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/timer.h>
12 #include <linux/workqueue.h>
13 #include <linux/pagemap.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>
29 struct scsi_ioctl_command;
32 struct elevator_queue;
38 struct blk_flush_queue;
41 struct blk_queue_stats;
42 struct blk_stat_callback;
43 struct blk_keyslot_manager;
45 #define BLKDEV_MIN_RQ 4
46 #define BLKDEV_MAX_RQ 128 /* Default maximum */
48 /* Must be consistent with blk_mq_poll_stats_bkt() */
49 #define BLK_MQ_POLL_STATS_BKTS 16
51 /* Doing classic polling */
52 #define BLK_MQ_POLL_CLASSIC -1
55 * Maximum number of blkcg policies allowed to be registered concurrently.
56 * Defined here to simplify include dependency.
58 #define BLKCG_MAX_POLS 5
60 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
64 typedef __u32 __bitwise req_flags_t;
66 /* elevator knows about this request */
67 #define RQF_SORTED ((__force req_flags_t)(1 << 0))
68 /* drive already may have started this one */
69 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
70 /* may not be passed by ioscheduler */
71 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
72 /* request for flush sequence */
73 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
74 /* merge of different types, fail separately */
75 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
76 /* track inflight for MQ */
77 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
78 /* don't call prep for this one */
79 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
80 /* set for "ide_preempt" requests and also for requests for which the SCSI
81 "quiesce" state must be ignored. */
82 #define RQF_PREEMPT ((__force req_flags_t)(1 << 8))
83 /* vaguely specified driver internal error. Ignored by the block layer */
84 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
85 /* don't warn about errors */
86 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
87 /* elevator private data attached */
88 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
89 /* account into disk and partition IO statistics */
90 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
91 /* request came from our alloc pool */
92 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
93 /* runtime pm request */
94 #define RQF_PM ((__force req_flags_t)(1 << 15))
95 /* on IO scheduler merge hash */
96 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
97 /* track IO completion time */
98 #define RQF_STATS ((__force req_flags_t)(1 << 17))
99 /* Look at ->special_vec for the actual data payload instead of the
101 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
102 /* The per-zone write lock is held for this request */
103 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
104 /* already slept for hybrid poll */
105 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
106 /* ->timeout has been called, don't expire again */
107 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
109 /* flags that prevent us from merging requests: */
110 #define RQF_NOMERGE_FLAGS \
111 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
114 * Request state for blk-mq.
123 * Try to put the fields that are referenced together in the same cacheline.
125 * If you modify this structure, make sure to update blk_rq_init() and
126 * especially blk_mq_rq_ctx_init() to take care of the added fields.
129 struct request_queue *q;
130 struct blk_mq_ctx *mq_ctx;
131 struct blk_mq_hw_ctx *mq_hctx;
133 unsigned int cmd_flags; /* op and common flags */
134 req_flags_t rq_flags;
139 /* the following two fields are internal, NEVER access directly */
140 unsigned int __data_len; /* total data len */
141 sector_t __sector; /* sector cursor */
146 struct list_head queuelist;
149 * The hash is used inside the scheduler, and killed once the
150 * request reaches the dispatch list. The ipi_list is only used
151 * to queue the request for softirq completion, which is long
152 * after the request has been unhashed (and even removed from
153 * the dispatch list).
156 struct hlist_node hash; /* merge hash */
157 struct list_head ipi_list;
161 * The rb_node is only used inside the io scheduler, requests
162 * are pruned when moved to the dispatch queue. So let the
163 * completion_data share space with the rb_node.
166 struct rb_node rb_node; /* sort/lookup */
167 struct bio_vec special_vec;
168 void *completion_data;
169 int error_count; /* for legacy drivers, don't use */
173 * Three pointers are available for the IO schedulers, if they need
174 * more they have to dynamically allocate it. Flush requests are
175 * never put on the IO scheduler. So let the flush fields share
176 * space with the elevator data.
186 struct list_head list;
187 rq_end_io_fn *saved_end_io;
191 struct gendisk *rq_disk;
192 struct hd_struct *part;
193 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
194 /* Time that the first bio started allocating this request. */
197 /* Time that this request was allocated for this IO. */
199 /* Time that I/O was submitted to the device. */
200 u64 io_start_time_ns;
202 #ifdef CONFIG_BLK_WBT
203 unsigned short wbt_flags;
206 * rq sectors used for blk stats. It has the same value
207 * with blk_rq_sectors(rq), except that it never be zeroed
210 unsigned short stats_sectors;
213 * Number of scatter-gather DMA addr+len pairs after
214 * physical address coalescing is performed.
216 unsigned short nr_phys_segments;
218 #if defined(CONFIG_BLK_DEV_INTEGRITY)
219 unsigned short nr_integrity_segments;
222 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
223 struct bio_crypt_ctx *crypt_ctx;
224 struct blk_ksm_keyslot *crypt_keyslot;
227 unsigned short write_hint;
228 unsigned short ioprio;
230 enum mq_rq_state state;
233 unsigned int timeout;
234 unsigned long deadline;
237 struct __call_single_data csd;
242 * completion callback.
