1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef _LINUX_BLKDEV_H
3 #define _LINUX_BLKDEV_H
5 #include <linux/sched.h>
6 #include <linux/sched/clock.h>
7 #include <linux/major.h>
8 #include <linux/genhd.h>
9 #include <linux/list.h>
10 #include <linux/llist.h>
11 #include <linux/minmax.h>
12 #include <linux/timer.h>
13 #include <linux/workqueue.h>
14 #include <linux/pagemap.h>
15 #include <linux/backing-dev-defs.h>
16 #include <linux/wait.h>
17 #include <linux/mempool.h>
18 #include <linux/pfn.h>
19 #include <linux/bio.h>
20 #include <linux/stringify.h>
21 #include <linux/gfp.h>
22 #include <linux/bsg.h>
23 #include <linux/smp.h>
24 #include <linux/rcupdate.h>
25 #include <linux/percpu-refcount.h>
26 #include <linux/scatterlist.h>
27 #include <linux/blkzoned.h>
31 struct scsi_ioctl_command;
34 struct elevator_queue;
40 struct blk_flush_queue;
43 struct blk_queue_stats;
44 struct blk_stat_callback;
45 struct blk_keyslot_manager;
47 #define BLKDEV_MIN_RQ 4
48 #define BLKDEV_MAX_RQ 128 /* Default maximum */
50 /* Must be consistent with blk_mq_poll_stats_bkt() */
51 #define BLK_MQ_POLL_STATS_BKTS 16
53 /* Doing classic polling */
54 #define BLK_MQ_POLL_CLASSIC -1
57 * Maximum number of blkcg policies allowed to be registered concurrently.
58 * Defined here to simplify include dependency.
60 #define BLKCG_MAX_POLS 5
62 static inline int blk_validate_block_size(unsigned int bsize)
64 if (bsize < 512 || bsize > PAGE_SIZE || !is_power_of_2(bsize))
70 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
74 typedef __u32 __bitwise req_flags_t;
76 /* elevator knows about this request */
77 #define RQF_SORTED ((__force req_flags_t)(1 << 0))
78 /* drive already may have started this one */
79 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
80 /* may not be passed by ioscheduler */
81 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
82 /* request for flush sequence */
83 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
84 /* merge of different types, fail separately */
85 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
86 /* track inflight for MQ */
87 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
88 /* don't call prep for this one */
89 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
90 /* vaguely specified driver internal error. Ignored by the block layer */
91 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
92 /* don't warn about errors */
93 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
94 /* elevator private data attached */
95 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
96 /* account into disk and partition IO statistics */
97 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
98 /* request came from our alloc pool */
99 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
100 /* runtime pm request */
101 #define RQF_PM ((__force req_flags_t)(1 << 15))
102 /* on IO scheduler merge hash */
103 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
104 /* track IO completion time */
105 #define RQF_STATS ((__force req_flags_t)(1 << 17))
106 /* Look at ->special_vec for the actual data payload instead of the
108 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
109 /* The per-zone write lock is held for this request */
110 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
111 /* already slept for hybrid poll */
112 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
113 /* ->timeout has been called, don't expire again */
114 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
116 /* flags that prevent us from merging requests: */
117 #define RQF_NOMERGE_FLAGS \
118 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
121 * Request state for blk-mq.
130 * Try to put the fields that are referenced together in the same cacheline.
132 * If you modify this structure, make sure to update blk_rq_init() and
133 * especially blk_mq_rq_ctx_init() to take care of the added fields.
136 struct request_queue *q;
137 struct blk_mq_ctx *mq_ctx;
138 struct blk_mq_hw_ctx *mq_hctx;
140 unsigned int cmd_flags; /* op and common flags */
141 req_flags_t rq_flags;
146 /* the following two fields are internal, NEVER access directly */
147 unsigned int __data_len; /* total data len */
148 sector_t __sector; /* sector cursor */
153 struct list_head queuelist;
156 * The hash is used inside the scheduler, and killed once the
157 * request reaches the dispatch list. The ipi_list is only used
158 * to queue the request for softirq completion, which is long
159 * after the request has been unhashed (and even removed from
160 * the dispatch list).
163 struct hlist_node hash; /* merge hash */
164 struct list_head ipi_list;
168 * The rb_node is only used inside the io scheduler, requests
169 * are pruned when moved to the dispatch queue. So let the
170 * completion_data share space with the rb_node.
173 struct rb_node rb_node; /* sort/lookup */
174 struct bio_vec special_vec;
175 void *completion_data;
176 int error_count; /* for legacy drivers, don't use */
180 * Three pointers are available for the IO schedulers, if they need
181 * more they have to dynamically allocate it. Flush requests are
182 * never put on the IO scheduler. So let the flush fields share
183 * space with the elevator data.
193 struct list_head list;
194 rq_end_io_fn *saved_end_io;
198 struct gendisk *rq_disk;
199 struct hd_struct *part;
200 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
201 /* Time that the first bio started allocating this request. */
204 /* Time that this request was allocated for this IO. */
206 /* Time that I/O was submitted to the device. */
207 u64 io_start_time_ns;
209 #ifdef CONFIG_BLK_WBT
210 unsigned short wbt_flags;
213 * rq sectors used for blk stats. It has the same value
214 * with blk_rq_sectors(rq), except that it never be zeroed
217 unsigned short stats_sectors;
220 * Number of scatter-gather DMA addr+len pairs after
221 * physical address coalescing is performed.
