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/sched/clock.h>
7 #include <linux/major.h>
8 #include <linux/genhd.h>
9 #include <linux/list.h>
10 #include <linux/llist.h>
11 #include <linux/minmax.h>
12 #include <linux/timer.h>
13 #include <linux/workqueue.h>
14 #include <linux/backing-dev-defs.h>
15 #include <linux/wait.h>
16 #include <linux/mempool.h>
17 #include <linux/pfn.h>
18 #include <linux/bio.h>
19 #include <linux/stringify.h>
20 #include <linux/gfp.h>
21 #include <linux/bsg.h>
22 #include <linux/smp.h>
23 #include <linux/rcupdate.h>
24 #include <linux/percpu-refcount.h>
25 #include <linux/scatterlist.h>
26 #include <linux/blkzoned.h>
30 struct scsi_ioctl_command;
33 struct elevator_queue;
39 struct blk_flush_queue;
42 struct blk_queue_stats;
43 struct blk_stat_callback;
44 struct blk_keyslot_manager;
46 #define BLKDEV_MIN_RQ 4
47 #define BLKDEV_MAX_RQ 128 /* Default maximum */
49 /* Must be consistent with blk_mq_poll_stats_bkt() */
50 #define BLK_MQ_POLL_STATS_BKTS 16
52 /* Doing classic polling */
53 #define BLK_MQ_POLL_CLASSIC -1
56 * Maximum number of blkcg policies allowed to be registered concurrently.
57 * Defined here to simplify include dependency.
59 #define BLKCG_MAX_POLS 5
61 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
65 typedef __u32 __bitwise req_flags_t;
67 /* drive already may have started this one */
68 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
69 /* may not be passed by ioscheduler */
70 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
71 /* request for flush sequence */
72 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
73 /* merge of different types, fail separately */
74 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
75 /* track inflight for MQ */
76 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
77 /* don't call prep for this one */
78 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
79 /* vaguely specified driver internal error. Ignored by the block layer */
80 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
81 /* don't warn about errors */
82 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
83 /* elevator private data attached */
84 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
85 /* account into disk and partition IO statistics */
86 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
87 /* runtime pm request */
88 #define RQF_PM ((__force req_flags_t)(1 << 15))
89 /* on IO scheduler merge hash */
90 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
91 /* track IO completion time */
92 #define RQF_STATS ((__force req_flags_t)(1 << 17))
93 /* Look at ->special_vec for the actual data payload instead of the
95 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
96 /* The per-zone write lock is held for this request */
97 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
98 /* already slept for hybrid poll */
99 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
100 /* ->timeout has been called, don't expire again */
101 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
103 /* flags that prevent us from merging requests: */
104 #define RQF_NOMERGE_FLAGS \
105 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
108 * Request state for blk-mq.
117 * Try to put the fields that are referenced together in the same cacheline.
119 * If you modify this structure, make sure to update blk_rq_init() and
120 * especially blk_mq_rq_ctx_init() to take care of the added fields.
123 struct request_queue *q;
124 struct blk_mq_ctx *mq_ctx;
125 struct blk_mq_hw_ctx *mq_hctx;
127 unsigned int cmd_flags; /* op and common flags */
128 req_flags_t rq_flags;
133 /* the following two fields are internal, NEVER access directly */
134 unsigned int __data_len; /* total data len */
135 sector_t __sector; /* sector cursor */
140 struct list_head queuelist;
143 * The hash is used inside the scheduler, and killed once the
144 * request reaches the dispatch list. The ipi_list is only used
145 * to queue the request for softirq completion, which is long
146 * after the request has been unhashed (and even removed from
147 * the dispatch list).
150 struct hlist_node hash; /* merge hash */
151 struct llist_node ipi_list;
155 * The rb_node is only used inside the io scheduler, requests
156 * are pruned when moved to the dispatch queue. So let the
157 * completion_data share space with the rb_node.
160 struct rb_node rb_node; /* sort/lookup */
161 struct bio_vec special_vec;
162 void *completion_data;
163 int error_count; /* for legacy drivers, don't use */
167 * Three pointers are available for the IO schedulers, if they need
168 * more they have to dynamically allocate it. Flush requests are
169 * never put on the IO scheduler. So let the flush fields share
170 * space with the elevator data.
180 struct list_head list;
181 rq_end_io_fn *saved_end_io;
185 struct gendisk *rq_disk;
186 struct block_device *part;
187 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
188 /* Time that the first bio started allocating this request. */
191 /* Time that this request was allocated for this IO. */
193 /* Time that I/O was submitted to the device. */
194 u64 io_start_time_ns;
196 #ifdef CONFIG_BLK_WBT
197 unsigned short wbt_flags;
200 * rq sectors used for blk stats. It has the same value
201 * with blk_rq_sectors(rq), except that it never be zeroed
204 unsigned short stats_sectors;
207 * Number of scatter-gather DMA addr+len pairs after
208 * physical address coalescing is performed.
210 unsigned short nr_phys_segments;
212 #if defined(CONFIG_BLK_DEV_INTEGRITY)
213 unsigned short nr_integrity_segments;
216 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
217 struct bio_crypt_ctx *crypt_ctx;
218 struct blk_ksm_keyslot *crypt_keyslot;
221 unsigned short write_hint;
222 unsigned short ioprio;
224 enum mq_rq_state state;
227 unsigned int timeout;
228 unsigned long deadline;
231 struct __call_single_data csd;
236 * completion callback.
