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>
10 #include <linux/major.h>
11 #include <linux/genhd.h>
12 #include <linux/list.h>
13 #include <linux/llist.h>
14 #include <linux/timer.h>
15 #include <linux/workqueue.h>
16 #include <linux/pagemap.h>
17 #include <linux/backing-dev-defs.h>
18 #include <linux/wait.h>
19 #include <linux/mempool.h>
20 #include <linux/pfn.h>
21 #include <linux/bio.h>
22 #include <linux/stringify.h>
23 #include <linux/gfp.h>
24 #include <linux/bsg.h>
25 #include <linux/smp.h>
26 #include <linux/rcupdate.h>
27 #include <linux/percpu-refcount.h>
28 #include <linux/scatterlist.h>
29 #include <linux/blkzoned.h>
32 struct scsi_ioctl_command;
35 struct elevator_queue;
41 struct blk_flush_queue;
44 struct blk_queue_stats;
45 struct blk_stat_callback;
46 struct blk_keyslot_manager;
48 #define BLKDEV_MIN_RQ 4
49 #define BLKDEV_MAX_RQ 128 /* Default maximum */
51 /* Must be consistent with blk_mq_poll_stats_bkt() */
52 #define BLK_MQ_POLL_STATS_BKTS 16
54 /* Doing classic polling */
55 #define BLK_MQ_POLL_CLASSIC -1
58 * Maximum number of blkcg policies allowed to be registered concurrently.
59 * Defined here to simplify include dependency.
61 #define BLKCG_MAX_POLS 5
63 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
67 typedef __u32 __bitwise req_flags_t;
69 /* elevator knows about this request */
70 #define RQF_SORTED ((__force req_flags_t)(1 << 0))
71 /* drive already may have started this one */
72 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
73 /* may not be passed by ioscheduler */
74 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
75 /* request for flush sequence */
76 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
77 /* merge of different types, fail separately */
78 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
79 /* track inflight for MQ */
80 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
81 /* don't call prep for this one */
82 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
83 /* set for "ide_preempt" requests and also for requests for which the SCSI
84 "quiesce" state must be ignored. */
85 #define RQF_PREEMPT ((__force req_flags_t)(1 << 8))
86 /* vaguely specified driver internal error. Ignored by the block layer */
87 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
88 /* don't warn about errors */
89 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
90 /* elevator private data attached */
91 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
92 /* account into disk and partition IO statistics */
93 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
94 /* request came from our alloc pool */
95 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
96 /* runtime pm request */
97 #define RQF_PM ((__force req_flags_t)(1 << 15))
98 /* on IO scheduler merge hash */
99 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
100 /* track IO completion time */
101 #define RQF_STATS ((__force req_flags_t)(1 << 17))
102 /* Look at ->special_vec for the actual data payload instead of the
104 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
105 /* The per-zone write lock is held for this request */
106 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
107 /* already slept for hybrid poll */
108 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
109 /* ->timeout has been called, don't expire again */
110 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
112 /* flags that prevent us from merging requests: */
113 #define RQF_NOMERGE_FLAGS \
114 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
117 * Request state for blk-mq.
126 * Try to put the fields that are referenced together in the same cacheline.
128 * If you modify this structure, make sure to update blk_rq_init() and
129 * especially blk_mq_rq_ctx_init() to take care of the added fields.
132 struct request_queue *q;
133 struct blk_mq_ctx *mq_ctx;
134 struct blk_mq_hw_ctx *mq_hctx;
136 unsigned int cmd_flags; /* op and common flags */
137 req_flags_t rq_flags;
142 /* the following two fields are internal, NEVER access directly */
143 unsigned int __data_len; /* total data len */
144 sector_t __sector; /* sector cursor */
149 struct list_head queuelist;
152 * The hash is used inside the scheduler, and killed once the
153 * request reaches the dispatch list. The ipi_list is only used
154 * to queue the request for softirq completion, which is long
155 * after the request has been unhashed (and even removed from
156 * the dispatch list).
159 struct hlist_node hash; /* merge hash */
160 struct list_head ipi_list;
164 * The rb_node is only used inside the io scheduler, requests
165 * are pruned when moved to the dispatch queue. So let the
166 * completion_data share space with the rb_node.
169 struct rb_node rb_node; /* sort/lookup */
170 struct bio_vec special_vec;
171 void *completion_data;
172 int error_count; /* for legacy drivers, don't use */
176 * Three pointers are available for the IO schedulers, if they need
177 * more they have to dynamically allocate it. Flush requests are
178 * never put on the IO scheduler. So let the flush fields share
179 * space with the elevator data.
189 struct list_head list;
190 rq_end_io_fn *saved_end_io;
194 struct gendisk *rq_disk;
195 struct hd_struct *part;
196 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
197 /* Time that the first bio started allocating this request. */
200 /* Time that this request was allocated for this IO. */
202 /* Time that I/O was submitted to the device. */
203 u64 io_start_time_ns;
205 #ifdef CONFIG_BLK_WBT
206 unsigned short wbt_flags;
209 * rq sectors used for blk stats. It has the same value
210 * with blk_rq_sectors(rq), except that it never be zeroed
213 unsigned short stats_sectors;
216 * Number of scatter-gather DMA addr+len pairs after
217 * physical address coalescing is performed.