223 unsigned short nr_phys_segments;
225 #if defined(CONFIG_BLK_DEV_INTEGRITY)
226 unsigned short nr_integrity_segments;
229 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
230 struct bio_crypt_ctx *crypt_ctx;
231 struct blk_ksm_keyslot *crypt_keyslot;
234 unsigned short write_hint;
235 unsigned short ioprio;
237 enum mq_rq_state state;
240 unsigned int timeout;
241 unsigned long deadline;
244 struct __call_single_data csd;
249 * completion callback.
251 rq_end_io_fn *end_io;
255 static inline bool blk_op_is_scsi(unsigned int op)
257 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
260 static inline bool blk_op_is_private(unsigned int op)
262 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
265 static inline bool blk_rq_is_scsi(struct request *rq)
267 return blk_op_is_scsi(req_op(rq));
270 static inline bool blk_rq_is_private(struct request *rq)
272 return blk_op_is_private(req_op(rq));
275 static inline bool blk_rq_is_passthrough(struct request *rq)
277 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
280 static inline bool bio_is_passthrough(struct bio *bio)
282 unsigned op = bio_op(bio);
284 return blk_op_is_scsi(op) || blk_op_is_private(op);
287 static inline unsigned short req_get_ioprio(struct request *req)
292 #include <linux/elevator.h>
294 struct blk_queue_ctx;
298 enum blk_eh_timer_return {
299 BLK_EH_DONE, /* drivers has completed the command */
300 BLK_EH_RESET_TIMER, /* reset timer and try again */
303 enum blk_queue_state {
308 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
309 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
311 #define BLK_SCSI_MAX_CMDS (256)
312 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
315 * Zoned block device models (zoned limit).
317 * Note: This needs to be ordered from the least to the most severe
318 * restrictions for the inheritance in blk_stack_limits() to work.
320 enum blk_zoned_model {
321 BLK_ZONED_NONE = 0, /* Regular block device */
322 BLK_ZONED_HA, /* Host-aware zoned block device */
323 BLK_ZONED_HM, /* Host-managed zoned block device */
326 struct queue_limits {
327 unsigned long bounce_pfn;
328 unsigned long seg_boundary_mask;
329 unsigned long virt_boundary_mask;
331 unsigned int max_hw_sectors;
332 unsigned int max_dev_sectors;
333 unsigned int chunk_sectors;
334 unsigned int max_sectors;
335 unsigned int max_segment_size;
336 unsigned int physical_block_size;
337 unsigned int logical_block_size;
338 unsigned int alignment_offset;
341 unsigned int max_discard_sectors;
342 unsigned int max_hw_discard_sectors;
343 unsigned int max_write_same_sectors;
344 unsigned int max_write_zeroes_sectors;
345 unsigned int max_zone_append_sectors;
346 unsigned int discard_granularity;
347 unsigned int discard_alignment;
348 unsigned int zone_write_granularity;
350 unsigned short max_segments;
351 unsigned short max_integrity_segments;
352 unsigned short max_discard_segments;
354 unsigned char misaligned;
355 unsigned char discard_misaligned;
356 unsigned char raid_partial_stripes_expensive;
357 enum blk_zoned_model zoned;
360 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
363 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
365 #ifdef CONFIG_BLK_DEV_ZONED
367 #define BLK_ALL_ZONES ((unsigned int)-1)
368 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
369 unsigned int nr_zones, report_zones_cb cb, void *data);
370 unsigned int blkdev_nr_zones(struct gendisk *disk);
371 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
372 sector_t sectors, sector_t nr_sectors,
374 int blk_revalidate_disk_zones(struct gendisk *disk,
375 void (*update_driver_data)(struct gendisk *disk));
377 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
378 unsigned int cmd, unsigned long arg);
379 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
380 unsigned int cmd, unsigned long arg);
382 #else /* CONFIG_BLK_DEV_ZONED */
384 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
389 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
390 fmode_t mode, unsigned int cmd,
396 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
397 fmode_t mode, unsigned int cmd,
403 #endif /* CONFIG_BLK_DEV_ZONED */
405 struct request_queue {
406 struct request *last_merge;
407 struct elevator_queue *elevator;
409 struct percpu_ref q_usage_counter;
411 struct blk_queue_stats *stats;
412 struct rq_qos *rq_qos;
414 const struct blk_mq_ops *mq_ops;
417 struct blk_mq_ctx __percpu *queue_ctx;
419 unsigned int queue_depth;
421 /* hw dispatch queues */
422 struct blk_mq_hw_ctx **queue_hw_ctx;
423 unsigned int nr_hw_queues;
425 struct backing_dev_info *backing_dev_info;
428 * The queue owner gets to use this for whatever they like.
429 * ll_rw_blk doesn't touch it.
434 * various queue flags, see QUEUE_* below
436 unsigned long queue_flags;
438 * Number of contexts that have called blk_set_pm_only(). If this
439 * counter is above zero then only RQF_PM requests are processed.
444 * ida allocated id for this queue. Used to index queues from
450 * queue needs bounce pages for pages above this limit
454 spinlock_t queue_lock;
464 struct kobject *mq_kobj;
466 #ifdef CONFIG_BLK_DEV_INTEGRITY
467 struct blk_integrity integrity;
468 #endif /* CONFIG_BLK_DEV_INTEGRITY */
472 enum rpm_status rpm_status;
473 unsigned int nr_pending;
479 unsigned long nr_requests; /* Max # of requests */
481 unsigned int dma_pad_mask;
482 unsigned int dma_alignment;
484 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
485 /* Inline crypto capabilities */
486 struct blk_keyslot_manager *ksm;
489 unsigned int rq_timeout;
492 struct blk_stat_callback *poll_cb;
493 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
495 struct timer_list timeout;
496 struct work_struct timeout_work;
498 atomic_t nr_active_requests_shared_sbitmap;
500 struct list_head icq_list;
501 #ifdef CONFIG_BLK_CGROUP
502 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
503 struct blkcg_gq *root_blkg;
504 struct list_head blkg_list;
507 struct queue_limits limits;
509 unsigned int required_elevator_features;
511 #ifdef CONFIG_BLK_DEV_ZONED
513 * Zoned block device information for request dispatch control.
