range
176 struct blk_independent_access_range {
178 struct request_queue *queue;
183 struct blk_independent_access_ranges {
185 bool sysfs_registered;
186 unsigned int nr_ia_ranges;
187 struct blk_independent_access_range ia_range[];
190 struct request_queue {
191 struct request *last_merge;
192 struct elevator_queue *elevator;
194 struct percpu_ref q_usage_counter;
196 struct blk_queue_stats *stats;
197 struct rq_qos *rq_qos;
199 const struct blk_mq_ops *mq_ops;
202 struct blk_mq_ctx __percpu *queue_ctx;
204 unsigned int queue_depth;
206 /* hw dispatch queues */
207 struct blk_mq_hw_ctx **queue_hw_ctx;
208 unsigned int nr_hw_queues;
211 * The queue owner gets to use this for whatever they like.
212 * ll_rw_blk doesn't touch it.
217 * various queue flags, see QUEUE_* below
219 unsigned long queue_flags;
221 * Number of contexts that have called blk_set_pm_only(). If this
222 * counter is above zero then only RQF_PM requests are processed.
227 * ida allocated id for this queue. Used to index queues from
232 spinlock_t queue_lock;
234 struct gendisk *disk;
244 struct kobject *mq_kobj;
246 #ifdef CONFIG_BLK_DEV_INTEGRITY
247 struct blk_integrity integrity;
248 #endif /* CONFIG_BLK_DEV_INTEGRITY */
252 enum rpm_status rpm_status;
258 unsigned long nr_requests; /* Max # of requests */
260 unsigned int dma_pad_mask;
261 unsigned int dma_alignment;
263 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
264 struct blk_crypto_profile *crypto_profile;
267 unsigned int rq_timeout;
270 struct blk_stat_callback *poll_cb;
271 struct blk_rq_stat *poll_stat;
273 struct timer_list timeout;
274 struct work_struct timeout_work;
276 atomic_t nr_active_requests_shared_tags;
278 struct blk_mq_tags *sched_shared_tags;
280 struct list_head icq_list;
281 #ifdef CONFIG_BLK_CGROUP
282 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
283 struct blkcg_gq *root_blkg;
284 struct list_head blkg_list;
287 struct queue_limits limits;
289 unsigned int required_elevator_features;
291 #ifdef CONFIG_BLK_DEV_ZONED
293 * Zoned block device information for request dispatch control.
294 * nr_zones is the total number of zones of the device. This is always
295 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
296 * bits which indicates if a zone is conventional (bit set) or
297 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
298 * bits which indicates if a zone is write locked, that is, if a write
299 * request targeting the zone was dispatched. All three fields are
300 * initialized by the low level device driver (e.g. scsi/sd.c).
301 * Stacking drivers (device mappers) may or may not initialize
304 * Reads of this information must be protected with blk_queue_enter() /
305 * blk_queue_exit(). Modifying this information is only allowed while
306 * no requests are being processed. See also blk_mq_freeze_queue() and
307 * blk_mq_unfreeze_queue().
309 unsigned int nr_zones;
310 unsigned long *conv_zones_bitmap;
311 unsigned long *seq_zones_wlock;
312 unsigned int max_open_zones;
313 unsigned int max_active_zones;
314 #endif /* CONFIG_BLK_DEV_ZONED */
317 struct mutex debugfs_mutex;
318 #ifdef CONFIG_BLK_DEV_IO_TRACE
319 struct blk_trace __rcu *blk_trace;
322 * for flush operations
324 struct blk_flush_queue *fq;
326 struct list_head requeue_list;
327 spinlock_t requeue_lock;
328 struct delayed_work requeue_work;
330 struct mutex sysfs_lock;
331 struct mutex sysfs_dir_lock;
334 * for reusing dead hctx instance in case of updating
337 struct list_head unused_hctx_list;
338 spinlock_t unused_hctx_lock;
342 #ifdef CONFIG_BLK_DEV_THROTTLING
344 struct throtl_data *td;
346 struct rcu_head rcu_head;
347 wait_queue_head_t mq_freeze_wq;
349 * Protect concurrent access to q_usage_counter by
350 * percpu_ref_kill() and percpu_ref_reinit().
