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>
30 struct scsi_ioctl_command;
33 struct elevator_queue;
39 struct blk_flush_queue;
42 struct blk_queue_stats;
43 struct blk_stat_callback;
44 struct blk_keyslot_manager;
46 #define BLKDEV_MIN_RQ 4
47 #define BLKDEV_MAX_RQ 128 /* Default maximum */
49 /* Must be consistent with blk_mq_poll_stats_bkt() */
50 #define BLK_MQ_POLL_STATS_BKTS 16
52 /* Doing classic polling */
53 #define BLK_MQ_POLL_CLASSIC -1
56 * Maximum number of blkcg policies allowed to be registered concurrently.
57 * Defined here to simplify include dependency.
59 #define BLKCG_MAX_POLS 5
61 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
65 typedef __u32 __bitwise req_flags_t;
67 /* elevator knows about this request */
68 #define RQF_SORTED ((__force req_flags_t)(1 << 0))
69 /* drive already may have started this one */
70 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
71 /* may not be passed by ioscheduler */
72 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
73 /* request for flush sequence */
74 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
75 /* merge of different types, fail separately */
76 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
77 /* track inflight for MQ */
78 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
79 /* don't call prep for this one */
80 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
81 /* set for "ide_preempt" requests and also for requests for which the SCSI
82 "quiesce" state must be ignored. */
83 #define RQF_PREEMPT ((__force req_flags_t)(1 << 8))
84 /* vaguely specified driver internal error. Ignored by the block layer */
85 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
86 /* don't warn about errors */
87 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
88 /* elevator private data attached */
89 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
90 /* account into disk and partition IO statistics */
91 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
92 /* request came from our alloc pool */
93 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
94 /* runtime pm request */
95 #define RQF_PM ((__force req_flags_t)(1 << 15))
96 /* on IO scheduler merge hash */
97 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
98 /* track IO completion time */
99 #define RQF_STATS ((__force req_flags_t)(1 << 17))
100 /* Look at ->special_vec for the actual data payload instead of the
102 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
103 /* The per-zone write lock is held for this request */
104 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
105 /* already slept for hybrid poll */
106 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
107 /* ->timeout has been called, don't expire again */
108 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
110 /* flags that prevent us from merging requests: */
111 #define RQF_NOMERGE_FLAGS \
112 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
115 * Request state for blk-mq.
124 * Try to put the fields that are referenced together in the same cacheline.
126 * If you modify this structure, make sure to update blk_rq_init() and
127 * especially blk_mq_rq_ctx_init() to take care of the added fields.
130 struct request_queue *q;
131 struct blk_mq_ctx *mq_ctx;
132 struct blk_mq_hw_ctx *mq_hctx;
134 unsigned int cmd_flags; /* op and common flags */
135 req_flags_t rq_flags;
140 /* the following two fields are internal, NEVER access directly */
141 unsigned int __data_len; /* total data len */
142 sector_t __sector; /* sector cursor */
147 struct list_head queuelist;
150 * The hash is used inside the scheduler, and killed once the
151 * request reaches the dispatch list. The ipi_list is only used
152 * to queue the request for softirq completion, which is long
153 * after the request has been unhashed (and even removed from
154 * the dispatch list).
157 struct hlist_node hash; /* merge hash */
158 struct list_head ipi_list;
162 * The rb_node is only used inside the io scheduler, requests
163 * are pruned when moved to the dispatch queue. So let the
164 * completion_data share space with the rb_node.
167 struct rb_node rb_node; /* sort/lookup */
168 struct bio_vec special_vec;
169 void *completion_data;
170 int error_count; /* for legacy drivers, don't use */
174 * Three pointers are available for the IO schedulers, if they need
175 * more they have to dynamically allocate it. Flush requests are
176 * never put on the IO scheduler. So let the flush fields share
177 * space with the elevator data.
187 struct list_head list;
188 rq_end_io_fn *saved_end_io;
192 struct gendisk *rq_disk;
193 struct hd_struct *part;
194 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
195 /* Time that the first bio started allocating this request. */
198 /* Time that this request was allocated for this IO. */
200 /* Time that I/O was submitted to the device. */
201 u64 io_start_time_ns;
203 #ifdef CONFIG_BLK_WBT
204 unsigned short wbt_flags;
207 * rq sectors used for blk stats. It has the same value
208 * with blk_rq_sectors(rq), except that it never be zeroed
211 unsigned short stats_sectors;
214 * Number of scatter-gather DMA addr+len pairs after
215 * physical address coalescing is performed.
217 unsigned short nr_phys_segments;
219 #if defined(CONFIG_BLK_DEV_INTEGRITY)
220 unsigned short nr_integrity_segments;
223 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
224 struct bio_crypt_ctx *crypt_ctx;
225 struct blk_ksm_keyslot *crypt_keyslot;
228 unsigned short write_hint;
229 unsigned short ioprio;
231 enum mq_rq_state state;
234 unsigned int timeout;
235 unsigned long deadline;
238 struct __call_single_data csd;
243 * completion callback.