219 unsigned short nr_phys_segments;
221 #if defined(CONFIG_BLK_DEV_INTEGRITY)
222 unsigned short nr_integrity_segments;
225 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
226 struct bio_crypt_ctx *crypt_ctx;
227 struct blk_ksm_keyslot *crypt_keyslot;
230 unsigned short write_hint;
231 unsigned short ioprio;
233 enum mq_rq_state state;
236 unsigned int timeout;
237 unsigned long deadline;
240 struct __call_single_data csd;
245 * completion callback.
247 rq_end_io_fn *end_io;
251 static inline bool blk_op_is_scsi(unsigned int op)
253 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
256 static inline bool blk_op_is_private(unsigned int op)
258 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
261 static inline bool blk_rq_is_scsi(struct request *rq)
263 return blk_op_is_scsi(req_op(rq));
266 static inline bool blk_rq_is_private(struct request *rq)
268 return blk_op_is_private(req_op(rq));
271 static inline bool blk_rq_is_passthrough(struct request *rq)
273 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
276 static inline bool bio_is_passthrough(struct bio *bio)
278 unsigned op = bio_op(bio);
280 return blk_op_is_scsi(op) || blk_op_is_private(op);
283 static inline unsigned short req_get_ioprio(struct request *req)
288 #include <linux/elevator.h>
290 struct blk_queue_ctx;
292 typedef blk_qc_t (make_request_fn) (struct request_queue *q, struct bio *bio);
296 enum blk_eh_timer_return {
297 BLK_EH_DONE, /* drivers has completed the command */
298 BLK_EH_RESET_TIMER, /* reset timer and try again */
301 enum blk_queue_state {
306 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
307 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
309 #define BLK_SCSI_MAX_CMDS (256)
310 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
313 * Zoned block device models (zoned limit).
315 enum blk_zoned_model {
316 BLK_ZONED_NONE, /* Regular block device */
317 BLK_ZONED_HA, /* Host-aware zoned block device */
318 BLK_ZONED_HM, /* Host-managed zoned block device */
321 struct queue_limits {
322 unsigned long bounce_pfn;
323 unsigned long seg_boundary_mask;
324 unsigned long virt_boundary_mask;
326 unsigned int max_hw_sectors;
327 unsigned int max_dev_sectors;
328 unsigned int chunk_sectors;
329 unsigned int max_sectors;
330 unsigned int max_segment_size;
331 unsigned int physical_block_size;
332 unsigned int logical_block_size;
333 unsigned int alignment_offset;
336 unsigned int max_discard_sectors;
337 unsigned int max_hw_discard_sectors;
338 unsigned int max_write_same_sectors;
339 unsigned int max_write_zeroes_sectors;
340 unsigned int max_zone_append_sectors;
341 unsigned int discard_granularity;
342 unsigned int discard_alignment;
344 unsigned short max_segments;
345 unsigned short max_integrity_segments;
346 unsigned short max_discard_segments;
348 unsigned char misaligned;
349 unsigned char discard_misaligned;
350 unsigned char raid_partial_stripes_expensive;
351 enum blk_zoned_model zoned;
354 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
357 #ifdef CONFIG_BLK_DEV_ZONED
359 #define BLK_ALL_ZONES ((unsigned int)-1)
360 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
361 unsigned int nr_zones, report_zones_cb cb, void *data);
362 unsigned int blkdev_nr_zones(struct gendisk *disk);
363 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
364 sector_t sectors, sector_t nr_sectors,
366 int blk_revalidate_disk_zones(struct gendisk *disk,
367 void (*update_driver_data)(struct gendisk *disk));
369 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
370 unsigned int cmd, unsigned long arg);
371 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
372 unsigned int cmd, unsigned long arg);
374 #else /* CONFIG_BLK_DEV_ZONED */
376 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
381 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
382 fmode_t mode, unsigned int cmd,
388 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
389 fmode_t mode, unsigned int cmd,
395 #endif /* CONFIG_BLK_DEV_ZONED */
397 struct request_queue {
398 struct request *last_merge;
399 struct elevator_queue *elevator;
401 struct blk_queue_stats *stats;
402 struct rq_qos *rq_qos;
404 make_request_fn *make_request_fn;
406 const struct blk_mq_ops *mq_ops;
409 struct blk_mq_ctx __percpu *queue_ctx;
411 unsigned int queue_depth;
413 /* hw dispatch queues */
414 struct blk_mq_hw_ctx **queue_hw_ctx;
415 unsigned int nr_hw_queues;
417 struct backing_dev_info *backing_dev_info;
420 * The queue owner gets to use this for whatever they like.
421 * ll_rw_blk doesn't touch it.