238 rq_end_io_fn *end_io;
242 static inline bool blk_op_is_scsi(unsigned int op)
244 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
247 static inline bool blk_op_is_private(unsigned int op)
249 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
252 static inline bool blk_rq_is_scsi(struct request *rq)
254 return blk_op_is_scsi(req_op(rq));
257 static inline bool blk_rq_is_private(struct request *rq)
259 return blk_op_is_private(req_op(rq));
262 static inline bool blk_rq_is_passthrough(struct request *rq)
264 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
267 static inline bool bio_is_passthrough(struct bio *bio)
269 unsigned op = bio_op(bio);
271 return blk_op_is_scsi(op) || blk_op_is_private(op);
274 static inline bool blk_op_is_passthrough(unsigned int op)
276 return (blk_op_is_scsi(op & REQ_OP_MASK) ||
277 blk_op_is_private(op & REQ_OP_MASK));
280 static inline unsigned short req_get_ioprio(struct request *req)
285 #include <linux/elevator.h>
287 struct blk_queue_ctx;
291 enum blk_eh_timer_return {
292 BLK_EH_DONE, /* drivers has completed the command */
293 BLK_EH_RESET_TIMER, /* reset timer and try again */
296 enum blk_queue_state {
301 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
302 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
304 #define BLK_SCSI_MAX_CMDS (256)
305 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
308 * Zoned block device models (zoned limit).
310 * Note: This needs to be ordered from the least to the most severe
311 * restrictions for the inheritance in blk_stack_limits() to work.
313 enum blk_zoned_model {
314 BLK_ZONED_NONE = 0, /* Regular block device */
315 BLK_ZONED_HA, /* Host-aware zoned block device */
316 BLK_ZONED_HM, /* Host-managed zoned block device */
320 * BLK_BOUNCE_NONE: never bounce (default)
321 * BLK_BOUNCE_HIGH: bounce all highmem pages
328 struct queue_limits {
329 enum blk_bounce bounce;
330 unsigned long seg_boundary_mask;
331 unsigned long virt_boundary_mask;
333 unsigned int max_hw_sectors;
334 unsigned int max_dev_sectors;
335 unsigned int chunk_sectors;
336 unsigned int max_sectors;
337 unsigned int max_segment_size;
338 unsigned int physical_block_size;
339 unsigned int logical_block_size;
340 unsigned int alignment_offset;
343 unsigned int max_discard_sectors;
344 unsigned int max_hw_discard_sectors;
345 unsigned int max_write_same_sectors;
346 unsigned int max_write_zeroes_sectors;
347 unsigned int max_zone_append_sectors;
348 unsigned int discard_granularity;
349 unsigned int discard_alignment;
350 unsigned int zone_write_granularity;
352 unsigned short max_segments;
353 unsigned short max_integrity_segments;
354 unsigned short max_discard_segments;
356 unsigned char misaligned;
357 unsigned char discard_misaligned;
358 unsigned char raid_partial_stripes_expensive;
359 enum blk_zoned_model zoned;
362 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
365 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
367 #ifdef CONFIG_BLK_DEV_ZONED
369 #define BLK_ALL_ZONES ((unsigned int)-1)
370 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
371 unsigned int nr_zones, report_zones_cb cb, void *data);
372 unsigned int blkdev_nr_zones(struct gendisk *disk);
373 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
374 sector_t sectors, sector_t nr_sectors,
376 int blk_revalidate_disk_zones(struct gendisk *disk,
377 void (*update_driver_data)(struct gendisk *disk));
379 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
380 unsigned int cmd, unsigned long arg);
381 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
382 unsigned int cmd, unsigned long arg);
384 #else /* CONFIG_BLK_DEV_ZONED */
386 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
391 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
392 fmode_t mode, unsigned int cmd,
398 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
399 fmode_t mode, unsigned int cmd,
405 #endif /* CONFIG_BLK_DEV_ZONED */
407 struct request_queue {
408 struct request *last_merge;
409 struct elevator_queue *elevator;
411 struct percpu_ref q_usage_counter;
413 struct blk_queue_stats *stats;
414 struct rq_qos *rq_qos;
416 const struct blk_mq_ops *mq_ops;
419 struct blk_mq_ctx __percpu *queue_ctx;
421 unsigned int queue_depth;
423 /* hw dispatch queues */
424 struct blk_mq_hw_ctx **queue_hw_ctx;
425 unsigned int nr_hw_queues;
427 struct backing_dev_info *backing_dev_info;
430 * The queue owner gets to use this for whatever they like.
431 * ll_rw_blk doesn't touch it.
436 * various queue flags, see QUEUE_* below
438 unsigned long queue_flags;
440 * Number of contexts that have called blk_set_pm_only(). If this
441 * counter is above zero then only RQF_PM requests are processed.