514 * nr_zones is the total number of zones of the device. This is always
515 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
516 * bits which indicates if a zone is conventional (bit set) or
517 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
518 * bits which indicates if a zone is write locked, that is, if a write
519 * request targeting the zone was dispatched. All three fields are
520 * initialized by the low level device driver (e.g. scsi/sd.c).
521 * Stacking drivers (device mappers) may or may not initialize
524 * Reads of this information must be protected with blk_queue_enter() /
525 * blk_queue_exit(). Modifying this information is only allowed while
526 * no requests are being processed. See also blk_mq_freeze_queue() and
527 * blk_mq_unfreeze_queue().
529 unsigned int nr_zones;
530 unsigned long *conv_zones_bitmap;
531 unsigned long *seq_zones_wlock;
532 unsigned int max_open_zones;
533 unsigned int max_active_zones;
534 #endif /* CONFIG_BLK_DEV_ZONED */
539 unsigned int sg_timeout;
540 unsigned int sg_reserved_size;
542 struct mutex debugfs_mutex;
543 #ifdef CONFIG_BLK_DEV_IO_TRACE
544 struct blk_trace __rcu *blk_trace;
547 * for flush operations
549 struct blk_flush_queue *fq;
551 struct list_head requeue_list;
552 spinlock_t requeue_lock;
553 struct delayed_work requeue_work;
555 struct mutex sysfs_lock;
556 struct mutex sysfs_dir_lock;
559 * for reusing dead hctx instance in case of updating
562 struct list_head unused_hctx_list;
563 spinlock_t unused_hctx_lock;
567 #if defined(CONFIG_BLK_DEV_BSG)
568 struct bsg_class_device bsg_dev;
571 #ifdef CONFIG_BLK_DEV_THROTTLING
573 struct throtl_data *td;
575 struct rcu_head rcu_head;
576 wait_queue_head_t mq_freeze_wq;
578 * Protect concurrent access to q_usage_counter by
579 * percpu_ref_kill() and percpu_ref_reinit().
581 struct mutex mq_freeze_lock;
583 struct blk_mq_tag_set *tag_set;
584 struct list_head tag_set_list;
585 struct bio_set bio_split;
587 struct dentry *debugfs_dir;
589 #ifdef CONFIG_BLK_DEBUG_FS
590 struct dentry *sched_debugfs_dir;
591 struct dentry *rqos_debugfs_dir;
594 bool mq_sysfs_init_done;
598 #define BLK_MAX_WRITE_HINTS 5
599 u64 write_hints[BLK_MAX_WRITE_HINTS];
602 /* Keep blk_queue_flag_name[] in sync with the definitions below */
603 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
604 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
605 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
606 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
607 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
608 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
609 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
610 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
611 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
612 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
613 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
614 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
615 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
616 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
617 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
618 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
619 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
620 #define QUEUE_FLAG_WC 17 /* Write back caching */
621 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
622 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
623 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
624 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
625 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
626 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
627 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
628 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
629 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
630 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
631 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
632 #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */
634 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
635 (1 << QUEUE_FLAG_SAME_COMP) | \
636 (1 << QUEUE_FLAG_NOWAIT))
638 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
639 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
640 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
642 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
643 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
644 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
645 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
646 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
647 #define blk_queue_noxmerges(q) \
648 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
649 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
650 #define blk_queue_stable_writes(q) \
651 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
652 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
653 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
654 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
655 #define blk_queue_zone_resetall(q) \
656 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
657 #define blk_queue_secure_erase(q) \
658 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
659 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
660 #define blk_queue_scsi_passthrough(q) \
661 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
662 #define blk_queue_pci_p2pdma(q) \
663 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
664 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
665 #define blk_queue_rq_alloc_time(q) \
666 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
668 #define blk_queue_rq_alloc_time(q) false
671 #define blk_noretry_request(rq) \
672 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
673 REQ_FAILFAST_DRIVER))
674 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
675 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
676 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
677 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
678 #define blk_queue_nowait(q) test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
680 extern void blk_set_pm_only(struct request_queue *q);
681 extern void blk_clear_pm_only(struct request_queue *q);
683 static inline bool blk_account_rq(struct request *rq)
685 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
688 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
690 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
692 #define rq_dma_dir(rq) \
693 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
695 #define dma_map_bvec(dev, bv, dir, attrs) \
696 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
699 static inline bool queue_is_mq(struct request_queue *q)
705 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
707 return q->rpm_status;
710 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
716 static inline enum blk_zoned_model
717 blk_queue_zoned_model(struct request_queue *q)
719 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
720 return q->limits.zoned;
721 return BLK_ZONED_NONE;
724 static inline bool blk_queue_is_zoned(struct request_queue *q)
726 switch (blk_queue_zoned_model(q)) {
735 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
737 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
740 #ifdef CONFIG_BLK_DEV_ZONED
741 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
743 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
746 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
749 if (!blk_queue_is_zoned(q))
751 return sector >> ilog2(q->limits.chunk_sectors);
754 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
757 if (!blk_queue_is_zoned(q))
759 if (!q->conv_zones_bitmap)
761 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