426 * various queue flags, see QUEUE_* below
428 unsigned long queue_flags;
430 * Number of contexts that have called blk_set_pm_only(). If this
431 * counter is above zero then only RQF_PM and RQF_PREEMPT requests are
437 * ida allocated id for this queue. Used to index queues from
443 * queue needs bounce pages for pages above this limit
447 spinlock_t queue_lock;
457 struct kobject *mq_kobj;
459 #ifdef CONFIG_BLK_DEV_INTEGRITY
460 struct blk_integrity integrity;
461 #endif /* CONFIG_BLK_DEV_INTEGRITY */
466 unsigned int nr_pending;
472 unsigned long nr_requests; /* Max # of requests */
474 unsigned int dma_pad_mask;
475 unsigned int dma_alignment;
477 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
478 /* Inline crypto capabilities */
479 struct blk_keyslot_manager *ksm;
482 unsigned int rq_timeout;
485 struct blk_stat_callback *poll_cb;
486 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
488 struct timer_list timeout;
489 struct work_struct timeout_work;
491 struct list_head icq_list;
492 #ifdef CONFIG_BLK_CGROUP
493 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
494 struct blkcg_gq *root_blkg;
495 struct list_head blkg_list;
498 struct queue_limits limits;
500 unsigned int required_elevator_features;
502 #ifdef CONFIG_BLK_DEV_ZONED
504 * Zoned block device information for request dispatch control.
505 * nr_zones is the total number of zones of the device. This is always
506 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
507 * bits which indicates if a zone is conventional (bit set) or
508 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
509 * bits which indicates if a zone is write locked, that is, if a write
510 * request targeting the zone was dispatched. All three fields are
511 * initialized by the low level device driver (e.g. scsi/sd.c).
512 * Stacking drivers (device mappers) may or may not initialize
515 * Reads of this information must be protected with blk_queue_enter() /
516 * blk_queue_exit(). Modifying this information is only allowed while
517 * no requests are being processed. See also blk_mq_freeze_queue() and
518 * blk_mq_unfreeze_queue().
520 unsigned int nr_zones;
521 unsigned long *conv_zones_bitmap;
522 unsigned long *seq_zones_wlock;
523 #endif /* CONFIG_BLK_DEV_ZONED */
528 unsigned int sg_timeout;
529 unsigned int sg_reserved_size;
531 #ifdef CONFIG_BLK_DEV_IO_TRACE
532 struct blk_trace __rcu *blk_trace;
533 struct mutex blk_trace_mutex;
536 * for flush operations
538 struct blk_flush_queue *fq;
540 struct list_head requeue_list;
541 spinlock_t requeue_lock;
542 struct delayed_work requeue_work;
544 struct mutex sysfs_lock;
545 struct mutex sysfs_dir_lock;
548 * for reusing dead hctx instance in case of updating
551 struct list_head unused_hctx_list;
552 spinlock_t unused_hctx_lock;
556 #if defined(CONFIG_BLK_DEV_BSG)
557 struct bsg_class_device bsg_dev;
560 #ifdef CONFIG_BLK_DEV_THROTTLING
562 struct throtl_data *td;
564 struct rcu_head rcu_head;
565 wait_queue_head_t mq_freeze_wq;
567 * Protect concurrent access to q_usage_counter by
568 * percpu_ref_kill() and percpu_ref_reinit().
570 struct mutex mq_freeze_lock;
571 struct percpu_ref q_usage_counter;
573 struct blk_mq_tag_set *tag_set;
574 struct list_head tag_set_list;
575 struct bio_set bio_split;
577 #ifdef CONFIG_BLK_DEBUG_FS
578 struct dentry *debugfs_dir;
579 struct dentry *sched_debugfs_dir;
580 struct dentry *rqos_debugfs_dir;
583 bool mq_sysfs_init_done;
587 struct work_struct release_work;
589 #define BLK_MAX_WRITE_HINTS 5
590 u64 write_hints[BLK_MAX_WRITE_HINTS];
593 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
594 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
595 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
596 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
597 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
598 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
599 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
600 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
601 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
602 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
603 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
604 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
605 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
606 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
607 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
608 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
609 #define QUEUE_FLAG_WC 17 /* Write back caching */
610 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
611 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
612 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
613 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
614 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
615 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
616 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
617 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
618 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
619 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
621 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
622 (1 << QUEUE_FLAG_SAME_COMP))
624 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
625 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
626 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
628 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
629 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
630 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
631 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
632 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
633 #define blk_queue_noxmerges(q) \
634 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
635 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
636 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
637 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
638 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
639 #define blk_queue_zone_resetall(q) \
640 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
641 #define blk_queue_secure_erase(q) \
642 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
643 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
644 #define blk_queue_scsi_passthrough(q) \
645 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
646 #define blk_queue_pci_p2pdma(q) \
647 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
648 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
649 #define blk_queue_rq_alloc_time(q) \
650 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
652 #define blk_queue_rq_alloc_time(q) false
655 #define blk_noretry_request(rq) \
656 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
657 REQ_FAILFAST_DRIVER))
658 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
659 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