352 struct mutex mq_freeze_lock;
356 struct blk_mq_tag_set *tag_set;
357 struct list_head tag_set_list;
358 struct bio_set bio_split;
360 struct dentry *debugfs_dir;
362 #ifdef CONFIG_BLK_DEBUG_FS
363 struct dentry *sched_debugfs_dir;
364 struct dentry *rqos_debugfs_dir;
367 bool mq_sysfs_init_done;
369 #define BLK_MAX_WRITE_HINTS 5
370 u64 write_hints[BLK_MAX_WRITE_HINTS];
373 * Independent sector access ranges. This is always NULL for
374 * devices that do not have multiple independent access ranges.
376 struct blk_independent_access_ranges *ia_ranges;
379 * @srcu: Sleepable RCU. Use as lock when type of the request queue
380 * is blocking (BLK_MQ_F_BLOCKING). Must be the last member
382 struct srcu_struct srcu[];
385 /* Keep blk_queue_flag_name[] in sync with the definitions below */
386 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
387 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
388 #define QUEUE_FLAG_HAS_SRCU 2 /* SRCU is allocated */
389 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
390 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
391 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
392 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
393 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
394 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
395 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
396 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
397 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
398 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
399 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
400 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
401 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
402 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
403 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
404 #define QUEUE_FLAG_WC 17 /* Write back caching */
405 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
406 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
407 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
408 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
409 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
410 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
411 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
412 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
413 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
414 #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */
416 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
417 (1 << QUEUE_FLAG_SAME_COMP) | \
418 (1 << QUEUE_FLAG_NOWAIT))
420 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
421 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
422 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
424 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
425 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
426 #define blk_queue_has_srcu(q) test_bit(QUEUE_FLAG_HAS_SRCU, &(q)->queue_flags)
427 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
428 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
429 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
430 #define blk_queue_noxmerges(q) \
431 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
432 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
433 #define blk_queue_stable_writes(q) \
434 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
435 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
436 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
437 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
438 #define blk_queue_zone_resetall(q) \
439 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
440 #define blk_queue_secure_erase(q) \
441 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
442 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
443 #define blk_queue_pci_p2pdma(q) \
444 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
445 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
446 #define blk_queue_rq_alloc_time(q) \
447 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
449 #define blk_queue_rq_alloc_time(q) false
452 #define blk_noretry_request(rq) \
453 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
454 REQ_FAILFAST_DRIVER))
455 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
456 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
457 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
458 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
459 #define blk_queue_nowait(q) test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
461 extern void blk_set_pm_only(struct request_queue *q);
462 extern void blk_clear_pm_only(struct request_queue *q);
464 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
466 #define dma_map_bvec(dev, bv, dir, attrs) \
467 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
470 static inline bool queue_is_mq(struct request_queue *q)
476 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
478 return q->rpm_status;
481 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
487 static inline enum blk_zoned_model
488 blk_queue_zoned_model(struct request_queue *q)
490 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
491 return q->limits.zoned;
492 return BLK_ZONED_NONE;
495 static inline bool blk_queue_is_zoned(struct request_queue *q)
497 switch (blk_queue_zoned_model(q)) {
506 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
508 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
511 #ifdef CONFIG_BLK_DEV_ZONED
512 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
514 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
517 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
520 if (!blk_queue_is_zoned(q))
522 return sector >> ilog2(q->limits.chunk_sectors);
525 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
528 if (!blk_queue_is_zoned(q))
530 if (!q->conv_zones_bitmap)
532 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
535 static inline void blk_queue_max_open_zones(struct request_queue *q,
536 unsigned int max_open_zones)
538 q->max_open_zones = max_open_zones;
541 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
543 return q->max_open_zones;
546 static inline void blk_queue_max_active_zones(struct request_queue *q,
547 unsigned int max_active_zones)
549 q->max_active_zones = max_active_zones;
552 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
554 return q->max_active_zones;
556 #else /* CONFIG_BLK_DEV_ZONED */
557 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
561 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
566 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