446 * ida allocated id for this queue. Used to index queues from
451 spinlock_t queue_lock;
461 struct kobject *mq_kobj;
463 #ifdef CONFIG_BLK_DEV_INTEGRITY
464 struct blk_integrity integrity;
465 #endif /* CONFIG_BLK_DEV_INTEGRITY */
469 enum rpm_status rpm_status;
475 unsigned long nr_requests; /* Max # of requests */
477 unsigned int dma_pad_mask;
478 unsigned int dma_alignment;
480 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
481 /* Inline crypto capabilities */
482 struct blk_keyslot_manager *ksm;
485 unsigned int rq_timeout;
488 struct blk_stat_callback *poll_cb;
489 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
491 struct timer_list timeout;
492 struct work_struct timeout_work;
494 atomic_t nr_active_requests_shared_sbitmap;
496 struct list_head icq_list;
497 #ifdef CONFIG_BLK_CGROUP
498 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
499 struct blkcg_gq *root_blkg;
500 struct list_head blkg_list;
503 struct queue_limits limits;
505 unsigned int required_elevator_features;
507 #ifdef CONFIG_BLK_DEV_ZONED
509 * Zoned block device information for request dispatch control.
510 * nr_zones is the total number of zones of the device. This is always
511 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
512 * bits which indicates if a zone is conventional (bit set) or
513 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
514 * bits which indicates if a zone is write locked, that is, if a write
515 * request targeting the zone was dispatched. All three fields are
516 * initialized by the low level device driver (e.g. scsi/sd.c).
517 * Stacking drivers (device mappers) may or may not initialize
520 * Reads of this information must be protected with blk_queue_enter() /
521 * blk_queue_exit(). Modifying this information is only allowed while
522 * no requests are being processed. See also blk_mq_freeze_queue() and
523 * blk_mq_unfreeze_queue().
525 unsigned int nr_zones;
526 unsigned long *conv_zones_bitmap;
527 unsigned long *seq_zones_wlock;
528 unsigned int max_open_zones;
529 unsigned int max_active_zones;
530 #endif /* CONFIG_BLK_DEV_ZONED */
535 unsigned int sg_timeout;
536 unsigned int sg_reserved_size;
538 struct mutex debugfs_mutex;
539 #ifdef CONFIG_BLK_DEV_IO_TRACE
540 struct blk_trace __rcu *blk_trace;
543 * for flush operations
545 struct blk_flush_queue *fq;
547 struct list_head requeue_list;
548 spinlock_t requeue_lock;
549 struct delayed_work requeue_work;
551 struct mutex sysfs_lock;
552 struct mutex sysfs_dir_lock;
555 * for reusing dead hctx instance in case of updating
558 struct list_head unused_hctx_list;
559 spinlock_t unused_hctx_lock;
563 #if defined(CONFIG_BLK_DEV_BSG)
564 struct bsg_class_device bsg_dev;
567 #ifdef CONFIG_BLK_DEV_THROTTLING
569 struct throtl_data *td;
571 struct rcu_head rcu_head;
572 wait_queue_head_t mq_freeze_wq;
574 * Protect concurrent access to q_usage_counter by
575 * percpu_ref_kill() and percpu_ref_reinit().
577 struct mutex mq_freeze_lock;
579 struct blk_mq_tag_set *tag_set;
580 struct list_head tag_set_list;
581 struct bio_set bio_split;
583 struct dentry *debugfs_dir;
585 #ifdef CONFIG_BLK_DEBUG_FS
586 struct dentry *sched_debugfs_dir;
587 struct dentry *rqos_debugfs_dir;
590 bool mq_sysfs_init_done;
594 #define BLK_MAX_WRITE_HINTS 5
595 u64 write_hints[BLK_MAX_WRITE_HINTS];
598 /* Keep blk_queue_flag_name[] in sync with the definitions below */
599 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
600 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
601 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
602 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
603 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
604 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
605 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
606 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
607 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
608 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
609 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
610 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
611 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
612 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
613 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
614 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
615 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
616 #define QUEUE_FLAG_WC 17 /* Write back caching */
617 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
618 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
619 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
620 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
621 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
622 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
623 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
624 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
625 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
626 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
627 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
628 #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */
630 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
631 (1 << QUEUE_FLAG_SAME_COMP) | \
632 (1 << QUEUE_FLAG_NOWAIT))
634 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
635 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
636 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
638 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
639 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
640 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
641 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
642 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
643 #define blk_queue_noxmerges(q) \
644 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
645 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
646 #define blk_queue_stable_writes(q) \
647 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
648 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
649 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
650 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
651 #define blk_queue_zone_resetall(q) \
652 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
653 #define blk_queue_secure_erase(q) \
654 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
655 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
656 #define blk_queue_scsi_passthrough(q) \
657 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
658 #define blk_queue_pci_p2pdma(q) \
659 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
660 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
661 #define blk_queue_rq_alloc_time(q) \
662 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
664 #define blk_queue_rq_alloc_time(q) false
667 #define blk_noretry_request(rq) \
668 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
669 REQ_FAILFAST_DRIVER))
670 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
671 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
672 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