245 rq_end_io_fn *end_io;
249 static inline bool blk_op_is_scsi(unsigned int op)
251 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
254 static inline bool blk_op_is_private(unsigned int op)
256 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
259 static inline bool blk_rq_is_scsi(struct request *rq)
261 return blk_op_is_scsi(req_op(rq));
264 static inline bool blk_rq_is_private(struct request *rq)
266 return blk_op_is_private(req_op(rq));
269 static inline bool blk_rq_is_passthrough(struct request *rq)
271 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
274 static inline bool bio_is_passthrough(struct bio *bio)
276 unsigned op = bio_op(bio);
278 return blk_op_is_scsi(op) || blk_op_is_private(op);
281 static inline unsigned short req_get_ioprio(struct request *req)
286 #include <linux/elevator.h>
288 struct blk_queue_ctx;
292 enum blk_eh_timer_return {
293 BLK_EH_DONE, /* drivers has completed the command */
294 BLK_EH_RESET_TIMER, /* reset timer and try again */
297 enum blk_queue_state {
302 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
303 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
305 #define BLK_SCSI_MAX_CMDS (256)
306 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
309 * Zoned block device models (zoned limit).
311 * Note: This needs to be ordered from the least to the most severe
312 * restrictions for the inheritance in blk_stack_limits() to work.
314 enum blk_zoned_model {
315 BLK_ZONED_NONE = 0, /* Regular block device */
316 BLK_ZONED_HA, /* Host-aware zoned block device */
317 BLK_ZONED_HM, /* Host-managed zoned block device */
320 struct queue_limits {
321 unsigned long bounce_pfn;
322 unsigned long seg_boundary_mask;
323 unsigned long virt_boundary_mask;
325 unsigned int max_hw_sectors;
326 unsigned int max_dev_sectors;
327 unsigned int chunk_sectors;
328 unsigned int max_sectors;
329 unsigned int max_segment_size;
330 unsigned int physical_block_size;
331 unsigned int logical_block_size;
332 unsigned int alignment_offset;
335 unsigned int max_discard_sectors;
336 unsigned int max_hw_discard_sectors;
337 unsigned int max_write_same_sectors;
338 unsigned int max_write_zeroes_sectors;
339 unsigned int max_zone_append_sectors;
340 unsigned int discard_granularity;
341 unsigned int discard_alignment;
343 unsigned short max_segments;
344 unsigned short max_integrity_segments;
345 unsigned short max_discard_segments;
347 unsigned char misaligned;
348 unsigned char discard_misaligned;
349 unsigned char raid_partial_stripes_expensive;
350 enum blk_zoned_model zoned;
353 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
356 #ifdef CONFIG_BLK_DEV_ZONED
358 #define BLK_ALL_ZONES ((unsigned int)-1)
359 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
360 unsigned int nr_zones, report_zones_cb cb, void *data);
361 unsigned int blkdev_nr_zones(struct gendisk *disk);
362 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
363 sector_t sectors, sector_t nr_sectors,
365 int blk_revalidate_disk_zones(struct gendisk *disk,
366 void (*update_driver_data)(struct gendisk *disk));
368 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
369 unsigned int cmd, unsigned long arg);
370 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
371 unsigned int cmd, unsigned long arg);
373 #else /* CONFIG_BLK_DEV_ZONED */
375 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
380 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
381 fmode_t mode, unsigned int cmd,
387 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
388 fmode_t mode, unsigned int cmd,
394 #endif /* CONFIG_BLK_DEV_ZONED */
396 struct request_queue {
397 struct request *last_merge;
398 struct elevator_queue *elevator;
400 struct blk_queue_stats *stats;
401 struct rq_qos *rq_qos;
403 const struct blk_mq_ops *mq_ops;
406 struct blk_mq_ctx __percpu *queue_ctx;
408 unsigned int queue_depth;
410 /* hw dispatch queues */
411 struct blk_mq_hw_ctx **queue_hw_ctx;
412 unsigned int nr_hw_queues;
414 struct backing_dev_info *backing_dev_info;
417 * The queue owner gets to use this for whatever they like.
418 * ll_rw_blk doesn't touch it.
423 * various queue flags, see QUEUE_* below
425 unsigned long queue_flags;
427 * Number of contexts that have called blk_set_pm_only(). If this
428 * counter is above zero then only RQF_PM and RQF_PREEMPT requests are
434 * ida allocated id for this queue. Used to index queues from
440 * queue needs bounce pages for pages above this limit
444 spinlock_t queue_lock;
454 struct kobject *mq_kobj;
456 #ifdef CONFIG_BLK_DEV_INTEGRITY
457 struct blk_integrity integrity;
458 #endif /* CONFIG_BLK_DEV_INTEGRITY */
462 enum rpm_status rpm_status;
463 unsigned int nr_pending;
469 unsigned long nr_requests; /* Max # of requests */
471 unsigned int dma_pad_mask;
472 unsigned int dma_alignment;
474 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
475 /* Inline crypto capabilities */
476 struct blk_keyslot_manager *ksm;
479 unsigned int rq_timeout;
482 struct blk_stat_callback *poll_cb;
483 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
485 struct timer_list timeout;
486 struct work_struct timeout_work;
488 atomic_t nr_active_requests_shared_sbitmap;
490 struct list_head icq_list;
491 #ifdef CONFIG_BLK_CGROUP
492 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
493 struct blkcg_gq *root_blkg;
494 struct list_head blkg_list;
497 struct queue_limits limits;
499 unsigned int required_elevator_features;
501 #ifdef CONFIG_BLK_DEV_ZONED
503 * Zoned block device information for request dispatch control.