571 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
575 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
579 #endif /* CONFIG_BLK_DEV_ZONED */
581 static inline unsigned int blk_queue_depth(struct request_queue *q)
584 return q->queue_depth;
586 return q->nr_requests;
590 * default timeout for SG_IO if none specified
592 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
593 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
595 /* This should not be used directly - use rq_for_each_segment */
596 #define for_each_bio(_bio) \
597 for (; _bio; _bio = _bio->bi_next)
600 extern int blk_register_queue(struct gendisk *disk);
601 extern void blk_unregister_queue(struct gendisk *disk);
602 void submit_bio_noacct(struct bio *bio);
604 extern int blk_lld_busy(struct request_queue *q);
605 extern void blk_queue_split(struct bio **);
606 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
607 extern void blk_queue_exit(struct request_queue *q);
608 extern void blk_sync_queue(struct request_queue *q);
610 /* Helper to convert REQ_OP_XXX to its string format XXX */
611 extern const char *blk_op_str(unsigned int op);
613 int blk_status_to_errno(blk_status_t status);
614 blk_status_t errno_to_blk_status(int errno);
616 /* only poll the hardware once, don't continue until a completion was found */
617 #define BLK_POLL_ONESHOT (1 << 0)
618 /* do not sleep to wait for the expected completion time */
619 #define BLK_POLL_NOSLEEP (1 << 1)
620 int bio_poll(struct bio *bio, struct io_comp_batch *iob, unsigned int flags);
621 int iocb_bio_iopoll(struct kiocb *kiocb, struct io_comp_batch *iob,
624 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
626 return bdev->bd_queue; /* this is never NULL */
629 #ifdef CONFIG_BLK_DEV_ZONED
631 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
632 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
634 static inline unsigned int bio_zone_no(struct bio *bio)
636 return blk_queue_zone_no(bdev_get_queue(bio->bi_bdev),
637 bio->bi_iter.bi_sector);
640 static inline unsigned int bio_zone_is_seq(struct bio *bio)
642 return blk_queue_zone_is_seq(bdev_get_queue(bio->bi_bdev),
643 bio->bi_iter.bi_sector);
645 #endif /* CONFIG_BLK_DEV_ZONED */
647 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
650 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
651 return min(q->limits.max_discard_sectors,
652 UINT_MAX >> SECTOR_SHIFT);
654 if (unlikely(op == REQ_OP_WRITE_SAME))
655 return q->limits.max_write_same_sectors;
657 if (unlikely(op == REQ_OP_WRITE_ZEROES))
658 return q->limits.max_write_zeroes_sectors;
660 return q->limits.max_sectors;
664 * Return maximum size of a request at given offset. Only valid for
665 * file system requests.
667 static inline unsigned int blk_max_size_offset(struct request_queue *q,
669 unsigned int chunk_sectors)
671 if (!chunk_sectors) {
672 if (q->limits.chunk_sectors)
673 chunk_sectors = q->limits.chunk_sectors;
675 return q->limits.max_sectors;
678 if (likely(is_power_of_2(chunk_sectors)))
679 chunk_sectors -= offset & (chunk_sectors - 1);
681 chunk_sectors -= sector_div(offset, chunk_sectors);
683 return min(q->limits.max_sectors, chunk_sectors);
687 * Access functions for manipulating queue properties
689 extern void blk_cleanup_queue(struct request_queue *);
690 void blk_queue_bounce_limit(struct request_queue *q, enum blk_bounce limit);
691 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
692 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
693 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
694 extern void blk_queue_max_discard_segments(struct request_queue *,
696 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
697 extern void blk_queue_max_discard_sectors(struct request_queue *q,
698 unsigned int max_discard_sectors);
699 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
700 unsigned int max_write_same_sectors);
701 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
702 unsigned int max_write_same_sectors);
703 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
704 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
705 unsigned int max_zone_append_sectors);
706 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
707 void blk_queue_zone_write_granularity(struct request_queue *q,
709 extern void blk_queue_alignment_offset(struct request_queue *q,
710 unsigned int alignment);
711 void disk_update_readahead(struct gendisk *disk);
712 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
713 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
714 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
715 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
716 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
717 extern void blk_set_default_limits(struct queue_limits *lim);
718 extern void blk_set_stacking_limits(struct queue_limits *lim);
719 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
721 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
723 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
724 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
725 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
726 extern void blk_queue_dma_alignment(struct request_queue *, int);
727 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
728 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
729 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
731 struct blk_independent_access_ranges *
732 disk_alloc_independent_access_ranges(struct gendisk *disk, int nr_ia_ranges);
733 void disk_set_independent_access_ranges(struct gendisk *disk,
734 struct blk_independent_access_ranges *iars);
737 * Elevator features for blk_queue_required_elevator_features:
739 /* Supports zoned block devices sequential write constraint */
740 #define ELEVATOR_F_ZBD_SEQ_WRITE (1U << 0)
741 /* Supports scheduling on multiple hardware queues */
742 #define ELEVATOR_F_MQ_AWARE (1U << 1)
744 extern void blk_queue_required_elevator_features(struct request_queue *q,
745 unsigned int features);
746 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
749 bool __must_check blk_get_queue(struct request_queue *);
750 extern void blk_put_queue(struct request_queue *);
751 extern void blk_set_queue_dying(struct request_queue *);
755 * blk_plug permits building a queue of related requests by holding the I/O
756 * fragments for a short period. This allows merging of sequential requests
757 * into single larger request. As the requests are moved from a per-task list to
758 * the device's request_queue in a batch, this results in improved scalability
759 * as the lock contention for request_queue lock is reduced.