673 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
674 #define blk_queue_nowait(q) test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
676 extern void blk_set_pm_only(struct request_queue *q);
677 extern void blk_clear_pm_only(struct request_queue *q);
679 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
681 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
683 #define rq_dma_dir(rq) \
684 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
686 #define dma_map_bvec(dev, bv, dir, attrs) \
687 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
690 #define queue_to_disk(q) (dev_to_disk(kobj_to_dev((q)->kobj.parent)))
692 static inline bool queue_is_mq(struct request_queue *q)
698 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
700 return q->rpm_status;
703 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
709 static inline enum blk_zoned_model
710 blk_queue_zoned_model(struct request_queue *q)
712 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
713 return q->limits.zoned;
714 return BLK_ZONED_NONE;
717 static inline bool blk_queue_is_zoned(struct request_queue *q)
719 switch (blk_queue_zoned_model(q)) {
728 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
730 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
733 #ifdef CONFIG_BLK_DEV_ZONED
734 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
736 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
739 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
742 if (!blk_queue_is_zoned(q))
744 return sector >> ilog2(q->limits.chunk_sectors);
747 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
750 if (!blk_queue_is_zoned(q))
752 if (!q->conv_zones_bitmap)
754 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
757 static inline void blk_queue_max_open_zones(struct request_queue *q,
758 unsigned int max_open_zones)
760 q->max_open_zones = max_open_zones;
763 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
765 return q->max_open_zones;
768 static inline void blk_queue_max_active_zones(struct request_queue *q,
769 unsigned int max_active_zones)
771 q->max_active_zones = max_active_zones;
774 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
776 return q->max_active_zones;
778 #else /* CONFIG_BLK_DEV_ZONED */
779 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
783 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
788 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
793 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
797 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
801 #endif /* CONFIG_BLK_DEV_ZONED */
803 static inline bool rq_is_sync(struct request *rq)
805 return op_is_sync(rq->cmd_flags);
808 static inline bool rq_mergeable(struct request *rq)
810 if (blk_rq_is_passthrough(rq))
813 if (req_op(rq) == REQ_OP_FLUSH)
816 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
819 if (req_op(rq) == REQ_OP_ZONE_APPEND)
822 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
824 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
830 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
832 if (bio_page(a) == bio_page(b) &&
833 bio_offset(a) == bio_offset(b))
839 static inline unsigned int blk_queue_depth(struct request_queue *q)
842 return q->queue_depth;
844 return q->nr_requests;
848 * default timeout for SG_IO if none specified
850 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
851 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
857 unsigned long offset;
862 struct req_iterator {
863 struct bvec_iter iter;
867 /* This should not be used directly - use rq_for_each_segment */
868 #define for_each_bio(_bio) \
869 for (; _bio; _bio = _bio->bi_next)
870 #define __rq_for_each_bio(_bio, rq) \
872 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
874 #define rq_for_each_segment(bvl, _rq, _iter) \
875 __rq_for_each_bio(_iter.bio, _rq) \
876 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
878 #define rq_for_each_bvec(bvl, _rq, _iter) \
879 __rq_for_each_bio(_iter.bio, _rq) \
880 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
882 #define rq_iter_last(bvec, _iter) \
883 (_iter.bio->bi_next == NULL && \
884 bio_iter_last(bvec, _iter.iter))
886 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
887 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
889 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
890 extern void rq_flush_dcache_pages(struct request *rq);
892 static inline void rq_flush_dcache_pages(struct request *rq)
897 extern int blk_register_queue(struct gendisk *disk);
898 extern void blk_unregister_queue(struct gendisk *disk);
899 blk_qc_t submit_bio_noacct(struct bio *bio);
900 extern void blk_rq_init(struct request_queue *q, struct request *rq);
901 extern void blk_put_request(struct request *);
902 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
903 blk_mq_req_flags_t flags);
904 extern int blk_lld_busy(struct request_queue *q);
905 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
906 struct bio_set *bs, gfp_t gfp_mask,
907 int (*bio_ctr)(struct bio *, struct bio *, void *),
909 extern void blk_rq_unprep_clone(struct request *rq);
910 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
912 int blk_rq_append_bio(struct request *rq, struct bio *bio);
913 extern void blk_queue_split(struct bio **);
914 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
915 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
916 unsigned int, void __user *);
917 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
918 unsigned int, void __user *);
919 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
920 struct scsi_ioctl_command __user *);
921 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
922 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
924 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
925 extern void blk_queue_exit(struct request_queue *q);
926 extern void blk_sync_queue(struct request_queue *q);
927 extern int blk_rq_map_user(struct request_queue *, struct request *,
928 struct rq_map_data *, void __user *, unsigned long,
930 extern int blk_rq_unmap_user(struct bio *);
931 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
932 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
933 struct rq_map_data *, const struct iov_iter *,
935 extern void blk_execute_rq(struct gendisk *, struct request *, int);
936 extern void blk_execute_rq_nowait(struct gendisk *,
937 struct request *, int, rq_end_io_fn *);
939 /* Helper to convert REQ_OP_XXX to its string format XXX */
940 extern const char *blk_op_str(unsigned int op);
942 int blk_status_to_errno(blk_status_t status);
943 blk_status_t errno_to_blk_status(int errno);
945 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
947 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
949 return bdev->bd_disk->queue; /* this is never NULL */
953 * The basic unit of block I/O is a sector. It is used in a number of contexts
954 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
955 * bytes. Variables of type sector_t represent an offset or size that is a
956 * multiple of 512 bytes. Hence these two constants.