764 static inline void blk_queue_max_open_zones(struct request_queue *q,
765 unsigned int max_open_zones)
767 q->max_open_zones = max_open_zones;
770 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
772 return q->max_open_zones;
775 static inline void blk_queue_max_active_zones(struct request_queue *q,
776 unsigned int max_active_zones)
778 q->max_active_zones = max_active_zones;
781 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
783 return q->max_active_zones;
785 #else /* CONFIG_BLK_DEV_ZONED */
786 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
790 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
795 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
800 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
804 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
808 #endif /* CONFIG_BLK_DEV_ZONED */
810 static inline bool rq_is_sync(struct request *rq)
812 return op_is_sync(rq->cmd_flags);
815 static inline bool rq_mergeable(struct request *rq)
817 if (blk_rq_is_passthrough(rq))
820 if (req_op(rq) == REQ_OP_FLUSH)
823 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
826 if (req_op(rq) == REQ_OP_ZONE_APPEND)
829 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
831 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
837 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
839 if (bio_page(a) == bio_page(b) &&
840 bio_offset(a) == bio_offset(b))
846 static inline unsigned int blk_queue_depth(struct request_queue *q)
849 return q->queue_depth;
851 return q->nr_requests;
854 extern unsigned long blk_max_low_pfn, blk_max_pfn;
857 * standard bounce addresses:
859 * BLK_BOUNCE_HIGH : bounce all highmem pages
860 * BLK_BOUNCE_ANY : don't bounce anything
861 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
864 #if BITS_PER_LONG == 32
865 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
867 #define BLK_BOUNCE_HIGH -1ULL
869 #define BLK_BOUNCE_ANY (-1ULL)
870 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
873 * default timeout for SG_IO if none specified
875 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
876 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
882 unsigned long offset;
887 struct req_iterator {
888 struct bvec_iter iter;
892 /* This should not be used directly - use rq_for_each_segment */
893 #define for_each_bio(_bio) \
894 for (; _bio; _bio = _bio->bi_next)
895 #define __rq_for_each_bio(_bio, rq) \
897 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
899 #define rq_for_each_segment(bvl, _rq, _iter) \
900 __rq_for_each_bio(_iter.bio, _rq) \
901 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
903 #define rq_for_each_bvec(bvl, _rq, _iter) \
904 __rq_for_each_bio(_iter.bio, _rq) \
905 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
907 #define rq_iter_last(bvec, _iter) \
908 (_iter.bio->bi_next == NULL && \
909 bio_iter_last(bvec, _iter.iter))
911 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
912 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
914 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
915 extern void rq_flush_dcache_pages(struct request *rq);
917 static inline void rq_flush_dcache_pages(struct request *rq)
922 extern int blk_register_queue(struct gendisk *disk);
923 extern void blk_unregister_queue(struct gendisk *disk);
924 blk_qc_t submit_bio_noacct(struct bio *bio);
925 extern void blk_rq_init(struct request_queue *q, struct request *rq);
926 extern void blk_put_request(struct request *);
927 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
928 blk_mq_req_flags_t flags);
929 extern int blk_lld_busy(struct request_queue *q);
930 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
931 struct bio_set *bs, gfp_t gfp_mask,
932 int (*bio_ctr)(struct bio *, struct bio *, void *),
934 extern void blk_rq_unprep_clone(struct request *rq);
935 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
937 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
938 extern void blk_queue_split(struct bio **);
939 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
940 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
941 unsigned int, void __user *);
942 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
943 unsigned int, void __user *);
944 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
945 struct scsi_ioctl_command __user *);
946 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
947 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
949 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
950 extern void blk_queue_exit(struct request_queue *q);
951 extern void blk_sync_queue(struct request_queue *q);
952 extern int blk_rq_map_user(struct request_queue *, struct request *,
953 struct rq_map_data *, void __user *, unsigned long,
955 extern int blk_rq_unmap_user(struct bio *);
956 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
957 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
958 struct rq_map_data *, const struct iov_iter *,
960 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
961 struct request *, int);
962 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
963 struct request *, int, rq_end_io_fn *);
965 /* Helper to convert REQ_OP_XXX to its string format XXX */
966 extern const char *blk_op_str(unsigned int op);
968 int blk_status_to_errno(blk_status_t status);
969 blk_status_t errno_to_blk_status(int errno);
971 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
973 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
975 return bdev->bd_disk->queue; /* this is never NULL */
979 * The basic unit of block I/O is a sector. It is used in a number of contexts
980 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
981 * bytes. Variables of type sector_t represent an offset or size that is a
982 * multiple of 512 bytes. Hence these two constants.
985 #define SECTOR_SHIFT 9
988 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
992 * blk_rq_pos() : the current sector
993 * blk_rq_bytes() : bytes left in the entire request
994 * blk_rq_cur_bytes() : bytes left in the current segment
995 * blk_rq_err_bytes() : bytes left till the next error boundary
996 * blk_rq_sectors() : sectors left in the entire request
997 * blk_rq_cur_sectors() : sectors left in the current segment
998 * blk_rq_stats_sectors() : sectors of the entire request used for stats
1000 static inline sector_t blk_rq_pos(const struct request *rq)
1002 return rq->__sector;
1005 static inline unsigned int blk_rq_bytes(const struct request *rq)
1007 return rq->__data_len;
1010 static inline int blk_rq_cur_bytes(const struct request *rq)
1012 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
1015 extern unsigned int blk_rq_err_bytes(const struct request *rq);
1017 static inline unsigned int blk_rq_sectors(const struct request *rq)
1019 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
1022 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
1024 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
1027 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
1029 return rq->stats_sectors;
1032 #ifdef CONFIG_BLK_DEV_ZONED
1034 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
1035 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
1037 static inline unsigned int blk_rq_zone_no(struct request *rq)
1039 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1042 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1044 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1046 #endif /* CONFIG_BLK_DEV_ZONED */
1049 * Some commands like WRITE SAME have a payload or data transfer size which
1050 * is different from the size of the request. Any driver that supports such
1051 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1052 * calculate the data transfer size.