761 * It is ok not to disable preemption when adding the request to the plug list
762 * or when attempting a merge. For details, please see schedule() where
763 * blk_flush_plug() is called.
766 struct request *mq_list; /* blk-mq requests */
768 /* if ios_left is > 1, we can batch tag/rq allocations */
769 struct request *cached_rq;
770 unsigned short nr_ios;
772 unsigned short rq_count;
774 bool multiple_queues;
778 struct list_head cb_list; /* md requires an unplug callback */
782 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
784 struct list_head list;
785 blk_plug_cb_fn callback;
788 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
789 void *data, int size);
790 extern void blk_start_plug(struct blk_plug *);
791 extern void blk_start_plug_nr_ios(struct blk_plug *, unsigned short);
792 extern void blk_finish_plug(struct blk_plug *);
794 void blk_flush_plug(struct blk_plug *plug, bool from_schedule);
796 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
798 struct blk_plug *plug = tsk->plug;
801 (plug->mq_list || !list_empty(&plug->cb_list));
804 int blkdev_issue_flush(struct block_device *bdev);
805 long nr_blockdev_pages(void);
806 #else /* CONFIG_BLOCK */
810 static inline void blk_start_plug_nr_ios(struct blk_plug *plug,
811 unsigned short nr_ios)
815 static inline void blk_start_plug(struct blk_plug *plug)
819 static inline void blk_finish_plug(struct blk_plug *plug)
823 static inline void blk_flush_plug(struct blk_plug *plug, bool async)
827 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
832 static inline int blkdev_issue_flush(struct block_device *bdev)
837 static inline long nr_blockdev_pages(void)
841 #endif /* CONFIG_BLOCK */
843 extern void blk_io_schedule(void);
845 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
846 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
848 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
850 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
851 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
852 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
853 sector_t nr_sects, gfp_t gfp_mask, int flags,
856 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
857 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
859 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
860 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
862 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
863 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
865 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
866 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
868 return blkdev_issue_discard(sb->s_bdev,
869 block << (sb->s_blocksize_bits -
871 nr_blocks << (sb->s_blocksize_bits -
875 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
876 sector_t nr_blocks, gfp_t gfp_mask)
878 return blkdev_issue_zeroout(sb->s_bdev,
879 block << (sb->s_blocksize_bits -
881 nr_blocks << (sb->s_blocksize_bits -
886 static inline bool bdev_is_partition(struct block_device *bdev)
888 return bdev->bd_partno;
891 enum blk_default_limits {
892 BLK_MAX_SEGMENTS = 128,
893 BLK_SAFE_MAX_SECTORS = 255,
894 BLK_DEF_MAX_SECTORS = 2560,
895 BLK_MAX_SEGMENT_SIZE = 65536,
896 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
899 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
901 return q->limits.seg_boundary_mask;
904 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
906 return q->limits.virt_boundary_mask;
909 static inline unsigned int queue_max_sectors(const struct request_queue *q)
911 return q->limits.max_sectors;
914 static inline unsigned int queue_max_bytes(struct request_queue *q)
916 return min_t(unsigned int, queue_max_sectors(q), INT_MAX >> 9) << 9;
919 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
921 return q->limits.max_hw_sectors;
924 static inline unsigned short queue_max_segments(const struct request_queue *q)
926 return q->limits.max_segments;
929 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
931 return q->limits.max_discard_segments;
934 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
936 return q->limits.max_segment_size;
939 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
942 const struct queue_limits *l = &q->limits;
944 return min(l->max_zone_append_sectors, l->max_sectors);
947 static inline unsigned queue_logical_block_size(const struct request_queue *q)
951 if (q && q->limits.logical_block_size)
952 retval = q->limits.logical_block_size;
957 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
959 return queue_logical_block_size(bdev_get_queue(bdev));
962 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
964 return q->limits.physical_block_size;
967 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
969 return queue_physical_block_size(bdev_get_queue(bdev));
972 static inline unsigned int queue_io_min(const struct request_queue *q)
974 return q->limits.io_min;
977 static inline int bdev_io_min(struct block_device *bdev)
979 return queue_io_min(bdev_get_queue(bdev));
982 static inline unsigned int queue_io_opt(const struct request_queue *q)
984 return q->limits.io_opt;
987 static inline int bdev_io_opt(struct block_device *bdev)
989 return queue_io_opt(bdev_get_queue(bdev));
992 static inline unsigned int
993 queue_zone_write_granularity(const struct request_queue *q)
995 return q->limits.zone_write_granularity;
998 static inline unsigned int
999 bdev_zone_write_granularity(struct block_device *bdev)
1001 return queue_zone_write_granularity(bdev_get_queue(bdev));