660 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
661 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
663 extern void blk_set_pm_only(struct request_queue *q);
664 extern void blk_clear_pm_only(struct request_queue *q);
666 static inline bool blk_account_rq(struct request *rq)
668 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
671 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
673 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
675 #define rq_dma_dir(rq) \
676 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
678 #define dma_map_bvec(dev, bv, dir, attrs) \
679 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
682 static inline bool queue_is_mq(struct request_queue *q)
687 static inline enum blk_zoned_model
688 blk_queue_zoned_model(struct request_queue *q)
690 return q->limits.zoned;
693 static inline bool blk_queue_is_zoned(struct request_queue *q)
695 switch (blk_queue_zoned_model(q)) {
704 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
706 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
709 #ifdef CONFIG_BLK_DEV_ZONED
710 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
712 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
715 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
718 if (!blk_queue_is_zoned(q))
720 return sector >> ilog2(q->limits.chunk_sectors);
723 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
726 if (!blk_queue_is_zoned(q))
728 if (!q->conv_zones_bitmap)
730 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
732 #else /* CONFIG_BLK_DEV_ZONED */
733 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
737 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
742 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
747 #endif /* CONFIG_BLK_DEV_ZONED */
749 static inline bool rq_is_sync(struct request *rq)
751 return op_is_sync(rq->cmd_flags);
754 static inline bool rq_mergeable(struct request *rq)
756 if (blk_rq_is_passthrough(rq))
759 if (req_op(rq) == REQ_OP_FLUSH)
762 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
765 if (req_op(rq) == REQ_OP_ZONE_APPEND)
768 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
770 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
776 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
778 if (bio_page(a) == bio_page(b) &&
779 bio_offset(a) == bio_offset(b))
785 static inline unsigned int blk_queue_depth(struct request_queue *q)
788 return q->queue_depth;
790 return q->nr_requests;
793 extern unsigned long blk_max_low_pfn, blk_max_pfn;
796 * standard bounce addresses:
798 * BLK_BOUNCE_HIGH : bounce all highmem pages
799 * BLK_BOUNCE_ANY : don't bounce anything
800 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
803 #if BITS_PER_LONG == 32
804 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
806 #define BLK_BOUNCE_HIGH -1ULL
808 #define BLK_BOUNCE_ANY (-1ULL)
809 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
812 * default timeout for SG_IO if none specified
814 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
815 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
821 unsigned long offset;
826 struct req_iterator {
827 struct bvec_iter iter;
831 /* This should not be used directly - use rq_for_each_segment */
832 #define for_each_bio(_bio) \
833 for (; _bio; _bio = _bio->bi_next)
834 #define __rq_for_each_bio(_bio, rq) \
836 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
838 #define rq_for_each_segment(bvl, _rq, _iter) \
839 __rq_for_each_bio(_iter.bio, _rq) \
840 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
842 #define rq_for_each_bvec(bvl, _rq, _iter) \
843 __rq_for_each_bio(_iter.bio, _rq) \
844 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
846 #define rq_iter_last(bvec, _iter) \
847 (_iter.bio->bi_next == NULL && \
848 bio_iter_last(bvec, _iter.iter))
850 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
851 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
853 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
854 extern void rq_flush_dcache_pages(struct request *rq);
856 static inline void rq_flush_dcache_pages(struct request *rq)
861 extern int blk_register_queue(struct gendisk *disk);
862 extern void blk_unregister_queue(struct gendisk *disk);
863 extern blk_qc_t generic_make_request(struct bio *bio);
864 extern blk_qc_t direct_make_request(struct bio *bio);
865 extern void blk_rq_init(struct request_queue *q, struct request *rq);
866 extern void blk_put_request(struct request *);
867 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
868 blk_mq_req_flags_t flags);
869 extern int blk_lld_busy(struct request_queue *q);
870 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
871 struct bio_set *bs, gfp_t gfp_mask,
872 int (*bio_ctr)(struct bio *, struct bio *, void *),
874 extern void blk_rq_unprep_clone(struct request *rq);
875 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
877 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
878 extern void blk_queue_split(struct request_queue *, struct bio **);
879 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
880 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
881 unsigned int, void __user *);
882 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
883 unsigned int, void __user *);
884 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
885 struct scsi_ioctl_command __user *);
886 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
887 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
889 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
890 extern void blk_queue_exit(struct request_queue *q);
891 extern void blk_sync_queue(struct request_queue *q);
892 extern int blk_rq_map_user(struct request_queue *, struct request *,
893 struct rq_map_data *, void __user *, unsigned long,
895 extern int blk_rq_unmap_user(struct bio *);
896 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
897 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
898 struct rq_map_data *, const struct iov_iter *,
900 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
901 struct request *, int);
902 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
903 struct request *, int, rq_end_io_fn *);
905 /* Helper to convert REQ_OP_XXX to its string format XXX */
906 extern const char *blk_op_str(unsigned int op);
908 int blk_status_to_errno(blk_status_t status);
909 blk_status_t errno_to_blk_status(int errno);
911 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
913 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
915 return bdev->bd_disk->queue; /* this is never NULL */
919 * The basic unit of block I/O is a sector. It is used in a number of contexts
920 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
921 * bytes. Variables of type sector_t represent an offset or size that is a
922 * multiple of 512 bytes. Hence these two constants.