504 * nr_zones is the total number of zones of the device. This is always
505 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
506 * bits which indicates if a zone is conventional (bit set) or
507 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
508 * bits which indicates if a zone is write locked, that is, if a write
509 * request targeting the zone was dispatched. All three fields are
510 * initialized by the low level device driver (e.g. scsi/sd.c).
511 * Stacking drivers (device mappers) may or may not initialize
514 * Reads of this information must be protected with blk_queue_enter() /
515 * blk_queue_exit(). Modifying this information is only allowed while
516 * no requests are being processed. See also blk_mq_freeze_queue() and
517 * blk_mq_unfreeze_queue().
519 unsigned int nr_zones;
520 unsigned long *conv_zones_bitmap;
521 unsigned long *seq_zones_wlock;
522 unsigned int max_open_zones;
523 unsigned int max_active_zones;
524 #endif /* CONFIG_BLK_DEV_ZONED */
529 unsigned int sg_timeout;
530 unsigned int sg_reserved_size;
532 struct mutex debugfs_mutex;
533 #ifdef CONFIG_BLK_DEV_IO_TRACE
534 struct blk_trace __rcu *blk_trace;
537 * for flush operations
539 struct blk_flush_queue *fq;
541 struct list_head requeue_list;
542 spinlock_t requeue_lock;
543 struct delayed_work requeue_work;
545 struct mutex sysfs_lock;
546 struct mutex sysfs_dir_lock;
549 * for reusing dead hctx instance in case of updating
552 struct list_head unused_hctx_list;
553 spinlock_t unused_hctx_lock;
557 #if defined(CONFIG_BLK_DEV_BSG)
558 struct bsg_class_device bsg_dev;
561 #ifdef CONFIG_BLK_DEV_THROTTLING
563 struct throtl_data *td;
565 struct rcu_head rcu_head;
566 wait_queue_head_t mq_freeze_wq;
568 * Protect concurrent access to q_usage_counter by
569 * percpu_ref_kill() and percpu_ref_reinit().
571 struct mutex mq_freeze_lock;
572 struct percpu_ref q_usage_counter;
574 struct blk_mq_tag_set *tag_set;
575 struct list_head tag_set_list;
576 struct bio_set bio_split;
578 struct dentry *debugfs_dir;
580 #ifdef CONFIG_BLK_DEBUG_FS
581 struct dentry *sched_debugfs_dir;
582 struct dentry *rqos_debugfs_dir;
585 bool mq_sysfs_init_done;
589 #define BLK_MAX_WRITE_HINTS 5
590 u64 write_hints[BLK_MAX_WRITE_HINTS];
593 /* Keep blk_queue_flag_name[] in sync with the definitions below */
594 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
595 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
596 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
597 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
598 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
599 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
600 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
601 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
602 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
603 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
604 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
605 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
606 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
607 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
608 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
609 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
610 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
611 #define QUEUE_FLAG_WC 17 /* Write back caching */
612 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
613 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
614 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
615 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
616 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
617 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
618 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
619 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
620 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
621 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
622 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
624 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
625 (1 << QUEUE_FLAG_SAME_COMP))
627 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
628 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
629 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
631 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
632 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
633 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
634 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
635 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
636 #define blk_queue_noxmerges(q) \
637 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
638 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
639 #define blk_queue_stable_writes(q) \
640 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
641 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
642 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
643 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
644 #define blk_queue_zone_resetall(q) \
645 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
646 #define blk_queue_secure_erase(q) \
647 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
648 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
649 #define blk_queue_scsi_passthrough(q) \
650 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
651 #define blk_queue_pci_p2pdma(q) \
652 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
653 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
654 #define blk_queue_rq_alloc_time(q) \
655 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
657 #define blk_queue_rq_alloc_time(q) false
660 #define blk_noretry_request(rq) \
661 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
662 REQ_FAILFAST_DRIVER))
663 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
664 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
665 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
666 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
668 extern void blk_set_pm_only(struct request_queue *q);
669 extern void blk_clear_pm_only(struct request_queue *q);
671 static inline bool blk_account_rq(struct request *rq)
673 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
676 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
678 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
680 #define rq_dma_dir(rq) \
681 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
683 #define dma_map_bvec(dev, bv, dir, attrs) \
684 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
687 static inline bool queue_is_mq(struct request_queue *q)
692 static inline enum blk_zoned_model
693 blk_queue_zoned_model(struct request_queue *q)
695 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
696 return q->limits.zoned;
697 return BLK_ZONED_NONE;
700 static inline bool blk_queue_is_zoned(struct request_queue *q)
702 switch (blk_queue_zoned_model(q)) {
711 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
713 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
716 #ifdef CONFIG_BLK_DEV_ZONED
717 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
719 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
722 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
725 if (!blk_queue_is_zoned(q))
727 return sector >> ilog2(q->limits.chunk_sectors);
730 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
733 if (!blk_queue_is_zoned(q))
735 if (!q->conv_zones_bitmap)
737 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
740 static inline void blk_queue_max_open_zones(struct request_queue *q,
741 unsigned int max_open_zones)