1004 static inline int queue_alignment_offset(const struct request_queue *q)
1006 if (q->limits.misaligned)
1009 return q->limits.alignment_offset;
1012 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1014 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1015 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1018 return (granularity + lim->alignment_offset - alignment) % granularity;
1021 static inline int bdev_alignment_offset(struct block_device *bdev)
1023 struct request_queue *q = bdev_get_queue(bdev);
1025 if (q->limits.misaligned)
1027 if (bdev_is_partition(bdev))
1028 return queue_limit_alignment_offset(&q->limits,
1029 bdev->bd_start_sect);
1030 return q->limits.alignment_offset;
1033 static inline int queue_discard_alignment(const struct request_queue *q)
1035 if (q->limits.discard_misaligned)
1038 return q->limits.discard_alignment;
1041 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1043 unsigned int alignment, granularity, offset;
1045 if (!lim->max_discard_sectors)
1048 /* Why are these in bytes, not sectors? */
1049 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1050 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1054 /* Offset of the partition start in 'granularity' sectors */
1055 offset = sector_div(sector, granularity);
1057 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1058 offset = (granularity + alignment - offset) % granularity;
1060 /* Turn it back into bytes, gaah */
1061 return offset << SECTOR_SHIFT;
1064 static inline int bdev_discard_alignment(struct block_device *bdev)
1066 struct request_queue *q = bdev_get_queue(bdev);
1068 if (bdev_is_partition(bdev))
1069 return queue_limit_discard_alignment(&q->limits,
1070 bdev->bd_start_sect);
1071 return q->limits.discard_alignment;
1074 static inline unsigned int bdev_write_same(struct block_device *bdev)
1076 struct request_queue *q = bdev_get_queue(bdev);
1079 return q->limits.max_write_same_sectors;
1084 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1086 struct request_queue *q = bdev_get_queue(bdev);
1089 return q->limits.max_write_zeroes_sectors;
1094 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1096 struct request_queue *q = bdev_get_queue(bdev);
1099 return blk_queue_zoned_model(q);
1101 return BLK_ZONED_NONE;
1104 static inline bool bdev_is_zoned(struct block_device *bdev)
1106 struct request_queue *q = bdev_get_queue(bdev);
1109 return blk_queue_is_zoned(q);
1114 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1116 struct request_queue *q = bdev_get_queue(bdev);
1119 return blk_queue_zone_sectors(q);
1123 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1125 struct request_queue *q = bdev_get_queue(bdev);
1128 return queue_max_open_zones(q);
1132 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1134 struct request_queue *q = bdev_get_queue(bdev);
1137 return queue_max_active_zones(q);
1141 static inline int queue_dma_alignment(const struct request_queue *q)
1143 return q ? q->dma_alignment : 511;
1146 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1149 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1150 return !(addr & alignment) && !(len & alignment);
1153 /* assumes size > 256 */
1154 static inline unsigned int blksize_bits(unsigned int size)
1156 unsigned int bits = 8;
1160 } while (size > 256);
1164 static inline unsigned int block_size(struct block_device *bdev)
1166 return 1 << bdev->bd_inode->i_blkbits;
1169 int kblockd_schedule_work(struct work_struct *work);
1170 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1172 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1173 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1174 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1175 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1177 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1179 bool blk_crypto_register(struct blk_crypto_profile *profile,
1180 struct request_queue *q);
1182 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1184 static inline bool blk_crypto_register(struct blk_crypto_profile *profile,
1185 struct request_queue *q)
1190 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1192 enum blk_unique_id {
1193 /* these match the Designator Types specified in SPC */
1199 #define NFL4_UFLG_MASK 0x0000003F
1201 struct block_device_operations {
1202 void (*submit_bio)(struct bio *bio);
1203 int (*open) (struct block_device *, fmode_t);
1204 void (*release) (struct gendisk *, fmode_t);
1205 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1206 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1207 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1208 unsigned int (*check_events) (struct gendisk *disk,
1209 unsigned int clearing);
1210 void (*unlock_native_capacity) (struct gendisk *);
1211 int (*getgeo)(struct block_device *, struct hd_geometry *);
1212 int (*set_read_only)(struct block_device *bdev, bool ro);
1213 /* this callback is with swap_lock and sometimes page table lock held */
1214 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1215 int (*report_zones)(struct gendisk *, sector_t sector,
1216 unsigned int nr_zones, report_zones_cb cb, void *data);
1217 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1218 /* returns the length of the identifier or a negative errno: */
1219 int (*get_unique_id)(struct gendisk *disk, u8 id[16],
1220 enum blk_unique_id id_type);
1221 struct module *owner;
1222 const struct pr_ops *pr_ops;
1225 * Special callback for probing GPT entry at a given sector.