1054 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1056 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1057 return rq->special_vec.bv_len;
1058 return blk_rq_bytes(rq);
1062 * Return the first full biovec in the request. The caller needs to check that
1063 * there are any bvecs before calling this helper.
1065 static inline struct bio_vec req_bvec(struct request *rq)
1067 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1068 return rq->special_vec;
1069 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1072 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1075 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1076 return min(q->limits.max_discard_sectors,
1077 UINT_MAX >> SECTOR_SHIFT);
1079 if (unlikely(op == REQ_OP_WRITE_SAME))
1080 return q->limits.max_write_same_sectors;
1082 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1083 return q->limits.max_write_zeroes_sectors;
1085 return q->limits.max_sectors;
1089 * Return maximum size of a request at given offset. Only valid for
1090 * file system requests.
1092 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1094 unsigned int chunk_sectors)
1096 if (!chunk_sectors) {
1097 if (q->limits.chunk_sectors)
1098 chunk_sectors = q->limits.chunk_sectors;
1100 return q->limits.max_sectors;
1103 if (likely(is_power_of_2(chunk_sectors)))
1104 chunk_sectors -= offset & (chunk_sectors - 1);
1106 chunk_sectors -= sector_div(offset, chunk_sectors);
1108 return min(q->limits.max_sectors, chunk_sectors);
1111 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1114 struct request_queue *q = rq->q;
1116 if (blk_rq_is_passthrough(rq))
1117 return q->limits.max_hw_sectors;
1119 if (!q->limits.chunk_sectors ||
1120 req_op(rq) == REQ_OP_DISCARD ||
1121 req_op(rq) == REQ_OP_SECURE_ERASE)
1122 return blk_queue_get_max_sectors(q, req_op(rq));
1124 return min(blk_max_size_offset(q, offset, 0),
1125 blk_queue_get_max_sectors(q, req_op(rq)));
1128 static inline unsigned int blk_rq_count_bios(struct request *rq)
1130 unsigned int nr_bios = 0;
1133 __rq_for_each_bio(bio, rq)
1139 void blk_steal_bios(struct bio_list *list, struct request *rq);
1142 * Request completion related functions.
1144 * blk_update_request() completes given number of bytes and updates
1145 * the request without completing it.
1147 extern bool blk_update_request(struct request *rq, blk_status_t error,
1148 unsigned int nr_bytes);
1150 extern void blk_abort_request(struct request *);
1153 * Access functions for manipulating queue properties
1155 extern void blk_cleanup_queue(struct request_queue *);
1156 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1157 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1158 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1159 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1160 extern void blk_queue_max_discard_segments(struct request_queue *,
1162 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1163 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1164 unsigned int max_discard_sectors);
1165 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1166 unsigned int max_write_same_sectors);
1167 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1168 unsigned int max_write_same_sectors);
1169 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1170 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1171 unsigned int max_zone_append_sectors);
1172 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1173 void blk_queue_zone_write_granularity(struct request_queue *q,
1175 extern void blk_queue_alignment_offset(struct request_queue *q,
1176 unsigned int alignment);
1177 void blk_queue_update_readahead(struct request_queue *q);
1178 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1179 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1180 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1181 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1182 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1183 extern void blk_set_default_limits(struct queue_limits *lim);
1184 extern void blk_set_stacking_limits(struct queue_limits *lim);
1185 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1187 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1189 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1190 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1191 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1192 extern void blk_queue_dma_alignment(struct request_queue *, int);
1193 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1194 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1195 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1196 extern void blk_queue_required_elevator_features(struct request_queue *q,
1197 unsigned int features);
1198 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1199 struct device *dev);
1202 * Number of physical segments as sent to the device.
1204 * Normally this is the number of discontiguous data segments sent by the
1205 * submitter. But for data-less command like discard we might have no
1206 * actual data segments submitted, but the driver might have to add it's
1207 * own special payload. In that case we still return 1 here so that this
1208 * special payload will be mapped.
1210 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1212 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1214 return rq->nr_phys_segments;
1218 * Number of discard segments (or ranges) the driver needs to fill in.
1219 * Each discard bio merged into a request is counted as one segment.
1221 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1223 return max_t(unsigned short, rq->nr_phys_segments, 1);
1226 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1227 struct scatterlist *sglist, struct scatterlist **last_sg);
1228 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1229 struct scatterlist *sglist)
1231 struct scatterlist *last_sg = NULL;
1233 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1235 extern void blk_dump_rq_flags(struct request *, char *);
1237 bool __must_check blk_get_queue(struct request_queue *);
1238 struct request_queue *blk_alloc_queue(int node_id);
1239 extern void blk_put_queue(struct request_queue *);
1240 extern void blk_set_queue_dying(struct request_queue *);
1244 * blk_plug permits building a queue of related requests by holding the I/O
1245 * fragments for a short period. This allows merging of sequential requests
1246 * into single larger request. As the requests are moved from a per-task list to
1247 * the device's request_queue in a batch, this results in improved scalability
1248 * as the lock contention for request_queue lock is reduced.