959 #define SECTOR_SHIFT 9
962 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
966 * blk_rq_pos() : the current sector
967 * blk_rq_bytes() : bytes left in the entire request
968 * blk_rq_cur_bytes() : bytes left in the current segment
969 * blk_rq_err_bytes() : bytes left till the next error boundary
970 * blk_rq_sectors() : sectors left in the entire request
971 * blk_rq_cur_sectors() : sectors left in the current segment
972 * blk_rq_stats_sectors() : sectors of the entire request used for stats
974 static inline sector_t blk_rq_pos(const struct request *rq)
979 static inline unsigned int blk_rq_bytes(const struct request *rq)
981 return rq->__data_len;
984 static inline int blk_rq_cur_bytes(const struct request *rq)
986 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
989 extern unsigned int blk_rq_err_bytes(const struct request *rq);
991 static inline unsigned int blk_rq_sectors(const struct request *rq)
993 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
996 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
998 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
1001 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
1003 return rq->stats_sectors;
1006 #ifdef CONFIG_BLK_DEV_ZONED
1008 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
1009 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
1011 static inline unsigned int blk_rq_zone_no(struct request *rq)
1013 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1016 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1018 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1020 #endif /* CONFIG_BLK_DEV_ZONED */
1023 * Some commands like WRITE SAME have a payload or data transfer size which
1024 * is different from the size of the request. Any driver that supports such
1025 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1026 * calculate the data transfer size.
1028 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1030 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1031 return rq->special_vec.bv_len;
1032 return blk_rq_bytes(rq);
1036 * Return the first full biovec in the request. The caller needs to check that
1037 * there are any bvecs before calling this helper.
1039 static inline struct bio_vec req_bvec(struct request *rq)
1041 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1042 return rq->special_vec;
1043 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1046 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1049 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1050 return min(q->limits.max_discard_sectors,
1051 UINT_MAX >> SECTOR_SHIFT);
1053 if (unlikely(op == REQ_OP_WRITE_SAME))
1054 return q->limits.max_write_same_sectors;
1056 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1057 return q->limits.max_write_zeroes_sectors;
1059 return q->limits.max_sectors;
1063 * Return maximum size of a request at given offset. Only valid for
1064 * file system requests.
1066 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1068 unsigned int chunk_sectors)
1070 if (!chunk_sectors) {
1071 if (q->limits.chunk_sectors)
1072 chunk_sectors = q->limits.chunk_sectors;
1074 return q->limits.max_sectors;
1077 if (likely(is_power_of_2(chunk_sectors)))
1078 chunk_sectors -= offset & (chunk_sectors - 1);
1080 chunk_sectors -= sector_div(offset, chunk_sectors);
1082 return min(q->limits.max_sectors, chunk_sectors);
1085 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1088 struct request_queue *q = rq->q;
1090 if (blk_rq_is_passthrough(rq))
1091 return q->limits.max_hw_sectors;
1093 if (!q->limits.chunk_sectors ||
1094 req_op(rq) == REQ_OP_DISCARD ||
1095 req_op(rq) == REQ_OP_SECURE_ERASE)
1096 return blk_queue_get_max_sectors(q, req_op(rq));
1098 return min(blk_max_size_offset(q, offset, 0),
1099 blk_queue_get_max_sectors(q, req_op(rq)));
1102 static inline unsigned int blk_rq_count_bios(struct request *rq)
1104 unsigned int nr_bios = 0;
1107 __rq_for_each_bio(bio, rq)
1113 void blk_steal_bios(struct bio_list *list, struct request *rq);
1116 * Request completion related functions.
1118 * blk_update_request() completes given number of bytes and updates
1119 * the request without completing it.
1121 extern bool blk_update_request(struct request *rq, blk_status_t error,
1122 unsigned int nr_bytes);
1124 extern void blk_abort_request(struct request *);
1127 * Access functions for manipulating queue properties
1129 extern void blk_cleanup_queue(struct request_queue *);
1130 void blk_queue_bounce_limit(struct request_queue *q, enum blk_bounce limit);
1131 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1132 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1133 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1134 extern void blk_queue_max_discard_segments(struct request_queue *,
1136 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1137 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1138 unsigned int max_discard_sectors);
1139 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1140 unsigned int max_write_same_sectors);
1141 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1142 unsigned int max_write_same_sectors);
1143 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1144 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1145 unsigned int max_zone_append_sectors);
1146 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1147 void blk_queue_zone_write_granularity(struct request_queue *q,
1149 extern void blk_queue_alignment_offset(struct request_queue *q,
1150 unsigned int alignment);
1151 void blk_queue_update_readahead(struct request_queue *q);
1152 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1153 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1154 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1155 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1156 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1157 extern void blk_set_default_limits(struct queue_limits *lim);
1158 extern void blk_set_stacking_limits(struct queue_limits *lim);
1159 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1161 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1163 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1164 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1165 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1166 extern void blk_queue_dma_alignment(struct request_queue *, int);
1167 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1168 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1169 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1170 extern void blk_queue_required_elevator_features(struct request_queue *q,
1171 unsigned int features);
1172 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1173 struct device *dev);
1176 * Number of physical segments as sent to the device.
1178 * Normally this is the number of discontiguous data segments sent by the
1179 * submitter. But for data-less command like discard we might have no
1180 * actual data segments submitted, but the driver might have to add it's
1181 * own special payload. In that case we still return 1 here so that this
1182 * special payload will be mapped.
1184 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1186 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1188 return rq->nr_phys_segments;
1192 * Number of discard segments (or ranges) the driver needs to fill in.
1193 * Each discard bio merged into a request is counted as one segment.