743 q->max_open_zones = max_open_zones;
746 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
748 return q->max_open_zones;
751 static inline void blk_queue_max_active_zones(struct request_queue *q,
752 unsigned int max_active_zones)
754 q->max_active_zones = max_active_zones;
757 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
759 return q->max_active_zones;
761 #else /* CONFIG_BLK_DEV_ZONED */
762 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
766 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
771 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
776 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
780 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
784 #endif /* CONFIG_BLK_DEV_ZONED */
786 static inline bool rq_is_sync(struct request *rq)
788 return op_is_sync(rq->cmd_flags);
791 static inline bool rq_mergeable(struct request *rq)
793 if (blk_rq_is_passthrough(rq))
796 if (req_op(rq) == REQ_OP_FLUSH)
799 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
802 if (req_op(rq) == REQ_OP_ZONE_APPEND)
805 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
807 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
813 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
815 if (bio_page(a) == bio_page(b) &&
816 bio_offset(a) == bio_offset(b))
822 static inline unsigned int blk_queue_depth(struct request_queue *q)
825 return q->queue_depth;
827 return q->nr_requests;
830 extern unsigned long blk_max_low_pfn, blk_max_pfn;
833 * standard bounce addresses:
835 * BLK_BOUNCE_HIGH : bounce all highmem pages
836 * BLK_BOUNCE_ANY : don't bounce anything
837 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
840 #if BITS_PER_LONG == 32
841 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
843 #define BLK_BOUNCE_HIGH -1ULL
845 #define BLK_BOUNCE_ANY (-1ULL)
846 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
849 * default timeout for SG_IO if none specified
851 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
852 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
858 unsigned long offset;
863 struct req_iterator {
864 struct bvec_iter iter;
868 /* This should not be used directly - use rq_for_each_segment */
869 #define for_each_bio(_bio) \
870 for (; _bio; _bio = _bio->bi_next)
871 #define __rq_for_each_bio(_bio, rq) \
873 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
875 #define rq_for_each_segment(bvl, _rq, _iter) \
876 __rq_for_each_bio(_iter.bio, _rq) \
877 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
879 #define rq_for_each_bvec(bvl, _rq, _iter) \
880 __rq_for_each_bio(_iter.bio, _rq) \
881 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
883 #define rq_iter_last(bvec, _iter) \
884 (_iter.bio->bi_next == NULL && \
885 bio_iter_last(bvec, _iter.iter))
887 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
888 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
890 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
891 extern void rq_flush_dcache_pages(struct request *rq);
893 static inline void rq_flush_dcache_pages(struct request *rq)
898 extern int blk_register_queue(struct gendisk *disk);
899 extern void blk_unregister_queue(struct gendisk *disk);
900 blk_qc_t submit_bio_noacct(struct bio *bio);
901 extern void blk_rq_init(struct request_queue *q, struct request *rq);
902 extern void blk_put_request(struct request *);
903 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
904 blk_mq_req_flags_t flags);
905 extern int blk_lld_busy(struct request_queue *q);
906 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
907 struct bio_set *bs, gfp_t gfp_mask,
908 int (*bio_ctr)(struct bio *, struct bio *, void *),
910 extern void blk_rq_unprep_clone(struct request *rq);
911 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
913 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
914 extern void blk_queue_split(struct bio **);
915 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
916 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
917 unsigned int, void __user *);
918 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
919 unsigned int, void __user *);
920 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
921 struct scsi_ioctl_command __user *);
922 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
923 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
925 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
926 extern void blk_queue_exit(struct request_queue *q);
927 extern void blk_sync_queue(struct request_queue *q);
928 extern int blk_rq_map_user(struct request_queue *, struct request *,
929 struct rq_map_data *, void __user *, unsigned long,
931 extern int blk_rq_unmap_user(struct bio *);
932 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
933 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
934 struct rq_map_data *, const struct iov_iter *,
936 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
937 struct request *, int);
938 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
939 struct request *, int, rq_end_io_fn *);
941 /* Helper to convert REQ_OP_XXX to its string format XXX */
942 extern const char *blk_op_str(unsigned int op);
944 int blk_status_to_errno(blk_status_t status);
945 blk_status_t errno_to_blk_status(int errno);
947 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
949 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
951 return bdev->bd_disk->queue; /* this is never NULL */
955 * The basic unit of block I/O is a sector. It is used in a number of contexts
956 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
957 * bytes. Variables of type sector_t represent an offset or size that is a
958 * multiple of 512 bytes. Hence these two constants.
961 #define SECTOR_SHIFT 9
964 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
968 * blk_rq_pos() : the current sector
969 * blk_rq_bytes() : bytes left in the entire request
970 * blk_rq_cur_bytes() : bytes left in the current segment
971 * blk_rq_err_bytes() : bytes left till the next error boundary
972 * blk_rq_sectors() : sectors left in the entire request
973 * blk_rq_cur_sectors() : sectors left in the current segment
974 * blk_rq_stats_sectors() : sectors of the entire request used for stats
976 static inline sector_t blk_rq_pos(const struct request *rq)
981 static inline unsigned int blk_rq_bytes(const struct request *rq)
983 return rq->__data_len;
986 static inline int blk_rq_cur_bytes(const struct request *rq)
988 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
991 extern unsigned int blk_rq_err_bytes(const struct request *rq);
993 static inline unsigned int blk_rq_sectors(const struct request *rq)
995 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
998 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
1000 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
1003 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
1005 return rq->stats_sectors;
1008 #ifdef CONFIG_BLK_DEV_ZONED
1010 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
1011 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
1013 static inline unsigned int blk_rq_zone_no(struct request *rq)
1015 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1018 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1020 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1022 #endif /* CONFIG_BLK_DEV_ZONED */
1025 * Some commands like WRITE SAME have a payload or data transfer size which
1026 * is different from the size of the request. Any driver that supports such
1027 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1028 * calculate the data transfer size.