925 #define SECTOR_SHIFT 9
928 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
932 * blk_rq_pos() : the current sector
933 * blk_rq_bytes() : bytes left in the entire request
934 * blk_rq_cur_bytes() : bytes left in the current segment
935 * blk_rq_err_bytes() : bytes left till the next error boundary
936 * blk_rq_sectors() : sectors left in the entire request
937 * blk_rq_cur_sectors() : sectors left in the current segment
938 * blk_rq_stats_sectors() : sectors of the entire request used for stats
940 static inline sector_t blk_rq_pos(const struct request *rq)
945 static inline unsigned int blk_rq_bytes(const struct request *rq)
947 return rq->__data_len;
950 static inline int blk_rq_cur_bytes(const struct request *rq)
952 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
955 extern unsigned int blk_rq_err_bytes(const struct request *rq);
957 static inline unsigned int blk_rq_sectors(const struct request *rq)
959 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
962 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
964 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
967 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
969 return rq->stats_sectors;
972 #ifdef CONFIG_BLK_DEV_ZONED
974 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
975 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
977 static inline unsigned int blk_rq_zone_no(struct request *rq)
979 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
982 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
984 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
986 #endif /* CONFIG_BLK_DEV_ZONED */
989 * Some commands like WRITE SAME have a payload or data transfer size which
990 * is different from the size of the request. Any driver that supports such
991 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
992 * calculate the data transfer size.
994 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
996 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
997 return rq->special_vec.bv_len;
998 return blk_rq_bytes(rq);
1002 * Return the first full biovec in the request. The caller needs to check that
1003 * there are any bvecs before calling this helper.
1005 static inline struct bio_vec req_bvec(struct request *rq)
1007 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1008 return rq->special_vec;
1009 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1012 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1015 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1016 return min(q->limits.max_discard_sectors,
1017 UINT_MAX >> SECTOR_SHIFT);
1019 if (unlikely(op == REQ_OP_WRITE_SAME))
1020 return q->limits.max_write_same_sectors;
1022 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1023 return q->limits.max_write_zeroes_sectors;
1025 return q->limits.max_sectors;
1029 * Return maximum size of a request at given offset. Only valid for
1030 * file system requests.
1032 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1035 if (!q->limits.chunk_sectors)
1036 return q->limits.max_sectors;
1038 return min(q->limits.max_sectors, (unsigned int)(q->limits.chunk_sectors -
1039 (offset & (q->limits.chunk_sectors - 1))));
1042 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1045 struct request_queue *q = rq->q;
1047 if (blk_rq_is_passthrough(rq))
1048 return q->limits.max_hw_sectors;
1050 if (!q->limits.chunk_sectors ||
1051 req_op(rq) == REQ_OP_DISCARD ||
1052 req_op(rq) == REQ_OP_SECURE_ERASE)
1053 return blk_queue_get_max_sectors(q, req_op(rq));
1055 return min(blk_max_size_offset(q, offset),
1056 blk_queue_get_max_sectors(q, req_op(rq)));
1059 static inline unsigned int blk_rq_count_bios(struct request *rq)
1061 unsigned int nr_bios = 0;
1064 __rq_for_each_bio(bio, rq)
1070 void blk_steal_bios(struct bio_list *list, struct request *rq);
1073 * Request completion related functions.
1075 * blk_update_request() completes given number of bytes and updates
1076 * the request without completing it.
1078 extern bool blk_update_request(struct request *rq, blk_status_t error,
1079 unsigned int nr_bytes);
1081 extern void __blk_complete_request(struct request *);
1082 extern void blk_abort_request(struct request *);
1085 * Access functions for manipulating queue properties
1087 extern void blk_cleanup_queue(struct request_queue *);
1088 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1089 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1090 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1091 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1092 extern void blk_queue_max_discard_segments(struct request_queue *,
1094 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1095 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1096 unsigned int max_discard_sectors);
1097 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1098 unsigned int max_write_same_sectors);
1099 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1100 unsigned int max_write_same_sectors);
1101 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1102 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1103 unsigned int max_zone_append_sectors);
1104 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1105 extern void blk_queue_alignment_offset(struct request_queue *q,
1106 unsigned int alignment);
1107 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1108 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1109 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1110 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1111 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1112 extern void blk_set_default_limits(struct queue_limits *lim);
1113 extern void blk_set_stacking_limits(struct queue_limits *lim);
1114 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1116 extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev,
1118 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1120 extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b);
1121 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1122 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1123 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1124 extern void blk_queue_dma_alignment(struct request_queue *, int);
1125 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1126 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1127 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1128 extern void blk_queue_required_elevator_features(struct request_queue *q,
1129 unsigned int features);
1130 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1131 struct device *dev);
1134 * Number of physical segments as sent to the device.