1195 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1197 return max_t(unsigned short, rq->nr_phys_segments, 1);
1200 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1201 struct scatterlist *sglist, struct scatterlist **last_sg);
1202 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1203 struct scatterlist *sglist)
1205 struct scatterlist *last_sg = NULL;
1207 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1209 extern void blk_dump_rq_flags(struct request *, char *);
1211 bool __must_check blk_get_queue(struct request_queue *);
1212 struct request_queue *blk_alloc_queue(int node_id);
1213 extern void blk_put_queue(struct request_queue *);
1214 extern void blk_set_queue_dying(struct request_queue *);
1218 * blk_plug permits building a queue of related requests by holding the I/O
1219 * fragments for a short period. This allows merging of sequential requests
1220 * into single larger request. As the requests are moved from a per-task list to
1221 * the device's request_queue in a batch, this results in improved scalability
1222 * as the lock contention for request_queue lock is reduced.
1224 * It is ok not to disable preemption when adding the request to the plug list
1225 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1226 * the plug list when the task sleeps by itself. For details, please see
1227 * schedule() where blk_schedule_flush_plug() is called.
1230 struct list_head mq_list; /* blk-mq requests */
1231 struct list_head cb_list; /* md requires an unplug callback */
1232 unsigned short rq_count;
1233 bool multiple_queues;
1236 #define BLK_MAX_REQUEST_COUNT 16
1237 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1240 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1241 struct blk_plug_cb {
1242 struct list_head list;
1243 blk_plug_cb_fn callback;
1246 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1247 void *data, int size);
1248 extern void blk_start_plug(struct blk_plug *);
1249 extern void blk_finish_plug(struct blk_plug *);
1250 extern void blk_flush_plug_list(struct blk_plug *, bool);
1252 static inline void blk_flush_plug(struct task_struct *tsk)
1254 struct blk_plug *plug = tsk->plug;
1257 blk_flush_plug_list(plug, false);
1260 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1262 struct blk_plug *plug = tsk->plug;
1265 blk_flush_plug_list(plug, true);
1268 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1270 struct blk_plug *plug = tsk->plug;
1273 (!list_empty(&plug->mq_list) ||
1274 !list_empty(&plug->cb_list));
1277 int blkdev_issue_flush(struct block_device *bdev);
1278 long nr_blockdev_pages(void);
1279 #else /* CONFIG_BLOCK */
1283 static inline void blk_start_plug(struct blk_plug *plug)
1287 static inline void blk_finish_plug(struct blk_plug *plug)
1291 static inline void blk_flush_plug(struct task_struct *task)
1295 static inline void blk_schedule_flush_plug(struct task_struct *task)
1300 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1305 static inline int blkdev_issue_flush(struct block_device *bdev)
1310 static inline long nr_blockdev_pages(void)
1314 #endif /* CONFIG_BLOCK */
1316 extern void blk_io_schedule(void);
1318 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1319 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1321 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1323 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1324 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1325 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1326 sector_t nr_sects, gfp_t gfp_mask, int flags,
1329 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1330 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1332 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1333 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1335 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1336 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1338 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1339 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1341 return blkdev_issue_discard(sb->s_bdev,
1342 block << (sb->s_blocksize_bits -
1344 nr_blocks << (sb->s_blocksize_bits -
1348 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1349 sector_t nr_blocks, gfp_t gfp_mask)
1351 return blkdev_issue_zeroout(sb->s_bdev,
1352 block << (sb->s_blocksize_bits -
1354 nr_blocks << (sb->s_blocksize_bits -
1359 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1361 static inline bool bdev_is_partition(struct block_device *bdev)
1363 return bdev->bd_partno;
1366 enum blk_default_limits {
1367 BLK_MAX_SEGMENTS = 128,
1368 BLK_SAFE_MAX_SECTORS = 255,
1369 BLK_DEF_MAX_SECTORS = 2560,
1370 BLK_MAX_SEGMENT_SIZE = 65536,
1371 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1374 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1376 return q->limits.seg_boundary_mask;
1379 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1381 return q->limits.virt_boundary_mask;
1384 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1386 return q->limits.max_sectors;
1389 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1391 return q->limits.max_hw_sectors;
1394 static inline unsigned short queue_max_segments(const struct request_queue *q)
1396 return q->limits.max_segments;
1399 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1401 return q->limits.max_discard_segments;
1404 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1406 return q->limits.max_segment_size;
1409 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1412 const struct queue_limits *l = &q->limits;
1414 return min(l->max_zone_append_sectors, l->max_sectors);
1417 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1421 if (q && q->limits.logical_block_size)
1422 retval = q->limits.logical_block_size;
1427 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1429 return queue_logical_block_size(bdev_get_queue(bdev));
1432 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1434 return q->limits.physical_block_size;
1437 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1439 return queue_physical_block_size(bdev_get_queue(bdev));
1442 static inline unsigned int queue_io_min(const struct request_queue *q)
1444 return q->limits.io_min;
1447 static inline int bdev_io_min(struct block_device *bdev)
1449 return queue_io_min(bdev_get_queue(bdev));
1452 static inline unsigned int queue_io_opt(const struct request_queue *q)
1454 return q->limits.io_opt;
1457 static inline int bdev_io_opt(struct block_device *bdev)
1459 return queue_io_opt(bdev_get_queue(bdev));
1462 static inline unsigned int
1463 queue_zone_write_granularity(const struct request_queue *q)