1030 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1032 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1033 return rq->special_vec.bv_len;
1034 return blk_rq_bytes(rq);
1038 * Return the first full biovec in the request. The caller needs to check that
1039 * there are any bvecs before calling this helper.
1041 static inline struct bio_vec req_bvec(struct request *rq)
1043 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1044 return rq->special_vec;
1045 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1048 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1051 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1052 return min(q->limits.max_discard_sectors,
1053 UINT_MAX >> SECTOR_SHIFT);
1055 if (unlikely(op == REQ_OP_WRITE_SAME))
1056 return q->limits.max_write_same_sectors;
1058 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1059 return q->limits.max_write_zeroes_sectors;
1061 return q->limits.max_sectors;
1065 * Return maximum size of a request at given offset. Only valid for
1066 * file system requests.
1068 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1071 unsigned int chunk_sectors = q->limits.chunk_sectors;
1074 return q->limits.max_sectors;
1076 if (likely(is_power_of_2(chunk_sectors)))
1077 chunk_sectors -= offset & (chunk_sectors - 1);
1079 chunk_sectors -= sector_div(offset, chunk_sectors);
1081 return min(q->limits.max_sectors, chunk_sectors);
1084 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1087 struct request_queue *q = rq->q;
1089 if (blk_rq_is_passthrough(rq))
1090 return q->limits.max_hw_sectors;
1092 if (!q->limits.chunk_sectors ||
1093 req_op(rq) == REQ_OP_DISCARD ||
1094 req_op(rq) == REQ_OP_SECURE_ERASE)
1095 return blk_queue_get_max_sectors(q, req_op(rq));
1097 return min(blk_max_size_offset(q, offset),
1098 blk_queue_get_max_sectors(q, req_op(rq)));
1101 static inline unsigned int blk_rq_count_bios(struct request *rq)
1103 unsigned int nr_bios = 0;
1106 __rq_for_each_bio(bio, rq)
1112 void blk_steal_bios(struct bio_list *list, struct request *rq);
1115 * Request completion related functions.
1117 * blk_update_request() completes given number of bytes and updates
1118 * the request without completing it.
1120 extern bool blk_update_request(struct request *rq, blk_status_t error,
1121 unsigned int nr_bytes);
1123 extern void blk_abort_request(struct request *);
1126 * Access functions for manipulating queue properties
1128 extern void blk_cleanup_queue(struct request_queue *);
1129 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1130 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1131 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1132 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1133 extern void blk_queue_max_discard_segments(struct request_queue *,
1135 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1136 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1137 unsigned int max_discard_sectors);
1138 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1139 unsigned int max_write_same_sectors);
1140 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1141 unsigned int max_write_same_sectors);
1142 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1143 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1144 unsigned int max_zone_append_sectors);
1145 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1146 extern void blk_queue_alignment_offset(struct request_queue *q,
1147 unsigned int alignment);
1148 void blk_queue_update_readahead(struct request_queue *q);
1149 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1150 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1151 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1152 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1153 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1154 extern void blk_set_default_limits(struct queue_limits *lim);
1155 extern void blk_set_stacking_limits(struct queue_limits *lim);
1156 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1158 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1160 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1161 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1162 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1163 extern void blk_queue_dma_alignment(struct request_queue *, int);
1164 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1165 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1166 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1167 extern void blk_queue_required_elevator_features(struct request_queue *q,
1168 unsigned int features);
1169 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1170 struct device *dev);
1173 * Number of physical segments as sent to the device.
1175 * Normally this is the number of discontiguous data segments sent by the
1176 * submitter. But for data-less command like discard we might have no
1177 * actual data segments submitted, but the driver might have to add it's
1178 * own special payload. In that case we still return 1 here so that this
1179 * special payload will be mapped.
1181 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1183 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1185 return rq->nr_phys_segments;
1189 * Number of discard segments (or ranges) the driver needs to fill in.
1190 * Each discard bio merged into a request is counted as one segment.
1192 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1194 return max_t(unsigned short, rq->nr_phys_segments, 1);
1197 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1198 struct scatterlist *sglist, struct scatterlist **last_sg);
1199 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1200 struct scatterlist *sglist)
1202 struct scatterlist *last_sg = NULL;
1204 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1206 extern void blk_dump_rq_flags(struct request *, char *);
1208 bool __must_check blk_get_queue(struct request_queue *);
1209 struct request_queue *blk_alloc_queue(int node_id);
1210 extern void blk_put_queue(struct request_queue *);
1211 extern void blk_set_queue_dying(struct request_queue *);
1215 * blk_plug permits building a queue of related requests by holding the I/O
1216 * fragments for a short period. This allows merging of sequential requests
1217 * into single larger request. As the requests are moved from a per-task list to
1218 * the device's request_queue in a batch, this results in improved scalability
1219 * as the lock contention for request_queue lock is reduced.