1136 * Normally this is the number of discontiguous data segments sent by the
1137 * submitter. But for data-less command like discard we might have no
1138 * actual data segments submitted, but the driver might have to add it's
1139 * own special payload. In that case we still return 1 here so that this
1140 * special payload will be mapped.
1142 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1144 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1146 return rq->nr_phys_segments;
1150 * Number of discard segments (or ranges) the driver needs to fill in.
1151 * Each discard bio merged into a request is counted as one segment.
1153 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1155 return max_t(unsigned short, rq->nr_phys_segments, 1);
1158 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1159 struct scatterlist *sglist, struct scatterlist **last_sg);
1160 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1161 struct scatterlist *sglist)
1163 struct scatterlist *last_sg = NULL;
1165 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1167 extern void blk_dump_rq_flags(struct request *, char *);
1168 extern long nr_blockdev_pages(void);
1170 bool __must_check blk_get_queue(struct request_queue *);
1171 struct request_queue *blk_alloc_queue(make_request_fn make_request, int node_id);
1172 extern void blk_put_queue(struct request_queue *);
1173 extern void blk_set_queue_dying(struct request_queue *);
1176 * blk_plug permits building a queue of related requests by holding the I/O
1177 * fragments for a short period. This allows merging of sequential requests
1178 * into single larger request. As the requests are moved from a per-task list to
1179 * the device's request_queue in a batch, this results in improved scalability
1180 * as the lock contention for request_queue lock is reduced.
1182 * It is ok not to disable preemption when adding the request to the plug list
1183 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1184 * the plug list when the task sleeps by itself. For details, please see
1185 * schedule() where blk_schedule_flush_plug() is called.
1188 struct list_head mq_list; /* blk-mq requests */
1189 struct list_head cb_list; /* md requires an unplug callback */
1190 unsigned short rq_count;
1191 bool multiple_queues;
1193 #define BLK_MAX_REQUEST_COUNT 16
1194 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1197 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1198 struct blk_plug_cb {
1199 struct list_head list;
1200 blk_plug_cb_fn callback;
1203 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1204 void *data, int size);
1205 extern void blk_start_plug(struct blk_plug *);
1206 extern void blk_finish_plug(struct blk_plug *);
1207 extern void blk_flush_plug_list(struct blk_plug *, bool);
1209 static inline void blk_flush_plug(struct task_struct *tsk)
1211 struct blk_plug *plug = tsk->plug;
1214 blk_flush_plug_list(plug, false);
1217 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1219 struct blk_plug *plug = tsk->plug;
1222 blk_flush_plug_list(plug, true);
1225 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1227 struct blk_plug *plug = tsk->plug;
1230 (!list_empty(&plug->mq_list) ||
1231 !list_empty(&plug->cb_list));
1234 extern void blk_io_schedule(void);
1236 int blkdev_issue_flush(struct block_device *, gfp_t);
1237 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1238 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1240 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1242 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1243 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1244 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1245 sector_t nr_sects, gfp_t gfp_mask, int flags,
1248 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1249 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1251 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1252 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1254 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1255 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1257 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1258 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1260 return blkdev_issue_discard(sb->s_bdev,
1261 block << (sb->s_blocksize_bits -
1263 nr_blocks << (sb->s_blocksize_bits -
1267 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1268 sector_t nr_blocks, gfp_t gfp_mask)
1270 return blkdev_issue_zeroout(sb->s_bdev,
1271 block << (sb->s_blocksize_bits -
1273 nr_blocks << (sb->s_blocksize_bits -
1278 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1280 enum blk_default_limits {
1281 BLK_MAX_SEGMENTS = 128,
1282 BLK_SAFE_MAX_SECTORS = 255,
1283 BLK_DEF_MAX_SECTORS = 2560,
1284 BLK_MAX_SEGMENT_SIZE = 65536,
1285 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1288 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1290 return q->limits.seg_boundary_mask;
1293 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1295 return q->limits.virt_boundary_mask;
1298 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1300 return q->limits.max_sectors;
1303 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1305 return q->limits.max_hw_sectors;
1308 static inline unsigned short queue_max_segments(const struct request_queue *q)
1310 return q->limits.max_segments;
1313 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1315 return q->limits.max_discard_segments;
1318 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1320 return q->limits.max_segment_size;
1323 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1325 return q->limits.max_zone_append_sectors;
1328 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1332 if (q && q->limits.logical_block_size)
1333 retval = q->limits.logical_block_size;
1338 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1340 return queue_logical_block_size(bdev_get_queue(bdev));
1343 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1345 return q->limits.physical_block_size;
1348 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1350 return queue_physical_block_size(bdev_get_queue(bdev));
1353 static inline unsigned int queue_io_min(const struct request_queue *q)
1355 return q->limits.io_min;
1358 static inline int bdev_io_min(struct block_device *bdev)
1360 return queue_io_min(bdev_get_queue(bdev));
1363 static inline unsigned int queue_io_opt(const struct request_queue *q)