244 rq_end_io_fn *end_io;
248 static inline bool blk_op_is_scsi(unsigned int op)
250 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
253 static inline bool blk_op_is_private(unsigned int op)
255 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
258 static inline bool blk_rq_is_scsi(struct request *rq)
260 return blk_op_is_scsi(req_op(rq));
263 static inline bool blk_rq_is_private(struct request *rq)
265 return blk_op_is_private(req_op(rq));
268 static inline bool blk_rq_is_passthrough(struct request *rq)
270 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
273 static inline bool bio_is_passthrough(struct bio *bio)
275 unsigned op = bio_op(bio);
277 return blk_op_is_scsi(op) || blk_op_is_private(op);
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 enum blk_zoned_model {
311 BLK_ZONED_NONE, /* Regular block device */
312 BLK_ZONED_HA, /* Host-aware zoned block device */
313 BLK_ZONED_HM, /* Host-managed zoned block device */
316 struct queue_limits {
317 unsigned long bounce_pfn;
318 unsigned long seg_boundary_mask;
319 unsigned long virt_boundary_mask;
321 unsigned int max_hw_sectors;
322 unsigned int max_dev_sectors;
323 unsigned int chunk_sectors;
324 unsigned int max_sectors;
325 unsigned int max_segment_size;
326 unsigned int physical_block_size;
327 unsigned int logical_block_size;
328 unsigned int alignment_offset;
331 unsigned int max_discard_sectors;
332 unsigned int max_hw_discard_sectors;
333 unsigned int max_write_same_sectors;
334 unsigned int max_write_zeroes_sectors;
335 unsigned int max_zone_append_sectors;
336 unsigned int discard_granularity;
337 unsigned int discard_alignment;
339 unsigned short max_segments;
340 unsigned short max_integrity_segments;
341 unsigned short max_discard_segments;
343 unsigned char misaligned;
344 unsigned char discard_misaligned;
345 unsigned char raid_partial_stripes_expensive;
346 enum blk_zoned_model zoned;
349 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
352 #ifdef CONFIG_BLK_DEV_ZONED
354 #define BLK_ALL_ZONES ((unsigned int)-1)
355 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
356 unsigned int nr_zones, report_zones_cb cb, void *data);
357 unsigned int blkdev_nr_zones(struct gendisk *disk);
358 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
359 sector_t sectors, sector_t nr_sectors,
361 int blk_revalidate_disk_zones(struct gendisk *disk,
362 void (*update_driver_data)(struct gendisk *disk));
364 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
365 unsigned int cmd, unsigned long arg);
366 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
367 unsigned int cmd, unsigned long arg);
369 #else /* CONFIG_BLK_DEV_ZONED */
371 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
376 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
377 fmode_t mode, unsigned int cmd,
383 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
384 fmode_t mode, unsigned int cmd,
390 #endif /* CONFIG_BLK_DEV_ZONED */
392 struct request_queue {
393 struct request *last_merge;
394 struct elevator_queue *elevator;
396 struct blk_queue_stats *stats;
397 struct rq_qos *rq_qos;
399 const struct blk_mq_ops *mq_ops;
402 struct blk_mq_ctx __percpu *queue_ctx;
404 unsigned int queue_depth;
406 /* hw dispatch queues */
407 struct blk_mq_hw_ctx **queue_hw_ctx;
408 unsigned int nr_hw_queues;
410 struct backing_dev_info *backing_dev_info;
413 * The queue owner gets to use this for whatever they like.
414 * ll_rw_blk doesn't touch it.
419 * various queue flags, see QUEUE_* below
421 unsigned long queue_flags;
423 * Number of contexts that have called blk_set_pm_only(). If this
424 * counter is above zero then only RQF_PM and RQF_PREEMPT requests are
430 * ida allocated id for this queue. Used to index queues from
436 * queue needs bounce pages for pages above this limit
440 spinlock_t queue_lock;
450 struct kobject *mq_kobj;
452 #ifdef CONFIG_BLK_DEV_INTEGRITY
453 struct blk_integrity integrity;
454 #endif /* CONFIG_BLK_DEV_INTEGRITY */
459 unsigned int nr_pending;
465 unsigned long nr_requests; /* Max # of requests */
467 unsigned int dma_pad_mask;
468 unsigned int dma_alignment;
470 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
471 /* Inline crypto capabilities */
472 struct blk_keyslot_manager *ksm;
475 unsigned int rq_timeout;
478 struct blk_stat_callback *poll_cb;
479 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
481 struct timer_list timeout;
482 struct work_struct timeout_work;
484 struct list_head icq_list;
485 #ifdef CONFIG_BLK_CGROUP
486 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
487 struct blkcg_gq *root_blkg;
488 struct list_head blkg_list;
491 struct queue_limits limits;
493 unsigned int required_elevator_features;
495 #ifdef CONFIG_BLK_DEV_ZONED
497 * Zoned block device information for request dispatch control.