1465 return q->limits.zone_write_granularity;
1468 static inline unsigned int
1469 bdev_zone_write_granularity(struct block_device *bdev)
1471 return queue_zone_write_granularity(bdev_get_queue(bdev));
1474 static inline int queue_alignment_offset(const struct request_queue *q)
1476 if (q->limits.misaligned)
1479 return q->limits.alignment_offset;
1482 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1484 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1485 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1488 return (granularity + lim->alignment_offset - alignment) % granularity;
1491 static inline int bdev_alignment_offset(struct block_device *bdev)
1493 struct request_queue *q = bdev_get_queue(bdev);
1495 if (q->limits.misaligned)
1497 if (bdev_is_partition(bdev))
1498 return queue_limit_alignment_offset(&q->limits,
1499 bdev->bd_start_sect);
1500 return q->limits.alignment_offset;
1503 static inline int queue_discard_alignment(const struct request_queue *q)
1505 if (q->limits.discard_misaligned)
1508 return q->limits.discard_alignment;
1511 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1513 unsigned int alignment, granularity, offset;
1515 if (!lim->max_discard_sectors)
1518 /* Why are these in bytes, not sectors? */
1519 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1520 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1524 /* Offset of the partition start in 'granularity' sectors */
1525 offset = sector_div(sector, granularity);
1527 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1528 offset = (granularity + alignment - offset) % granularity;
1530 /* Turn it back into bytes, gaah */
1531 return offset << SECTOR_SHIFT;
1534 static inline int bdev_discard_alignment(struct block_device *bdev)
1536 struct request_queue *q = bdev_get_queue(bdev);
1538 if (bdev_is_partition(bdev))
1539 return queue_limit_discard_alignment(&q->limits,
1540 bdev->bd_start_sect);
1541 return q->limits.discard_alignment;
1544 static inline unsigned int bdev_write_same(struct block_device *bdev)
1546 struct request_queue *q = bdev_get_queue(bdev);
1549 return q->limits.max_write_same_sectors;
1554 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1556 struct request_queue *q = bdev_get_queue(bdev);
1559 return q->limits.max_write_zeroes_sectors;
1564 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1566 struct request_queue *q = bdev_get_queue(bdev);
1569 return blk_queue_zoned_model(q);
1571 return BLK_ZONED_NONE;
1574 static inline bool bdev_is_zoned(struct block_device *bdev)
1576 struct request_queue *q = bdev_get_queue(bdev);
1579 return blk_queue_is_zoned(q);
1584 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1586 struct request_queue *q = bdev_get_queue(bdev);
1589 return blk_queue_zone_sectors(q);
1593 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1595 struct request_queue *q = bdev_get_queue(bdev);
1598 return queue_max_open_zones(q);
1602 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1604 struct request_queue *q = bdev_get_queue(bdev);
1607 return queue_max_active_zones(q);
1611 static inline int queue_dma_alignment(const struct request_queue *q)
1613 return q ? q->dma_alignment : 511;
1616 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1619 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1620 return !(addr & alignment) && !(len & alignment);
1623 /* assumes size > 256 */
1624 static inline unsigned int blksize_bits(unsigned int size)
1626 unsigned int bits = 8;
1630 } while (size > 256);
1634 static inline unsigned int block_size(struct block_device *bdev)
1636 return 1 << bdev->bd_inode->i_blkbits;
1639 int kblockd_schedule_work(struct work_struct *work);
1640 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1642 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1643 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1644 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1645 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1647 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1649 enum blk_integrity_flags {
1650 BLK_INTEGRITY_VERIFY = 1 << 0,
1651 BLK_INTEGRITY_GENERATE = 1 << 1,
1652 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1653 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1656 struct blk_integrity_iter {
1660 unsigned int data_size;
1661 unsigned short interval;
1662 const char *disk_name;
1665 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1666 typedef void (integrity_prepare_fn) (struct request *);
1667 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1669 struct blk_integrity_profile {
1670 integrity_processing_fn *generate_fn;
1671 integrity_processing_fn *verify_fn;
1672 integrity_prepare_fn *prepare_fn;
1673 integrity_complete_fn *complete_fn;
1677 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1678 extern void blk_integrity_unregister(struct gendisk *);
1679 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1680 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1681 struct scatterlist *);
1682 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1684 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1686 struct blk_integrity *bi = &disk->queue->integrity;
1695 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1697 return blk_get_integrity(bdev->bd_disk);
1701 blk_integrity_queue_supports_integrity(struct request_queue *q)
1703 return q->integrity.profile;
1706 static inline bool blk_integrity_rq(struct request *rq)
1708 return rq->cmd_flags & REQ_INTEGRITY;
1711 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1714 q->limits.max_integrity_segments = segs;
1717 static inline unsigned short
1718 queue_max_integrity_segments(const struct request_queue *q)
1720 return q->limits.max_integrity_segments;
1724 * bio_integrity_intervals - Return number of integrity intervals for a bio
1725 * @bi: blk_integrity profile for device
1726 * @sectors: Size of the bio in 512-byte sectors
1728 * Description: The block layer calculates everything in 512 byte
1729 * sectors but integrity metadata is done in terms of the data integrity
1730 * interval size of the storage device. Convert the block layer sectors
1731 * to the appropriate number of integrity intervals.
1733 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1734 unsigned int sectors)
1736 return sectors >> (bi->interval_exp - 9);
1739 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1740 unsigned int sectors)
1742 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1746 * Return the first bvec that contains integrity data. Only drivers that are
1747 * limited to a single integrity segment should use this helper.