1365 return q->limits.io_opt;
1368 static inline int bdev_io_opt(struct block_device *bdev)
1370 return queue_io_opt(bdev_get_queue(bdev));
1373 static inline int queue_alignment_offset(const struct request_queue *q)
1375 if (q->limits.misaligned)
1378 return q->limits.alignment_offset;
1381 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1383 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1384 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1387 return (granularity + lim->alignment_offset - alignment) % granularity;
1390 static inline int bdev_alignment_offset(struct block_device *bdev)
1392 struct request_queue *q = bdev_get_queue(bdev);
1394 if (q->limits.misaligned)
1397 if (bdev != bdev->bd_contains)
1398 return bdev->bd_part->alignment_offset;
1400 return q->limits.alignment_offset;
1403 static inline int queue_discard_alignment(const struct request_queue *q)
1405 if (q->limits.discard_misaligned)
1408 return q->limits.discard_alignment;
1411 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1413 unsigned int alignment, granularity, offset;
1415 if (!lim->max_discard_sectors)
1418 /* Why are these in bytes, not sectors? */
1419 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1420 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1424 /* Offset of the partition start in 'granularity' sectors */
1425 offset = sector_div(sector, granularity);
1427 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1428 offset = (granularity + alignment - offset) % granularity;
1430 /* Turn it back into bytes, gaah */
1431 return offset << SECTOR_SHIFT;
1434 static inline int bdev_discard_alignment(struct block_device *bdev)
1436 struct request_queue *q = bdev_get_queue(bdev);
1438 if (bdev != bdev->bd_contains)
1439 return bdev->bd_part->discard_alignment;
1441 return q->limits.discard_alignment;
1444 static inline unsigned int bdev_write_same(struct block_device *bdev)
1446 struct request_queue *q = bdev_get_queue(bdev);
1449 return q->limits.max_write_same_sectors;
1454 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1456 struct request_queue *q = bdev_get_queue(bdev);
1459 return q->limits.max_write_zeroes_sectors;
1464 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1466 struct request_queue *q = bdev_get_queue(bdev);
1469 return blk_queue_zoned_model(q);
1471 return BLK_ZONED_NONE;
1474 static inline bool bdev_is_zoned(struct block_device *bdev)
1476 struct request_queue *q = bdev_get_queue(bdev);
1479 return blk_queue_is_zoned(q);
1484 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1486 struct request_queue *q = bdev_get_queue(bdev);
1489 return blk_queue_zone_sectors(q);
1493 static inline int queue_dma_alignment(const struct request_queue *q)
1495 return q ? q->dma_alignment : 511;
1498 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1501 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1502 return !(addr & alignment) && !(len & alignment);
1505 /* assumes size > 256 */
1506 static inline unsigned int blksize_bits(unsigned int size)
1508 unsigned int bits = 8;
1512 } while (size > 256);
1516 static inline unsigned int block_size(struct block_device *bdev)
1518 return bdev->bd_block_size;
1521 int kblockd_schedule_work(struct work_struct *work);
1522 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1524 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1525 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1526 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1527 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1529 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1531 enum blk_integrity_flags {
1532 BLK_INTEGRITY_VERIFY = 1 << 0,
1533 BLK_INTEGRITY_GENERATE = 1 << 1,
1534 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1535 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1538 struct blk_integrity_iter {
1542 unsigned int data_size;
1543 unsigned short interval;
1544 const char *disk_name;
1547 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1548 typedef void (integrity_prepare_fn) (struct request *);
1549 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1551 struct blk_integrity_profile {
1552 integrity_processing_fn *generate_fn;
1553 integrity_processing_fn *verify_fn;
1554 integrity_prepare_fn *prepare_fn;
1555 integrity_complete_fn *complete_fn;
1559 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1560 extern void blk_integrity_unregister(struct gendisk *);
1561 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1562 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1563 struct scatterlist *);
1564 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1565 extern bool blk_integrity_merge_rq(struct request_queue *, struct request *,
1567 extern bool blk_integrity_merge_bio(struct request_queue *, struct request *,
1570 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1572 struct blk_integrity *bi = &disk->queue->integrity;
1581 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1583 return blk_get_integrity(bdev->bd_disk);
1587 blk_integrity_queue_supports_integrity(struct request_queue *q)
1589 return q->integrity.profile;
1592 static inline bool blk_integrity_rq(struct request *rq)
1594 return rq->cmd_flags & REQ_INTEGRITY;
1597 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1600 q->limits.max_integrity_segments = segs;
1603 static inline unsigned short
1604 queue_max_integrity_segments(const struct request_queue *q)
1606 return q->limits.max_integrity_segments;
1610 * bio_integrity_intervals - Return number of integrity intervals for a bio
1611 * @bi: blk_integrity profile for device
1612 * @sectors: Size of the bio in 512-byte sectors
1614 * Description: The block layer calculates everything in 512 byte
1615 * sectors but integrity metadata is done in terms of the data integrity
1616 * interval size of the storage device. Convert the block layer sectors
1617 * to the appropriate number of integrity intervals.
1619 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1620 unsigned int sectors)
1622 return sectors >> (bi->interval_exp - 9);
1625 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1626 unsigned int sectors)
1628 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1632 * Return the first bvec that contains integrity data. Only drivers that are
1633 * limited to a single integrity segment should use this helper.