498 * nr_zones is the total number of zones of the device. This is always
499 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
500 * bits which indicates if a zone is conventional (bit set) or
501 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
502 * bits which indicates if a zone is write locked, that is, if a write
503 * request targeting the zone was dispatched. All three fields are
504 * initialized by the low level device driver (e.g. scsi/sd.c).
505 * Stacking drivers (device mappers) may or may not initialize
508 * Reads of this information must be protected with blk_queue_enter() /
509 * blk_queue_exit(). Modifying this information is only allowed while
510 * no requests are being processed. See also blk_mq_freeze_queue() and
511 * blk_mq_unfreeze_queue().
513 unsigned int nr_zones;
514 unsigned long *conv_zones_bitmap;
515 unsigned long *seq_zones_wlock;
516 #endif /* CONFIG_BLK_DEV_ZONED */
521 unsigned int sg_timeout;
522 unsigned int sg_reserved_size;
524 struct mutex debugfs_mutex;
525 #ifdef CONFIG_BLK_DEV_IO_TRACE
526 struct blk_trace __rcu *blk_trace;
529 * for flush operations
531 struct blk_flush_queue *fq;
533 struct list_head requeue_list;
534 spinlock_t requeue_lock;
535 struct delayed_work requeue_work;
537 struct mutex sysfs_lock;
538 struct mutex sysfs_dir_lock;
541 * for reusing dead hctx instance in case of updating
544 struct list_head unused_hctx_list;
545 spinlock_t unused_hctx_lock;
549 #if defined(CONFIG_BLK_DEV_BSG)
550 struct bsg_class_device bsg_dev;
553 #ifdef CONFIG_BLK_DEV_THROTTLING
555 struct throtl_data *td;
557 struct rcu_head rcu_head;
558 wait_queue_head_t mq_freeze_wq;
560 * Protect concurrent access to q_usage_counter by
561 * percpu_ref_kill() and percpu_ref_reinit().
563 struct mutex mq_freeze_lock;
564 struct percpu_ref q_usage_counter;
566 struct blk_mq_tag_set *tag_set;
567 struct list_head tag_set_list;
568 struct bio_set bio_split;
570 struct dentry *debugfs_dir;
572 #ifdef CONFIG_BLK_DEBUG_FS
573 struct dentry *sched_debugfs_dir;
574 struct dentry *rqos_debugfs_dir;
577 bool mq_sysfs_init_done;
581 #define BLK_MAX_WRITE_HINTS 5
582 u64 write_hints[BLK_MAX_WRITE_HINTS];
585 /* Keep blk_queue_flag_name[] in sync with the definitions below */
586 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
587 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
588 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
589 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
590 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
591 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
592 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
593 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
594 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
595 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
596 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
597 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
598 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
599 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
600 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
601 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
602 #define QUEUE_FLAG_WC 17 /* Write back caching */
603 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
604 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
605 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
606 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
607 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
608 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
609 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
610 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
611 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
612 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
614 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
615 (1 << QUEUE_FLAG_SAME_COMP))
617 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
618 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
619 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
621 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
622 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
623 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
624 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
625 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
626 #define blk_queue_noxmerges(q) \
627 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
628 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
629 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
630 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
631 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
632 #define blk_queue_zone_resetall(q) \
633 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
634 #define blk_queue_secure_erase(q) \
635 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
636 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
637 #define blk_queue_scsi_passthrough(q) \
638 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
639 #define blk_queue_pci_p2pdma(q) \
640 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
641 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
642 #define blk_queue_rq_alloc_time(q) \
643 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
645 #define blk_queue_rq_alloc_time(q) false
648 #define blk_noretry_request(rq) \
649 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
650 REQ_FAILFAST_DRIVER))
651 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
652 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
653 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
654 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
656 extern void blk_set_pm_only(struct request_queue *q);
657 extern void blk_clear_pm_only(struct request_queue *q);
659 static inline bool blk_account_rq(struct request *rq)
661 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
664 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
666 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
668 #define rq_dma_dir(rq) \
669 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
671 #define dma_map_bvec(dev, bv, dir, attrs) \
672 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
675 static inline bool queue_is_mq(struct request_queue *q)
680 static inline enum blk_zoned_model
681 blk_queue_zoned_model(struct request_queue *q)
683 return q->limits.zoned;
686 static inline bool blk_queue_is_zoned(struct request_queue *q)
688 switch (blk_queue_zoned_model(q)) {
697 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
699 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
702 #ifdef CONFIG_BLK_DEV_ZONED
703 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
705 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
708 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
711 if (!blk_queue_is_zoned(q))
713 return sector >> ilog2(q->limits.chunk_sectors);
716 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
719 if (!blk_queue_is_zoned(q))
721 if (!q->conv_zones_bitmap)
723 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