1250 * It is ok not to disable preemption when adding the request to the plug list
1251 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1252 * the plug list when the task sleeps by itself. For details, please see
1253 * schedule() where blk_schedule_flush_plug() is called.
1256 struct list_head mq_list; /* blk-mq requests */
1257 struct list_head cb_list; /* md requires an unplug callback */
1258 unsigned short rq_count;
1259 bool multiple_queues;
1264 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1265 struct blk_plug_cb {
1266 struct list_head list;
1267 blk_plug_cb_fn callback;
1270 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1271 void *data, int size);
1272 extern void blk_start_plug(struct blk_plug *);
1273 extern void blk_finish_plug(struct blk_plug *);
1274 extern void blk_flush_plug_list(struct blk_plug *, bool);
1276 static inline void blk_flush_plug(struct task_struct *tsk)
1278 struct blk_plug *plug = tsk->plug;
1281 blk_flush_plug_list(plug, false);
1284 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1286 struct blk_plug *plug = tsk->plug;
1289 blk_flush_plug_list(plug, true);
1292 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1294 struct blk_plug *plug = tsk->plug;
1297 (!list_empty(&plug->mq_list) ||
1298 !list_empty(&plug->cb_list));
1301 int blkdev_issue_flush(struct block_device *, gfp_t);
1302 long nr_blockdev_pages(void);
1303 #else /* CONFIG_BLOCK */
1307 static inline void blk_start_plug(struct blk_plug *plug)
1311 static inline void blk_finish_plug(struct blk_plug *plug)
1315 static inline void blk_flush_plug(struct task_struct *task)
1319 static inline void blk_schedule_flush_plug(struct task_struct *task)
1324 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1329 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask)
1334 static inline long nr_blockdev_pages(void)
1338 #endif /* CONFIG_BLOCK */
1340 extern void blk_io_schedule(void);
1342 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1343 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1345 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1347 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1348 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1349 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1350 sector_t nr_sects, gfp_t gfp_mask, int flags,
1353 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1354 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1356 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1357 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1359 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1360 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1362 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1363 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1365 return blkdev_issue_discard(sb->s_bdev,
1366 block << (sb->s_blocksize_bits -
1368 nr_blocks << (sb->s_blocksize_bits -
1372 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1373 sector_t nr_blocks, gfp_t gfp_mask)
1375 return blkdev_issue_zeroout(sb->s_bdev,
1376 block << (sb->s_blocksize_bits -
1378 nr_blocks << (sb->s_blocksize_bits -
1383 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1385 static inline bool bdev_is_partition(struct block_device *bdev)
1387 return bdev->bd_partno;
1390 enum blk_default_limits {
1391 BLK_MAX_SEGMENTS = 128,
1392 BLK_SAFE_MAX_SECTORS = 255,
1393 BLK_DEF_MAX_SECTORS = 2560,
1394 BLK_MAX_SEGMENT_SIZE = 65536,
1395 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1398 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1400 return q->limits.seg_boundary_mask;
1403 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1405 return q->limits.virt_boundary_mask;
1408 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1410 return q->limits.max_sectors;
1413 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1415 return q->limits.max_hw_sectors;
1418 static inline unsigned short queue_max_segments(const struct request_queue *q)
1420 return q->limits.max_segments;
1423 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1425 return q->limits.max_discard_segments;
1428 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1430 return q->limits.max_segment_size;
1433 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1436 const struct queue_limits *l = &q->limits;
1438 return min(l->max_zone_append_sectors, l->max_sectors);
1441 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1445 if (q && q->limits.logical_block_size)
1446 retval = q->limits.logical_block_size;
1451 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1453 return queue_logical_block_size(bdev_get_queue(bdev));
1456 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1458 return q->limits.physical_block_size;
1461 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1463 return queue_physical_block_size(bdev_get_queue(bdev));
1466 static inline unsigned int queue_io_min(const struct request_queue *q)
1468 return q->limits.io_min;
1471 static inline int bdev_io_min(struct block_device *bdev)
1473 return queue_io_min(bdev_get_queue(bdev));
1476 static inline unsigned int queue_io_opt(const struct request_queue *q)
1478 return q->limits.io_opt;
1481 static inline int bdev_io_opt(struct block_device *bdev)
1483 return queue_io_opt(bdev_get_queue(bdev));
1486 static inline unsigned int
1487 queue_zone_write_granularity(const struct request_queue *q)
1489 return q->limits.zone_write_granularity;
1492 static inline unsigned int
1493 bdev_zone_write_granularity(struct block_device *bdev)
1495 return queue_zone_write_granularity(bdev_get_queue(bdev));
1498 static inline int queue_alignment_offset(const struct request_queue *q)
1500 if (q->limits.misaligned)
1503 return q->limits.alignment_offset;
1506 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1508 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1509 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1512 return (granularity + lim->alignment_offset - alignment) % granularity;
1515 static inline int bdev_alignment_offset(struct block_device *bdev)
1517 struct request_queue *q = bdev_get_queue(bdev);
1519 if (q->limits.misaligned)
1521 if (bdev_is_partition(bdev))
1522 return queue_limit_alignment_offset(&q->limits,
1523 bdev->bd_part->start_sect);
1524 return q->limits.alignment_offset;
1527 static inline int queue_discard_alignment(const struct request_queue *q)
1529 if (q->limits.discard_misaligned)
1532 return q->limits.discard_alignment;
1535 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1537 unsigned int alignment, granularity, offset;
1539 if (!lim->max_discard_sectors)
1542 /* Why are these in bytes, not sectors? */
1543 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1544 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1548 /* Offset of the partition start in 'granularity' sectors */
1549 offset = sector_div(sector, granularity);
1551 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1552 offset = (granularity + alignment - offset) % granularity;
1554 /* Turn it back into bytes, gaah */
1555 return offset << SECTOR_SHIFT;
1559 * Two cases of handling DISCARD merge:
1560 * If max_discard_segments > 1, the driver takes every bio
1561 * as a range and send them to controller together. The ranges
1562 * needn't to be contiguous.