1749 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1751 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1753 return rq->bio->bi_integrity->bip_vec;
1756 #else /* CONFIG_BLK_DEV_INTEGRITY */
1759 struct block_device;
1761 struct blk_integrity;
1763 static inline int blk_integrity_rq(struct request *rq)
1767 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1772 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1774 struct scatterlist *s)
1778 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1782 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1787 blk_integrity_queue_supports_integrity(struct request_queue *q)
1791 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1795 static inline void blk_integrity_register(struct gendisk *d,
1796 struct blk_integrity *b)
1799 static inline void blk_integrity_unregister(struct gendisk *d)
1802 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1806 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1811 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1812 unsigned int sectors)
1817 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1818 unsigned int sectors)
1823 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1828 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1830 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1832 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1834 void blk_ksm_unregister(struct request_queue *q);
1836 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1838 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1839 struct request_queue *q)
1844 static inline void blk_ksm_unregister(struct request_queue *q) { }
1846 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1849 struct block_device_operations {
1850 blk_qc_t (*submit_bio) (struct bio *bio);
1851 int (*open) (struct block_device *, fmode_t);
1852 void (*release) (struct gendisk *, fmode_t);
1853 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1854 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1855 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1856 unsigned int (*check_events) (struct gendisk *disk,
1857 unsigned int clearing);
1858 void (*unlock_native_capacity) (struct gendisk *);
1859 int (*getgeo)(struct block_device *, struct hd_geometry *);
1860 int (*set_read_only)(struct block_device *bdev, bool ro);
1861 /* this callback is with swap_lock and sometimes page table lock held */
1862 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1863 int (*report_zones)(struct gendisk *, sector_t sector,
1864 unsigned int nr_zones, report_zones_cb cb, void *data);
1865 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1866 struct module *owner;
1867 const struct pr_ops *pr_ops;
1870 #ifdef CONFIG_COMPAT
1871 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1872 unsigned int, unsigned long);
1874 #define blkdev_compat_ptr_ioctl NULL
1877 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1878 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1879 struct writeback_control *);
1881 #ifdef CONFIG_BLK_DEV_ZONED
1882 bool blk_req_needs_zone_write_lock(struct request *rq);
1883 bool blk_req_zone_write_trylock(struct request *rq);
1884 void __blk_req_zone_write_lock(struct request *rq);
1885 void __blk_req_zone_write_unlock(struct request *rq);
1887 static inline void blk_req_zone_write_lock(struct request *rq)
1889 if (blk_req_needs_zone_write_lock(rq))
1890 __blk_req_zone_write_lock(rq);
1893 static inline void blk_req_zone_write_unlock(struct request *rq)
1895 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1896 __blk_req_zone_write_unlock(rq);
1899 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1901 return rq->q->seq_zones_wlock &&
1902 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1905 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1907 if (!blk_req_needs_zone_write_lock(rq))
1909 return !blk_req_zone_is_write_locked(rq);
1912 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1917 static inline void blk_req_zone_write_lock(struct request *rq)
1921 static inline void blk_req_zone_write_unlock(struct request *rq)
1924 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1929 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1933 #endif /* CONFIG_BLK_DEV_ZONED */
1935 static inline void blk_wake_io_task(struct task_struct *waiter)
1938 * If we're polling, the task itself is doing the completions. For
1939 * that case, we don't need to signal a wakeup, it's enough to just
1940 * mark us as RUNNING.
1942 if (waiter == current)
1943 __set_current_state(TASK_RUNNING);
1945 wake_up_process(waiter);
1948 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1950 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1951 unsigned long start_time);
1953 unsigned long bio_start_io_acct(struct bio *bio);
1954 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1955 struct block_device *orig_bdev);
1958 * bio_end_io_acct - end I/O accounting for bio based drivers
1959 * @bio: bio to end account for
1960 * @start: start time returned by bio_start_io_acct()
1962 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1964 return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1967 int bdev_read_only(struct block_device *bdev);
1968 int set_blocksize(struct block_device *bdev, int size);
1970 const char *bdevname(struct block_device *bdev, char *buffer);
1971 int lookup_bdev(const char *pathname, dev_t *dev);
1973 void blkdev_show(struct seq_file *seqf, off_t offset);
1975 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1976 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1978 #define BLKDEV_MAJOR_MAX 512
1980 #define BLKDEV_MAJOR_MAX 0
1983 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1985 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1986 int bd_prepare_to_claim(struct block_device *bdev, void *holder);
1987 void bd_abort_claiming(struct block_device *bdev, void *holder);
1988 void blkdev_put(struct block_device *bdev, fmode_t mode);
1990 /* just for blk-cgroup, don't use elsewhere */
1991 struct block_device *blkdev_get_no_open(dev_t dev);
1992 void blkdev_put_no_open(struct block_device *bdev);
1994 struct block_device *bdev_alloc(struct gendisk *disk, u8 partno);
1995 void bdev_add(struct block_device *bdev, dev_t dev);
1996 struct block_device *I_BDEV(struct inode *inode);
1997 struct block_device *bdgrab(struct block_device *bdev);
1998 void bdput(struct block_device *);
1999 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
2003 void invalidate_bdev(struct block_device *bdev);
2004 int sync_blockdev(struct block_device *bdev);
2006 static inline void invalidate_bdev(struct block_device *bdev)
2009 static inline int sync_blockdev(struct block_device *bdev)
2014 int fsync_bdev(struct block_device *bdev);
2016 int freeze_bdev(struct block_device *bdev);
2017 int thaw_bdev(struct block_device *bdev);
2019 #endif /* _LINUX_BLKDEV_H */