725 #else /* CONFIG_BLK_DEV_ZONED */
726 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
730 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
735 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
740 #endif /* CONFIG_BLK_DEV_ZONED */
742 static inline bool rq_is_sync(struct request *rq)
744 return op_is_sync(rq->cmd_flags);
747 static inline bool rq_mergeable(struct request *rq)
749 if (blk_rq_is_passthrough(rq))
752 if (req_op(rq) == REQ_OP_FLUSH)
755 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
758 if (req_op(rq) == REQ_OP_ZONE_APPEND)
761 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
763 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
769 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
771 if (bio_page(a) == bio_page(b) &&
772 bio_offset(a) == bio_offset(b))
778 static inline unsigned int blk_queue_depth(struct request_queue *q)
781 return q->queue_depth;
783 return q->nr_requests;
786 extern unsigned long blk_max_low_pfn, blk_max_pfn;
789 * standard bounce addresses:
791 * BLK_BOUNCE_HIGH : bounce all highmem pages
792 * BLK_BOUNCE_ANY : don't bounce anything
793 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
796 #if BITS_PER_LONG == 32
797 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
799 #define BLK_BOUNCE_HIGH -1ULL
801 #define BLK_BOUNCE_ANY (-1ULL)
802 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
805 * default timeout for SG_IO if none specified
807 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
808 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
814 unsigned long offset;
819 struct req_iterator {
820 struct bvec_iter iter;
824 /* This should not be used directly - use rq_for_each_segment */
825 #define for_each_bio(_bio) \
826 for (; _bio; _bio = _bio->bi_next)
827 #define __rq_for_each_bio(_bio, rq) \
829 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
831 #define rq_for_each_segment(bvl, _rq, _iter) \
832 __rq_for_each_bio(_iter.bio, _rq) \
833 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
835 #define rq_for_each_bvec(bvl, _rq, _iter) \
836 __rq_for_each_bio(_iter.bio, _rq) \
837 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
839 #define rq_iter_last(bvec, _iter) \
840 (_iter.bio->bi_next == NULL && \
841 bio_iter_last(bvec, _iter.iter))
843 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
844 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
846 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
847 extern void rq_flush_dcache_pages(struct request *rq);
849 static inline void rq_flush_dcache_pages(struct request *rq)
854 extern int blk_register_queue(struct gendisk *disk);
855 extern void blk_unregister_queue(struct gendisk *disk);
856 blk_qc_t submit_bio_noacct(struct bio *bio);
857 extern void blk_rq_init(struct request_queue *q, struct request *rq);
858 extern void blk_put_request(struct request *);
859 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
860 blk_mq_req_flags_t flags);
861 extern int blk_lld_busy(struct request_queue *q);
862 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
863 struct bio_set *bs, gfp_t gfp_mask,
864 int (*bio_ctr)(struct bio *, struct bio *, void *),
866 extern void blk_rq_unprep_clone(struct request *rq);
867 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
869 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
870 extern void blk_queue_split(struct bio **);
871 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
872 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
873 unsigned int, void __user *);
874 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
875 unsigned int, void __user *);
876 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
877 struct scsi_ioctl_command __user *);
878 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
879 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
881 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
882 extern void blk_queue_exit(struct request_queue *q);
883 extern void blk_sync_queue(struct request_queue *q);
884 extern int blk_rq_map_user(struct request_queue *, struct request *,
885 struct rq_map_data *, void __user *, unsigned long,
887 extern int blk_rq_unmap_user(struct bio *);
888 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
889 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
890 struct rq_map_data *, const struct iov_iter *,
892 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
893 struct request *, int);
894 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
895 struct request *, int, rq_end_io_fn *);
897 /* Helper to convert REQ_OP_XXX to its string format XXX */
898 extern const char *blk_op_str(unsigned int op);
900 int blk_status_to_errno(blk_status_t status);
901 blk_status_t errno_to_blk_status(int errno);
903 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
905 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
907 return bdev->bd_disk->queue; /* this is never NULL */
911 * The basic unit of block I/O is a sector. It is used in a number of contexts
912 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
913 * bytes. Variables of type sector_t represent an offset or size that is a
914 * multiple of 512 bytes. Hence these two constants.
917 #define SECTOR_SHIFT 9
920 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
924 * blk_rq_pos() : the current sector
925 * blk_rq_bytes() : bytes left in the entire request
926 * blk_rq_cur_bytes() : bytes left in the current segment
927 * blk_rq_err_bytes() : bytes left till the next error boundary
928 * blk_rq_sectors() : sectors left in the entire request
929 * blk_rq_cur_sectors() : sectors left in the current segment
930 * blk_rq_stats_sectors() : sectors of the entire request used for stats
932 static inline sector_t blk_rq_pos(const struct request *rq)
937 static inline unsigned int blk_rq_bytes(const struct request *rq)
939 return rq->__data_len;
942 static inline int blk_rq_cur_bytes(const struct request *rq)
944 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
947 extern unsigned int blk_rq_err_bytes(const struct request *rq);
949 static inline unsigned int blk_rq_sectors(const struct request *rq)
951 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
954 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
956 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
959 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
961 return rq->stats_sectors;
964 #ifdef CONFIG_BLK_DEV_ZONED
966 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
967 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
969 static inline unsigned int blk_rq_zone_no(struct request *rq)
971 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
974 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
976 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
978 #endif /* CONFIG_BLK_DEV_ZONED */
981 * Some commands like WRITE SAME have a payload or data transfer size which
982 * is different from the size of the request. Any driver that supports such
983 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
984 * calculate the data transfer size.