1221 * It is ok not to disable preemption when adding the request to the plug list
1222 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1223 * the plug list when the task sleeps by itself. For details, please see
1224 * schedule() where blk_schedule_flush_plug() is called.
1227 struct list_head mq_list; /* blk-mq requests */
1228 struct list_head cb_list; /* md requires an unplug callback */
1229 unsigned short rq_count;
1230 bool multiple_queues;
1233 #define BLK_MAX_REQUEST_COUNT 16
1234 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1237 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1238 struct blk_plug_cb {
1239 struct list_head list;
1240 blk_plug_cb_fn callback;
1243 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1244 void *data, int size);
1245 extern void blk_start_plug(struct blk_plug *);
1246 extern void blk_finish_plug(struct blk_plug *);
1247 extern void blk_flush_plug_list(struct blk_plug *, bool);
1249 static inline void blk_flush_plug(struct task_struct *tsk)
1251 struct blk_plug *plug = tsk->plug;
1254 blk_flush_plug_list(plug, false);
1257 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1259 struct blk_plug *plug = tsk->plug;
1262 blk_flush_plug_list(plug, true);
1265 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1267 struct blk_plug *plug = tsk->plug;
1270 (!list_empty(&plug->mq_list) ||
1271 !list_empty(&plug->cb_list));
1274 int blkdev_issue_flush(struct block_device *, gfp_t);
1275 long nr_blockdev_pages(void);
1276 #else /* CONFIG_BLOCK */
1280 static inline void blk_start_plug(struct blk_plug *plug)
1284 static inline void blk_finish_plug(struct blk_plug *plug)
1288 static inline void blk_flush_plug(struct task_struct *task)
1292 static inline void blk_schedule_flush_plug(struct task_struct *task)
1297 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1302 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask)
1307 static inline long nr_blockdev_pages(void)
1311 #endif /* CONFIG_BLOCK */
1313 extern void blk_io_schedule(void);
1315 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1316 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1318 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1320 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1321 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1322 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1323 sector_t nr_sects, gfp_t gfp_mask, int flags,
1326 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1327 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1329 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1330 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1332 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1333 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1335 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1336 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1338 return blkdev_issue_discard(sb->s_bdev,
1339 block << (sb->s_blocksize_bits -
1341 nr_blocks << (sb->s_blocksize_bits -
1345 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1346 sector_t nr_blocks, gfp_t gfp_mask)
1348 return blkdev_issue_zeroout(sb->s_bdev,
1349 block << (sb->s_blocksize_bits -
1351 nr_blocks << (sb->s_blocksize_bits -
1356 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1358 enum blk_default_limits {
1359 BLK_MAX_SEGMENTS = 128,
1360 BLK_SAFE_MAX_SECTORS = 255,
1361 BLK_DEF_MAX_SECTORS = 2560,
1362 BLK_MAX_SEGMENT_SIZE = 65536,
1363 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1366 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1368 return q->limits.seg_boundary_mask;
1371 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1373 return q->limits.virt_boundary_mask;
1376 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1378 return q->limits.max_sectors;
1381 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1383 return q->limits.max_hw_sectors;
1386 static inline unsigned short queue_max_segments(const struct request_queue *q)
1388 return q->limits.max_segments;
1391 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1393 return q->limits.max_discard_segments;
1396 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1398 return q->limits.max_segment_size;
1401 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1403 return q->limits.max_zone_append_sectors;
1406 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1410 if (q && q->limits.logical_block_size)
1411 retval = q->limits.logical_block_size;
1416 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1418 return queue_logical_block_size(bdev_get_queue(bdev));
1421 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1423 return q->limits.physical_block_size;
1426 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1428 return queue_physical_block_size(bdev_get_queue(bdev));
1431 static inline unsigned int queue_io_min(const struct request_queue *q)
1433 return q->limits.io_min;
1436 static inline int bdev_io_min(struct block_device *bdev)
1438 return queue_io_min(bdev_get_queue(bdev));
1441 static inline unsigned int queue_io_opt(const struct request_queue *q)
1443 return q->limits.io_opt;
1446 static inline int bdev_io_opt(struct block_device *bdev)
1448 return queue_io_opt(bdev_get_queue(bdev));
1451 static inline int queue_alignment_offset(const struct request_queue *q)
1453 if (q->limits.misaligned)
1456 return q->limits.alignment_offset;
1459 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1461 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1462 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1465 return (granularity + lim->alignment_offset - alignment) % granularity;
1468 static inline int bdev_alignment_offset(struct block_device *bdev)
1470 struct request_queue *q = bdev_get_queue(bdev);
1472 if (q->limits.misaligned)
1474 if (bdev != bdev->bd_contains)
1475 return queue_limit_alignment_offset(&q->limits,
1476 bdev->bd_part->start_sect);