1563 * Otherwise, the bios/requests will be handled as same as
1564 * others which should be contiguous.
1566 static inline bool blk_discard_mergable(struct request *req)
1568 if (req_op(req) == REQ_OP_DISCARD &&
1569 queue_max_discard_segments(req->q) > 1)
1574 static inline int bdev_discard_alignment(struct block_device *bdev)
1576 struct request_queue *q = bdev_get_queue(bdev);
1578 if (bdev_is_partition(bdev))
1579 return queue_limit_discard_alignment(&q->limits,
1580 bdev->bd_part->start_sect);
1581 return q->limits.discard_alignment;
1584 static inline unsigned int bdev_write_same(struct block_device *bdev)
1586 struct request_queue *q = bdev_get_queue(bdev);
1589 return q->limits.max_write_same_sectors;
1594 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1596 struct request_queue *q = bdev_get_queue(bdev);
1599 return q->limits.max_write_zeroes_sectors;
1604 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1606 struct request_queue *q = bdev_get_queue(bdev);
1609 return blk_queue_zoned_model(q);
1611 return BLK_ZONED_NONE;
1614 static inline bool bdev_is_zoned(struct block_device *bdev)
1616 struct request_queue *q = bdev_get_queue(bdev);
1619 return blk_queue_is_zoned(q);
1624 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1626 struct request_queue *q = bdev_get_queue(bdev);
1629 return blk_queue_zone_sectors(q);
1633 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1635 struct request_queue *q = bdev_get_queue(bdev);
1638 return queue_max_open_zones(q);
1642 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1644 struct request_queue *q = bdev_get_queue(bdev);
1647 return queue_max_active_zones(q);
1651 static inline int queue_dma_alignment(const struct request_queue *q)
1653 return q ? q->dma_alignment : 511;
1656 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1659 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1660 return !(addr & alignment) && !(len & alignment);
1663 /* assumes size > 256 */
1664 static inline unsigned int blksize_bits(unsigned int size)
1666 unsigned int bits = 8;
1670 } while (size > 256);
1674 static inline unsigned int block_size(struct block_device *bdev)
1676 return 1 << bdev->bd_inode->i_blkbits;
1679 int kblockd_schedule_work(struct work_struct *work);
1680 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1682 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1683 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1684 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1685 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1687 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1689 enum blk_integrity_flags {
1690 BLK_INTEGRITY_VERIFY = 1 << 0,
1691 BLK_INTEGRITY_GENERATE = 1 << 1,
1692 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1693 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1696 struct blk_integrity_iter {
1700 unsigned int data_size;
1701 unsigned short interval;
1702 const char *disk_name;
1705 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1706 typedef void (integrity_prepare_fn) (struct request *);
1707 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1709 struct blk_integrity_profile {
1710 integrity_processing_fn *generate_fn;
1711 integrity_processing_fn *verify_fn;
1712 integrity_prepare_fn *prepare_fn;
1713 integrity_complete_fn *complete_fn;
1717 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1718 extern void blk_integrity_unregister(struct gendisk *);
1719 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1720 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1721 struct scatterlist *);
1722 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1724 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1726 struct blk_integrity *bi = &disk->queue->integrity;
1735 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1737 return blk_get_integrity(bdev->bd_disk);
1741 blk_integrity_queue_supports_integrity(struct request_queue *q)
1743 return q->integrity.profile;
1746 static inline bool blk_integrity_rq(struct request *rq)
1748 return rq->cmd_flags & REQ_INTEGRITY;
1751 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1754 q->limits.max_integrity_segments = segs;
1757 static inline unsigned short
1758 queue_max_integrity_segments(const struct request_queue *q)
1760 return q->limits.max_integrity_segments;
1764 * bio_integrity_intervals - Return number of integrity intervals for a bio
1765 * @bi: blk_integrity profile for device
1766 * @sectors: Size of the bio in 512-byte sectors
1768 * Description: The block layer calculates everything in 512 byte
1769 * sectors but integrity metadata is done in terms of the data integrity
1770 * interval size of the storage device. Convert the block layer sectors
1771 * to the appropriate number of integrity intervals.
1773 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1774 unsigned int sectors)
1776 return sectors >> (bi->interval_exp - 9);
1779 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1780 unsigned int sectors)
1782 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1786 * Return the first bvec that contains integrity data. Only drivers that are
1787 * limited to a single integrity segment should use this helper.