1226 * Needed by Android devices, used by GPT scanner and MMC blk
1229 int (*alternative_gpt_sector)(struct gendisk *disk, sector_t *sector);
1232 #ifdef CONFIG_COMPAT
1233 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1234 unsigned int, unsigned long);
1236 #define blkdev_compat_ptr_ioctl NULL
1239 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1240 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1241 struct writeback_control *);
1243 static inline void blk_wake_io_task(struct task_struct *waiter)
1246 * If we're polling, the task itself is doing the completions. For
1247 * that case, we don't need to signal a wakeup, it's enough to just
1248 * mark us as RUNNING.
1250 if (waiter == current)
1251 __set_current_state(TASK_RUNNING);
1253 wake_up_process(waiter);
1256 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1258 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1259 unsigned long start_time);
1261 unsigned long bio_start_io_acct(struct bio *bio);
1262 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1263 struct block_device *orig_bdev);
1266 * bio_end_io_acct - end I/O accounting for bio based drivers
1267 * @bio: bio to end account for
1268 * @start: start time returned by bio_start_io_acct()
1270 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1272 return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1275 int bdev_read_only(struct block_device *bdev);
1276 int set_blocksize(struct block_device *bdev, int size);
1278 const char *bdevname(struct block_device *bdev, char *buffer);
1279 int lookup_bdev(const char *pathname, dev_t *dev);
1281 void blkdev_show(struct seq_file *seqf, off_t offset);
1283 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1284 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1286 #define BLKDEV_MAJOR_MAX 512
1288 #define BLKDEV_MAJOR_MAX 0
1291 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1293 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1294 int bd_prepare_to_claim(struct block_device *bdev, void *holder);
1295 void bd_abort_claiming(struct block_device *bdev, void *holder);
1296 void blkdev_put(struct block_device *bdev, fmode_t mode);
1298 /* just for blk-cgroup, don't use elsewhere */
1299 struct block_device *blkdev_get_no_open(dev_t dev);
1300 void blkdev_put_no_open(struct block_device *bdev);
1302 struct block_device *bdev_alloc(struct gendisk *disk, u8 partno);
1303 void bdev_add(struct block_device *bdev, dev_t dev);
1304 struct block_device *I_BDEV(struct inode *inode);
1305 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
1309 void invalidate_bdev(struct block_device *bdev);
1310 int sync_blockdev(struct block_device *bdev);
1311 int sync_blockdev_nowait(struct block_device *bdev);
1312 void sync_bdevs(bool wait);
1314 static inline void invalidate_bdev(struct block_device *bdev)
1317 static inline int sync_blockdev(struct block_device *bdev)
1321 static inline int sync_blockdev_nowait(struct block_device *bdev)
1325 static inline void sync_bdevs(bool wait)
1329 int fsync_bdev(struct block_device *bdev);
1331 int freeze_bdev(struct block_device *bdev);
1332 int thaw_bdev(struct block_device *bdev);
1334 struct io_comp_batch {
1335 struct request *req_list;
1337 void (*complete)(struct io_comp_batch *);
1340 #define DEFINE_IO_COMP_BATCH(name) struct io_comp_batch name = { }
1342 #endif /* _LINUX_BLKDEV_H */