1635 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1637 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1639 return rq->bio->bi_integrity->bip_vec;
1642 #else /* CONFIG_BLK_DEV_INTEGRITY */
1645 struct block_device;
1647 struct blk_integrity;
1649 static inline int blk_integrity_rq(struct request *rq)
1653 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1658 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1660 struct scatterlist *s)
1664 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1668 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1673 blk_integrity_queue_supports_integrity(struct request_queue *q)
1677 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1681 static inline void blk_integrity_register(struct gendisk *d,
1682 struct blk_integrity *b)
1685 static inline void blk_integrity_unregister(struct gendisk *d)
1688 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1692 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1696 static inline bool blk_integrity_merge_rq(struct request_queue *rq,
1702 static inline bool blk_integrity_merge_bio(struct request_queue *rq,
1709 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1710 unsigned int sectors)
1715 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1716 unsigned int sectors)
1721 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1726 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1728 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1730 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1732 void blk_ksm_unregister(struct request_queue *q);
1734 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1736 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1737 struct request_queue *q)
1742 static inline void blk_ksm_unregister(struct request_queue *q) { }
1744 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1747 struct block_device_operations {
1748 int (*open) (struct block_device *, fmode_t);
1749 void (*release) (struct gendisk *, fmode_t);
1750 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1751 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1752 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1753 unsigned int (*check_events) (struct gendisk *disk,
1754 unsigned int clearing);
1755 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1756 int (*media_changed) (struct gendisk *);
1757 void (*unlock_native_capacity) (struct gendisk *);
1758 int (*revalidate_disk) (struct gendisk *);
1759 int (*getgeo)(struct block_device *, struct hd_geometry *);
1760 /* this callback is with swap_lock and sometimes page table lock held */
1761 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1762 int (*report_zones)(struct gendisk *, sector_t sector,
1763 unsigned int nr_zones, report_zones_cb cb, void *data);
1764 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1765 struct module *owner;
1766 const struct pr_ops *pr_ops;
1769 #ifdef CONFIG_COMPAT
1770 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1771 unsigned int, unsigned long);
1773 #define blkdev_compat_ptr_ioctl NULL
1776 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1778 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1779 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1780 struct writeback_control *);
1782 #ifdef CONFIG_BLK_DEV_ZONED
1783 bool blk_req_needs_zone_write_lock(struct request *rq);
1784 bool blk_req_zone_write_trylock(struct request *rq);
1785 void __blk_req_zone_write_lock(struct request *rq);
1786 void __blk_req_zone_write_unlock(struct request *rq);
1788 static inline void blk_req_zone_write_lock(struct request *rq)
1790 if (blk_req_needs_zone_write_lock(rq))
1791 __blk_req_zone_write_lock(rq);
1794 static inline void blk_req_zone_write_unlock(struct request *rq)
1796 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1797 __blk_req_zone_write_unlock(rq);
1800 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1802 return rq->q->seq_zones_wlock &&
1803 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1806 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1808 if (!blk_req_needs_zone_write_lock(rq))
1810 return !blk_req_zone_is_write_locked(rq);
1813 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1818 static inline void blk_req_zone_write_lock(struct request *rq)
1822 static inline void blk_req_zone_write_unlock(struct request *rq)
1825 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1830 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1834 #endif /* CONFIG_BLK_DEV_ZONED */
1836 #else /* CONFIG_BLOCK */
1838 struct block_device;
1841 * stubs for when the block layer is configured out
1843 #define buffer_heads_over_limit 0
1845 static inline long nr_blockdev_pages(void)
1853 static inline void blk_start_plug(struct blk_plug *plug)
1857 static inline void blk_finish_plug(struct blk_plug *plug)
1861 static inline void blk_flush_plug(struct task_struct *task)
1865 static inline void blk_schedule_flush_plug(struct task_struct *task)
1870 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1875 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask)
1880 #endif /* CONFIG_BLOCK */
1882 static inline void blk_wake_io_task(struct task_struct *waiter)
1885 * If we're polling, the task itself is doing the completions. For
1886 * that case, we don't need to signal a wakeup, it's enough to just
1887 * mark us as RUNNING.
1889 if (waiter == current)
1890 __set_current_state(TASK_RUNNING);
1892 wake_up_process(waiter);
1896 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1898 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1899 unsigned long start_time);
1902 * bio_start_io_acct - start I/O accounting for bio based drivers
1903 * @bio: bio to start account for
1905 * Returns the start time that should be passed back to bio_end_io_acct().
1907 static inline unsigned long bio_start_io_acct(struct bio *bio)
1909 return disk_start_io_acct(bio->bi_disk, bio_sectors(bio), bio_op(bio));
1913 * bio_end_io_acct - end I/O accounting for bio based drivers
1914 * @bio: bio to end account for
1915 * @start: start time returned by bio_start_io_acct()
1917 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1919 return disk_end_io_acct(bio->bi_disk, bio_op(bio), start_time);
1921 #endif /* CONFIG_BLOCK */