1477 return q->limits.alignment_offset;
1480 static inline int queue_discard_alignment(const struct request_queue *q)
1482 if (q->limits.discard_misaligned)
1485 return q->limits.discard_alignment;
1488 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1490 unsigned int alignment, granularity, offset;
1492 if (!lim->max_discard_sectors)
1495 /* Why are these in bytes, not sectors? */
1496 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1497 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1501 /* Offset of the partition start in 'granularity' sectors */
1502 offset = sector_div(sector, granularity);
1504 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1505 offset = (granularity + alignment - offset) % granularity;
1507 /* Turn it back into bytes, gaah */
1508 return offset << SECTOR_SHIFT;
1511 static inline int bdev_discard_alignment(struct block_device *bdev)
1513 struct request_queue *q = bdev_get_queue(bdev);
1515 if (bdev != bdev->bd_contains)
1516 return queue_limit_discard_alignment(&q->limits,
1517 bdev->bd_part->start_sect);
1518 return q->limits.discard_alignment;
1521 static inline unsigned int bdev_write_same(struct block_device *bdev)
1523 struct request_queue *q = bdev_get_queue(bdev);
1526 return q->limits.max_write_same_sectors;
1531 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1533 struct request_queue *q = bdev_get_queue(bdev);
1536 return q->limits.max_write_zeroes_sectors;
1541 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1543 struct request_queue *q = bdev_get_queue(bdev);
1546 return blk_queue_zoned_model(q);
1548 return BLK_ZONED_NONE;
1551 static inline bool bdev_is_zoned(struct block_device *bdev)
1553 struct request_queue *q = bdev_get_queue(bdev);
1556 return blk_queue_is_zoned(q);
1561 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1563 struct request_queue *q = bdev_get_queue(bdev);
1566 return blk_queue_zone_sectors(q);
1570 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1572 struct request_queue *q = bdev_get_queue(bdev);
1575 return queue_max_open_zones(q);
1579 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1581 struct request_queue *q = bdev_get_queue(bdev);
1584 return queue_max_active_zones(q);
1588 static inline int queue_dma_alignment(const struct request_queue *q)
1590 return q ? q->dma_alignment : 511;
1593 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1596 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1597 return !(addr & alignment) && !(len & alignment);
1600 /* assumes size > 256 */
1601 static inline unsigned int blksize_bits(unsigned int size)
1603 unsigned int bits = 8;
1607 } while (size > 256);
1611 static inline unsigned int block_size(struct block_device *bdev)
1613 return 1 << bdev->bd_inode->i_blkbits;
1616 int kblockd_schedule_work(struct work_struct *work);
1617 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1619 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1620 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1621 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1622 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1624 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1626 enum blk_integrity_flags {
1627 BLK_INTEGRITY_VERIFY = 1 << 0,
1628 BLK_INTEGRITY_GENERATE = 1 << 1,
1629 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1630 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1633 struct blk_integrity_iter {
1637 unsigned int data_size;
1638 unsigned short interval;
1639 const char *disk_name;
1642 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1643 typedef void (integrity_prepare_fn) (struct request *);
1644 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1646 struct blk_integrity_profile {
1647 integrity_processing_fn *generate_fn;
1648 integrity_processing_fn *verify_fn;
1649 integrity_prepare_fn *prepare_fn;
1650 integrity_complete_fn *complete_fn;
1654 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1655 extern void blk_integrity_unregister(struct gendisk *);
1656 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1657 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1658 struct scatterlist *);
1659 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1660 extern bool blk_integrity_merge_rq(struct request_queue *, struct request *,
1662 extern bool blk_integrity_merge_bio(struct request_queue *, struct request *,
1665 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1667 struct blk_integrity *bi = &disk->queue->integrity;
1676 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1678 return blk_get_integrity(bdev->bd_disk);
1682 blk_integrity_queue_supports_integrity(struct request_queue *q)
1684 return q->integrity.profile;
1687 static inline bool blk_integrity_rq(struct request *rq)
1689 return rq->cmd_flags & REQ_INTEGRITY;
1692 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1695 q->limits.max_integrity_segments = segs;
1698 static inline unsigned short
1699 queue_max_integrity_segments(const struct request_queue *q)
1701 return q->limits.max_integrity_segments;
1705 * bio_integrity_intervals - Return number of integrity intervals for a bio
1706 * @bi: blk_integrity profile for device
1707 * @sectors: Size of the bio in 512-byte sectors
1709 * Description: The block layer calculates everything in 512 byte
1710 * sectors but integrity metadata is done in terms of the data integrity
1711 * interval size of the storage device. Convert the block layer sectors
1712 * to the appropriate number of integrity intervals.
1714 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1715 unsigned int sectors)
1717 return sectors >> (bi->interval_exp - 9);
1720 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1721 unsigned int sectors)
1723 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1727 * Return the first bvec that contains integrity data. Only drivers that are
1728 * limited to a single integrity segment should use this helper.