986 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
988 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
989 return rq->special_vec.bv_len;
990 return blk_rq_bytes(rq);
994 * Return the first full biovec in the request. The caller needs to check that
995 * there are any bvecs before calling this helper.
997 static inline struct bio_vec req_bvec(struct request *rq)
999 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1000 return rq->special_vec;
1001 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1004 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1007 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1008 return min(q->limits.max_discard_sectors,
1009 UINT_MAX >> SECTOR_SHIFT);
1011 if (unlikely(op == REQ_OP_WRITE_SAME))
1012 return q->limits.max_write_same_sectors;
1014 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1015 return q->limits.max_write_zeroes_sectors;
1017 return q->limits.max_sectors;
1021 * Return maximum size of a request at given offset. Only valid for
1022 * file system requests.
1024 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1027 if (!q->limits.chunk_sectors)
1028 return q->limits.max_sectors;
1030 return min(q->limits.max_sectors, (unsigned int)(q->limits.chunk_sectors -
1031 (offset & (q->limits.chunk_sectors - 1))));
1034 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1037 struct request_queue *q = rq->q;
1039 if (blk_rq_is_passthrough(rq))
1040 return q->limits.max_hw_sectors;
1042 if (!q->limits.chunk_sectors ||
1043 req_op(rq) == REQ_OP_DISCARD ||
1044 req_op(rq) == REQ_OP_SECURE_ERASE)
1045 return blk_queue_get_max_sectors(q, req_op(rq));
1047 return min(blk_max_size_offset(q, offset),
1048 blk_queue_get_max_sectors(q, req_op(rq)));
1051 static inline unsigned int blk_rq_count_bios(struct request *rq)
1053 unsigned int nr_bios = 0;
1056 __rq_for_each_bio(bio, rq)
1062 void blk_steal_bios(struct bio_list *list, struct request *rq);
1065 * Request completion related functions.
1067 * blk_update_request() completes given number of bytes and updates
1068 * the request without completing it.
1070 extern bool blk_update_request(struct request *rq, blk_status_t error,
1071 unsigned int nr_bytes);
1073 extern void blk_abort_request(struct request *);
1076 * Access functions for manipulating queue properties
1078 extern void blk_cleanup_queue(struct request_queue *);
1079 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1080 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1081 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1082 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1083 extern void blk_queue_max_discard_segments(struct request_queue *,
1085 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1086 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1087 unsigned int max_discard_sectors);
1088 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1089 unsigned int max_write_same_sectors);
1090 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1091 unsigned int max_write_same_sectors);
1092 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1093 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1094 unsigned int max_zone_append_sectors);
1095 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1096 extern void blk_queue_alignment_offset(struct request_queue *q,
1097 unsigned int alignment);
1098 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1099 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1100 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1101 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1102 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1103 extern void blk_set_default_limits(struct queue_limits *lim);
1104 extern void blk_set_stacking_limits(struct queue_limits *lim);
1105 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1107 extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev,
1109 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1111 extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b);
1112 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1113 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1114 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1115 extern void blk_queue_dma_alignment(struct request_queue *, int);
1116 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1117 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1118 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1119 extern void blk_queue_required_elevator_features(struct request_queue *q,
1120 unsigned int features);
1121 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1122 struct device *dev);
1125 * Number of physical segments as sent to the device.
1127 * Normally this is the number of discontiguous data segments sent by the
1128 * submitter. But for data-less command like discard we might have no
1129 * actual data segments submitted, but the driver might have to add it's
1130 * own special payload. In that case we still return 1 here so that this
1131 * special payload will be mapped.
1133 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1135 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1137 return rq->nr_phys_segments;
1141 * Number of discard segments (or ranges) the driver needs to fill in.
1142 * Each discard bio merged into a request is counted as one segment.
1144 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1146 return max_t(unsigned short, rq->nr_phys_segments, 1);
1149 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1150 struct scatterlist *sglist, struct scatterlist **last_sg);
1151 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1152 struct scatterlist *sglist)
1154 struct scatterlist *last_sg = NULL;
1156 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1158 extern void blk_dump_rq_flags(struct request *, char *);
1160 bool __must_check blk_get_queue(struct request_queue *);
1161 struct request_queue *blk_alloc_queue(int node_id);
1162 extern void blk_put_queue(struct request_queue *);
1163 extern void blk_set_queue_dying(struct request_queue *);
1167 * blk_plug permits building a queue of related requests by holding the I/O
1168 * fragments for a short period. This allows merging of sequential requests
1169 * into single larger request. As the requests are moved from a per-task list to
1170 * the device's request_queue in a batch, this results in improved scalability
1171 * as the lock contention for request_queue lock is reduced.
1173 * It is ok not to disable preemption when adding the request to the plug list
1174 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1175 * the plug list when the task sleeps by itself. For details, please see
1176 * schedule() where blk_schedule_flush_plug() is called.