1789 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1791 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1793 return rq->bio->bi_integrity->bip_vec;
1796 #else /* CONFIG_BLK_DEV_INTEGRITY */
1799 struct block_device;
1801 struct blk_integrity;
1803 static inline int blk_integrity_rq(struct request *rq)
1807 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1812 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1814 struct scatterlist *s)
1818 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1822 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1827 blk_integrity_queue_supports_integrity(struct request_queue *q)
1831 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1835 static inline void blk_integrity_register(struct gendisk *d,
1836 struct blk_integrity *b)
1839 static inline void blk_integrity_unregister(struct gendisk *d)
1842 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1846 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1851 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1852 unsigned int sectors)
1857 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1858 unsigned int sectors)
1863 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1868 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1870 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1872 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1874 void blk_ksm_unregister(struct request_queue *q);
1876 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1878 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1879 struct request_queue *q)
1884 static inline void blk_ksm_unregister(struct request_queue *q) { }
1886 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1889 struct block_device_operations {
1890 blk_qc_t (*submit_bio) (struct bio *bio);
1891 int (*open) (struct block_device *, fmode_t);
1892 void (*release) (struct gendisk *, fmode_t);
1893 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1894 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1895 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1896 unsigned int (*check_events) (struct gendisk *disk,
1897 unsigned int clearing);
1898 void (*unlock_native_capacity) (struct gendisk *);
1899 int (*revalidate_disk) (struct gendisk *);
1900 int (*getgeo)(struct block_device *, struct hd_geometry *);
1901 int (*set_read_only)(struct block_device *bdev, bool ro);
1902 /* this callback is with swap_lock and sometimes page table lock held */
1903 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1904 int (*report_zones)(struct gendisk *, sector_t sector,
1905 unsigned int nr_zones, report_zones_cb cb, void *data);
1906 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1907 struct module *owner;
1908 const struct pr_ops *pr_ops;
1911 #ifdef CONFIG_COMPAT
1912 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1913 unsigned int, unsigned long);
1915 #define blkdev_compat_ptr_ioctl NULL
1918 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1920 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1921 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1922 struct writeback_control *);
1924 #ifdef CONFIG_BLK_DEV_ZONED
1925 bool blk_req_needs_zone_write_lock(struct request *rq);
1926 bool blk_req_zone_write_trylock(struct request *rq);
1927 void __blk_req_zone_write_lock(struct request *rq);
1928 void __blk_req_zone_write_unlock(struct request *rq);
1930 static inline void blk_req_zone_write_lock(struct request *rq)
1932 if (blk_req_needs_zone_write_lock(rq))
1933 __blk_req_zone_write_lock(rq);
1936 static inline void blk_req_zone_write_unlock(struct request *rq)
1938 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1939 __blk_req_zone_write_unlock(rq);
1942 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1944 return rq->q->seq_zones_wlock &&
1945 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1948 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1950 if (!blk_req_needs_zone_write_lock(rq))
1952 return !blk_req_zone_is_write_locked(rq);
1955 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1960 static inline void blk_req_zone_write_lock(struct request *rq)
1964 static inline void blk_req_zone_write_unlock(struct request *rq)
1967 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1972 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1976 #endif /* CONFIG_BLK_DEV_ZONED */
1978 static inline void blk_wake_io_task(struct task_struct *waiter)
1981 * If we're polling, the task itself is doing the completions. For
1982 * that case, we don't need to signal a wakeup, it's enough to just
1983 * mark us as RUNNING.
1985 if (waiter == current)
1986 __set_current_state(TASK_RUNNING);
1988 wake_up_process(waiter);
1991 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1993 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1994 unsigned long start_time);
1996 unsigned long part_start_io_acct(struct gendisk *disk, struct hd_struct **part,
1998 void part_end_io_acct(struct hd_struct *part, struct bio *bio,
1999 unsigned long start_time);
2002 * bio_start_io_acct - start I/O accounting for bio based drivers
2003 * @bio: bio to start account for
2005 * Returns the start time that should be passed back to bio_end_io_acct().
2007 static inline unsigned long bio_start_io_acct(struct bio *bio)
2009 return disk_start_io_acct(bio->bi_disk, bio_sectors(bio), bio_op(bio));
2013 * bio_end_io_acct - end I/O accounting for bio based drivers
2014 * @bio: bio to end account for
2015 * @start: start time returned by bio_start_io_acct()
2017 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
2019 return disk_end_io_acct(bio->bi_disk, bio_op(bio), start_time);
2022 int bdev_read_only(struct block_device *bdev);
2023 int set_blocksize(struct block_device *bdev, int size);
2025 const char *bdevname(struct block_device *bdev, char *buffer);
2026 struct block_device *lookup_bdev(const char *);
2028 void blkdev_show(struct seq_file *seqf, off_t offset);
2030 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2031 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2033 #define BLKDEV_MAJOR_MAX 512
2035 #define BLKDEV_MAJOR_MAX 0
2038 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2040 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
2041 int bd_prepare_to_claim(struct block_device *bdev, struct block_device *whole,
2043 void bd_abort_claiming(struct block_device *bdev, struct block_device *whole,
2045 void blkdev_put(struct block_device *bdev, fmode_t mode);
2047 struct block_device *I_BDEV(struct inode *inode);
2048 struct block_device *bdget_part(struct hd_struct *part);
2049 struct block_device *bdgrab(struct block_device *bdev);
2050 void bdput(struct block_device *);
2053 void invalidate_bdev(struct block_device *bdev);
2054 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
2056 int sync_blockdev(struct block_device *bdev);
2058 static inline void invalidate_bdev(struct block_device *bdev)
2061 static inline int truncate_bdev_range(struct block_device *bdev, fmode_t mode,
2062 loff_t lstart, loff_t lend)
2066 static inline int sync_blockdev(struct block_device *bdev)
2071 int fsync_bdev(struct block_device *bdev);
2073 struct super_block *freeze_bdev(struct block_device *bdev);
2074 int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2076 #endif /* _LINUX_BLKDEV_H */