1730 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1732 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1734 return rq->bio->bi_integrity->bip_vec;
1737 #else /* CONFIG_BLK_DEV_INTEGRITY */
1740 struct block_device;
1742 struct blk_integrity;
1744 static inline int blk_integrity_rq(struct request *rq)
1748 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1753 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1755 struct scatterlist *s)
1759 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1763 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1768 blk_integrity_queue_supports_integrity(struct request_queue *q)
1772 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1776 static inline void blk_integrity_register(struct gendisk *d,
1777 struct blk_integrity *b)
1780 static inline void blk_integrity_unregister(struct gendisk *d)
1783 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1787 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1791 static inline bool blk_integrity_merge_rq(struct request_queue *rq,
1797 static inline bool blk_integrity_merge_bio(struct request_queue *rq,
1804 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1805 unsigned int sectors)
1810 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1811 unsigned int sectors)
1816 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1821 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1823 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1825 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1827 void blk_ksm_unregister(struct request_queue *q);
1829 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1831 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1832 struct request_queue *q)
1837 static inline void blk_ksm_unregister(struct request_queue *q) { }
1839 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1842 struct block_device_operations {
1843 blk_qc_t (*submit_bio) (struct bio *bio);
1844 int (*open) (struct block_device *, fmode_t);
1845 void (*release) (struct gendisk *, fmode_t);
1846 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1847 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1848 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1849 unsigned int (*check_events) (struct gendisk *disk,
1850 unsigned int clearing);
1851 void (*unlock_native_capacity) (struct gendisk *);
1852 int (*revalidate_disk) (struct gendisk *);
1853 int (*getgeo)(struct block_device *, struct hd_geometry *);
1854 /* this callback is with swap_lock and sometimes page table lock held */
1855 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1856 int (*report_zones)(struct gendisk *, sector_t sector,
1857 unsigned int nr_zones, report_zones_cb cb, void *data);
1858 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1859 struct module *owner;
1860 const struct pr_ops *pr_ops;
1863 #ifdef CONFIG_COMPAT
1864 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1865 unsigned int, unsigned long);
1867 #define blkdev_compat_ptr_ioctl NULL
1870 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1872 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1873 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1874 struct writeback_control *);
1876 #ifdef CONFIG_BLK_DEV_ZONED
1877 bool blk_req_needs_zone_write_lock(struct request *rq);
1878 bool blk_req_zone_write_trylock(struct request *rq);
1879 void __blk_req_zone_write_lock(struct request *rq);
1880 void __blk_req_zone_write_unlock(struct request *rq);
1882 static inline void blk_req_zone_write_lock(struct request *rq)
1884 if (blk_req_needs_zone_write_lock(rq))
1885 __blk_req_zone_write_lock(rq);
1888 static inline void blk_req_zone_write_unlock(struct request *rq)
1890 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1891 __blk_req_zone_write_unlock(rq);
1894 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1896 return rq->q->seq_zones_wlock &&
1897 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1900 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1902 if (!blk_req_needs_zone_write_lock(rq))
1904 return !blk_req_zone_is_write_locked(rq);
1907 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1912 static inline void blk_req_zone_write_lock(struct request *rq)
1916 static inline void blk_req_zone_write_unlock(struct request *rq)
1919 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1924 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1928 #endif /* CONFIG_BLK_DEV_ZONED */
1930 static inline void blk_wake_io_task(struct task_struct *waiter)
1933 * If we're polling, the task itself is doing the completions. For
1934 * that case, we don't need to signal a wakeup, it's enough to just
1935 * mark us as RUNNING.
1937 if (waiter == current)
1938 __set_current_state(TASK_RUNNING);
1940 wake_up_process(waiter);
1943 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1945 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1946 unsigned long start_time);
1948 unsigned long part_start_io_acct(struct gendisk *disk, struct hd_struct **part,
1950 void part_end_io_acct(struct hd_struct *part, struct bio *bio,
1951 unsigned long start_time);
1954 * bio_start_io_acct - start I/O accounting for bio based drivers
1955 * @bio: bio to start account for
1957 * Returns the start time that should be passed back to bio_end_io_acct().
1959 static inline unsigned long bio_start_io_acct(struct bio *bio)
1961 return disk_start_io_acct(bio->bi_disk, bio_sectors(bio), bio_op(bio));
1965 * bio_end_io_acct - end I/O accounting for bio based drivers
1966 * @bio: bio to end account for
1967 * @start: start time returned by bio_start_io_acct()
1969 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1971 return disk_end_io_acct(bio->bi_disk, bio_op(bio), start_time);
1974 int bdev_read_only(struct block_device *bdev);
1975 int set_blocksize(struct block_device *bdev, int size);
1977 const char *bdevname(struct block_device *bdev, char *buffer);
1978 struct block_device *lookup_bdev(const char *);
1980 void blkdev_show(struct seq_file *seqf, off_t offset);
1982 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1983 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1985 #define BLKDEV_MAJOR_MAX 512
1987 #define BLKDEV_MAJOR_MAX 0
1990 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1992 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1993 int bd_prepare_to_claim(struct block_device *bdev, struct block_device *whole,
1995 void bd_abort_claiming(struct block_device *bdev, struct block_device *whole,
1997 void blkdev_put(struct block_device *bdev, fmode_t mode);
1999 struct block_device *I_BDEV(struct inode *inode);
2000 struct block_device *bdget(dev_t);
2001 struct block_device *bdgrab(struct block_device *bdev);
2002 void bdput(struct block_device *);
2005 void invalidate_bdev(struct block_device *bdev);
2006 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
2008 int sync_blockdev(struct block_device *bdev);
2010 static inline void invalidate_bdev(struct block_device *bdev)
2013 static inline int truncate_bdev_range(struct block_device *bdev, fmode_t mode,
2014 loff_t lstart, loff_t lend)
2018 static inline int sync_blockdev(struct block_device *bdev)
2023 int fsync_bdev(struct block_device *bdev);
2025 struct super_block *freeze_bdev(struct block_device *bdev);
2026 int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2028 #endif /* _LINUX_BLKDEV_H */