1179 struct list_head mq_list; /* blk-mq requests */
1180 struct list_head cb_list; /* md requires an unplug callback */
1181 unsigned short rq_count;
1182 bool multiple_queues;
1185 #define BLK_MAX_REQUEST_COUNT 16
1186 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1189 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1190 struct blk_plug_cb {
1191 struct list_head list;
1192 blk_plug_cb_fn callback;
1195 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1196 void *data, int size);
1197 extern void blk_start_plug(struct blk_plug *);
1198 extern void blk_finish_plug(struct blk_plug *);
1199 extern void blk_flush_plug_list(struct blk_plug *, bool);
1201 static inline void blk_flush_plug(struct task_struct *tsk)
1203 struct blk_plug *plug = tsk->plug;
1206 blk_flush_plug_list(plug, false);
1209 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1211 struct blk_plug *plug = tsk->plug;
1214 blk_flush_plug_list(plug, true);
1217 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1219 struct blk_plug *plug = tsk->plug;
1222 (!list_empty(&plug->mq_list) ||
1223 !list_empty(&plug->cb_list));
1226 int blkdev_issue_flush(struct block_device *, gfp_t);
1227 long nr_blockdev_pages(void);
1228 #else /* CONFIG_BLOCK */
1232 static inline void blk_start_plug(struct blk_plug *plug)
1236 static inline void blk_finish_plug(struct blk_plug *plug)
1240 static inline void blk_flush_plug(struct task_struct *task)
1244 static inline void blk_schedule_flush_plug(struct task_struct *task)
1249 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1254 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask)
1259 static inline long nr_blockdev_pages(void)
1263 #endif /* CONFIG_BLOCK */
1265 extern void blk_io_schedule(void);
1267 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1268 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1270 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1272 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1273 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1274 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1275 sector_t nr_sects, gfp_t gfp_mask, int flags,
1278 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1279 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1281 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1282 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1284 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1285 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1287 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1288 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1290 return blkdev_issue_discard(sb->s_bdev,
1291 block << (sb->s_blocksize_bits -
1293 nr_blocks << (sb->s_blocksize_bits -
1297 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1298 sector_t nr_blocks, gfp_t gfp_mask)
1300 return blkdev_issue_zeroout(sb->s_bdev,
1301 block << (sb->s_blocksize_bits -
1303 nr_blocks << (sb->s_blocksize_bits -
1308 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1310 enum blk_default_limits {
1311 BLK_MAX_SEGMENTS = 128,
1312 BLK_SAFE_MAX_SECTORS = 255,
1313 BLK_DEF_MAX_SECTORS = 2560,
1314 BLK_MAX_SEGMENT_SIZE = 65536,
1315 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1318 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1320 return q->limits.seg_boundary_mask;
1323 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1325 return q->limits.virt_boundary_mask;
1328 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1330 return q->limits.max_sectors;
1333 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1335 return q->limits.max_hw_sectors;
1338 static inline unsigned short queue_max_segments(const struct request_queue *q)
1340 return q->limits.max_segments;
1343 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1345 return q->limits.max_discard_segments;
1348 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1350 return q->limits.max_segment_size;
1353 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1355 return q->limits.max_zone_append_sectors;
1358 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1362 if (q && q->limits.logical_block_size)
1363 retval = q->limits.logical_block_size;
1368 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1370 return queue_logical_block_size(bdev_get_queue(bdev));
1373 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1375 return q->limits.physical_block_size;
1378 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1380 return queue_physical_block_size(bdev_get_queue(bdev));
1383 static inline unsigned int queue_io_min(const struct request_queue *q)
1385 return q->limits.io_min;
1388 static inline int bdev_io_min(struct block_device *bdev)
1390 return queue_io_min(bdev_get_queue(bdev));
1393 static inline unsigned int queue_io_opt(const struct request_queue *q)
1395 return q->limits.io_opt;
1398 static inline int bdev_io_opt(struct block_device *bdev)
1400 return queue_io_opt(bdev_get_queue(bdev));
1403 static inline int queue_alignment_offset(const struct request_queue *q)
1405 if (q->limits.misaligned)
1408 return q->limits.alignment_offset;
1411 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1413 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1414 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1417 return (granularity + lim->alignment_offset - alignment) % granularity;
1420 static inline int bdev_alignment_offset(struct block_device *bdev)
1422 struct request_queue *q = bdev_get_queue(bdev);
1424 if (q->limits.misaligned)
1427 if (bdev != bdev->bd_contains)
1428 return bdev->bd_part->alignment_offset;
1430 return q->limits.alignment_offset;
1433 static inline int queue_discard_alignment(const struct request_queue *q)
1435 if (q->limits.discard_misaligned)
1438 return q->limits.discard_alignment;
1441 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1443 unsigned int alignment, granularity, offset;
1445 if (!lim->max_discard_sectors)
1448 /* Why are these in bytes, not sectors? */
1449 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1450 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1454 /* Offset of the partition start in 'granularity' sectors */
1455 offset = sector_div(sector, granularity);
1457 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1458 offset = (granularity + alignment - offset) % granularity;
1460 /* Turn it back into bytes, gaah */
1461 return offset << SECTOR_SHIFT;
1464 static inline int bdev_discard_alignment(struct block_device *bdev)
1466 struct request_queue *q = bdev_get_queue(bdev);
1468 if (bdev != bdev->bd_contains)
1469 return bdev->bd_part->discard_alignment;
1471 return q->limits.discard_alignment;
1474 static inline unsigned int bdev_write_same(struct block_device *bdev)
1476 struct request_queue *q = bdev_get_queue(bdev);
1479 return q->limits.max_write_same_sectors;
1484 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1486 struct request_queue *q = bdev_get_queue(bdev);
1489 return q->limits.max_write_zeroes_sectors;
1494 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1496 struct request_queue *q = bdev_get_queue(bdev);
1499 return blk_queue_zoned_model(q);
1501 return BLK_ZONED_NONE;
1504 static inline bool bdev_is_zoned(struct block_device *bdev)
1506 struct request_queue *q = bdev_get_queue(bdev);
1509 return blk_queue_is_zoned(q);
1514 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1516 struct request_queue *q = bdev_get_queue(bdev);
1519 return blk_queue_zone_sectors(q);
1523 static inline int queue_dma_alignment(const struct request_queue *q)
1525 return q ? q->dma_alignment : 511;
1528 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1531 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1532 return !(addr & alignment) && !(len & alignment);
1535 /* assumes size > 256 */
1536 static inline unsigned int blksize_bits(unsigned int size)
1538 unsigned int bits = 8;
1542 } while (size > 256);
1546 static inline unsigned int block_size(struct block_device *bdev)
1548 return 1 << bdev->bd_inode->i_blkbits;
1551 int kblockd_schedule_work(struct work_struct *work);
1552 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1554 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1555 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1556 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1557 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1559 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1561 enum blk_integrity_flags {
1562 BLK_INTEGRITY_VERIFY = 1 << 0,
1563 BLK_INTEGRITY_GENERATE = 1 << 1,
1564 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1565 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1568 struct blk_integrity_iter {
1572 unsigned int data_size;
1573 unsigned short interval;
1574 const char *disk_name;
1577 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1578 typedef void (integrity_prepare_fn) (struct request *);
1579 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1581 struct blk_integrity_profile {
1582 integrity_processing_fn *generate_fn;
1583 integrity_processing_fn *verify_fn;
1584 integrity_prepare_fn *prepare_fn;
1585 integrity_complete_fn *complete_fn;
1589 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1590 extern void blk_integrity_unregister(struct gendisk *);
1591 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1592 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1593 struct scatterlist *);
1594 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1595 extern bool blk_integrity_merge_rq(struct request_queue *, struct request *,
1597 extern bool blk_integrity_merge_bio(struct request_queue *, struct request *,
1600 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1602 struct blk_integrity *bi = &disk->queue->integrity;
1611 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1613 return blk_get_integrity(bdev->bd_disk);
1617 blk_integrity_queue_supports_integrity(struct request_queue *q)
1619 return q->integrity.profile;
1622 static inline bool blk_integrity_rq(struct request *rq)
1624 return rq->cmd_flags & REQ_INTEGRITY;
1627 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1630 q->limits.max_integrity_segments = segs;
1633 static inline unsigned short
1634 queue_max_integrity_segments(const struct request_queue *q)
1636 return q->limits.max_integrity_segments;
1640 * bio_integrity_intervals - Return number of integrity intervals for a bio
1641 * @bi: blk_integrity profile for device
1642 * @sectors: Size of the bio in 512-byte sectors
1644 * Description: The block layer calculates everything in 512 byte
1645 * sectors but integrity metadata is done in terms of the data integrity
1646 * interval size of the storage device. Convert the block layer sectors
1647 * to the appropriate number of integrity intervals.
1649 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1650 unsigned int sectors)
1652 return sectors >> (bi->interval_exp - 9);
1655 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1656 unsigned int sectors)
1658 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1662 * Return the first bvec that contains integrity data. Only drivers that are
1663 * limited to a single integrity segment should use this helper.
1665 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1667 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1669 return rq->bio->bi_integrity->bip_vec;
1672 #else /* CONFIG_BLK_DEV_INTEGRITY */
1675 struct block_device;
1677 struct blk_integrity;
1679 static inline int blk_integrity_rq(struct request *rq)
1683 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1688 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1690 struct scatterlist *s)
1694 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1698 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1703 blk_integrity_queue_supports_integrity(struct request_queue *q)
1707 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1711 static inline void blk_integrity_register(struct gendisk *d,
1712 struct blk_integrity *b)
1715 static inline void blk_integrity_unregister(struct gendisk *d)
1718 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1722 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1726 static inline bool blk_integrity_merge_rq(struct request_queue *rq,
1732 static inline bool blk_integrity_merge_bio(struct request_queue *rq,
1739 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1740 unsigned int sectors)
1745 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1746 unsigned int sectors)
1751 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1756 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1758 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1760 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1762 void blk_ksm_unregister(struct request_queue *q);
1764 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1766 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1767 struct request_queue *q)
1772 static inline void blk_ksm_unregister(struct request_queue *q) { }
1774 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1777 struct block_device_operations {
1778 blk_qc_t (*submit_bio) (struct bio *bio);
1779 int (*open) (struct block_device *, fmode_t);
1780 void (*release) (struct gendisk *, fmode_t);
1781 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1782 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1783 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1784 unsigned int (*check_events) (struct gendisk *disk,
1785 unsigned int clearing);
1786 void (*unlock_native_capacity) (struct gendisk *);
1787 int (*revalidate_disk) (struct gendisk *);
1788 int (*getgeo)(struct block_device *, struct hd_geometry *);
1789 /* this callback is with swap_lock and sometimes page table lock held */
1790 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1791 int (*report_zones)(struct gendisk *, sector_t sector,
1792 unsigned int nr_zones, report_zones_cb cb, void *data);
1793 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1794 struct module *owner;
1795 const struct pr_ops *pr_ops;
1798 #ifdef CONFIG_COMPAT
1799 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1800 unsigned int, unsigned long);
1802 #define blkdev_compat_ptr_ioctl NULL
1805 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1807 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1808 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1809 struct writeback_control *);
1811 #ifdef CONFIG_BLK_DEV_ZONED
1812 bool blk_req_needs_zone_write_lock(struct request *rq);
1813 bool blk_req_zone_write_trylock(struct request *rq);
1814 void __blk_req_zone_write_lock(struct request *rq);
1815 void __blk_req_zone_write_unlock(struct request *rq);
1817 static inline void blk_req_zone_write_lock(struct request *rq)
1819 if (blk_req_needs_zone_write_lock(rq))
1820 __blk_req_zone_write_lock(rq);
1823 static inline void blk_req_zone_write_unlock(struct request *rq)
1825 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1826 __blk_req_zone_write_unlock(rq);
1829 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1831 return rq->q->seq_zones_wlock &&
1832 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1835 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1837 if (!blk_req_needs_zone_write_lock(rq))
1839 return !blk_req_zone_is_write_locked(rq);
1842 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1847 static inline void blk_req_zone_write_lock(struct request *rq)
1851 static inline void blk_req_zone_write_unlock(struct request *rq)
1854 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1859 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1863 #endif /* CONFIG_BLK_DEV_ZONED */
1865 static inline void blk_wake_io_task(struct task_struct *waiter)
1868 * If we're polling, the task itself is doing the completions. For
1869 * that case, we don't need to signal a wakeup, it's enough to just
1870 * mark us as RUNNING.
1872 if (waiter == current)
1873 __set_current_state(TASK_RUNNING);
1875 wake_up_process(waiter);
1878 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1880 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1881 unsigned long start_time);
1884 * bio_start_io_acct - start I/O accounting for bio based drivers
1885 * @bio: bio to start account for
1887 * Returns the start time that should be passed back to bio_end_io_acct().
1889 static inline unsigned long bio_start_io_acct(struct bio *bio)
1891 return disk_start_io_acct(bio->bi_disk, bio_sectors(bio), bio_op(bio));
1895 * bio_end_io_acct - end I/O accounting for bio based drivers
1896 * @bio: bio to end account for
1897 * @start: start time returned by bio_start_io_acct()
1899 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1901 return disk_end_io_acct(bio->bi_disk, bio_op(bio), start_time);
1904 int bdev_read_only(struct block_device *bdev);
1905 int set_blocksize(struct block_device *bdev, int size);
1907 const char *bdevname(struct block_device *bdev, char *buffer);
1908 struct block_device *lookup_bdev(const char *);
1910 void blkdev_show(struct seq_file *seqf, off_t offset);
1912 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1913 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1915 #define BLKDEV_MAJOR_MAX 512
1917 #define BLKDEV_MAJOR_MAX 0
1920 int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
1921 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1923 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1924 int bd_prepare_to_claim(struct block_device *bdev, struct block_device *whole,
1926 void bd_abort_claiming(struct block_device *bdev, struct block_device *whole,
1928 void blkdev_put(struct block_device *bdev, fmode_t mode);
1930 struct block_device *I_BDEV(struct inode *inode);
1931 struct block_device *bdget(dev_t);
1932 struct block_device *bdgrab(struct block_device *bdev);
1933 void bdput(struct block_device *);
1936 void invalidate_bdev(struct block_device *bdev);
1937 int sync_blockdev(struct block_device *bdev);
1939 static inline void invalidate_bdev(struct block_device *bdev)
1942 static inline int sync_blockdev(struct block_device *bdev)
1947 int fsync_bdev(struct block_device *bdev);
1949 struct super_block *freeze_bdev(struct block_device *bdev);
1950 int thaw_bdev(struct block_device *bdev, struct super_block *sb);
1952 #endif /* _LINUX_BLKDEV_H */