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 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
66 typedef __u32 __bitwise req_flags_t;
68 /* drive already may have started this one */
69 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
70 /* may not be passed by ioscheduler */
71 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
72 /* request for flush sequence */
73 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
74 /* merge of different types, fail separately */
75 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
76 /* track inflight for MQ */
77 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
78 /* don't call prep for this one */
79 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
80 /* vaguely specified driver internal error. Ignored by the block layer */
81 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
82 /* don't warn about errors */
83 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
84 /* elevator private data attached */
85 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
86 /* account into disk and partition IO statistics */
87 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
88 /* request came from our alloc pool */
89 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
90 /* runtime pm request */
91 #define RQF_PM ((__force req_flags_t)(1 << 15))
92 /* on IO scheduler merge hash */
93 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
94 /* track IO completion time */
95 #define RQF_STATS ((__force req_flags_t)(1 << 17))
96 /* Look at ->special_vec for the actual data payload instead of the
98 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
99 /* The per-zone write lock is held for this request */
100 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
101 /* already slept for hybrid poll */
102 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
103 /* ->timeout has been called, don't expire again */
104 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
106 /* flags that prevent us from merging requests: */
107 #define RQF_NOMERGE_FLAGS \
108 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
111 * Request state for blk-mq.
120 * Try to put the fields that are referenced together in the same cacheline.
122 * If you modify this structure, make sure to update blk_rq_init() and
123 * especially blk_mq_rq_ctx_init() to take care of the added fields.
126 struct request_queue *q;
127 struct blk_mq_ctx *mq_ctx;
128 struct blk_mq_hw_ctx *mq_hctx;
130 unsigned int cmd_flags; /* op and common flags */
131 req_flags_t rq_flags;
136 /* the following two fields are internal, NEVER access directly */
137 unsigned int __data_len; /* total data len */
138 sector_t __sector; /* sector cursor */
143 struct list_head queuelist;
146 * The hash is used inside the scheduler, and killed once the
147 * request reaches the dispatch list. The ipi_list is only used
148 * to queue the request for softirq completion, which is long
149 * after the request has been unhashed (and even removed from
150 * the dispatch list).
153 struct hlist_node hash; /* merge hash */
154 struct llist_node ipi_list;
158 * The rb_node is only used inside the io scheduler, requests
159 * are pruned when moved to the dispatch queue. So let the
160 * completion_data share space with the rb_node.
163 struct rb_node rb_node; /* sort/lookup */
164 struct bio_vec special_vec;
165 void *completion_data;
166 int error_count; /* for legacy drivers, don't use */
170 * Three pointers are available for the IO schedulers, if they need
171 * more they have to dynamically allocate it. Flush requests are
172 * never put on the IO scheduler. So let the flush fields share
173 * space with the elevator data.
183 struct list_head list;
184 rq_end_io_fn *saved_end_io;
188 struct gendisk *rq_disk;
189 struct block_device *part;
190 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
191 /* Time that the first bio started allocating this request. */
194 /* Time that this request was allocated for this IO. */
196 /* Time that I/O was submitted to the device. */
197 u64 io_start_time_ns;
199 #ifdef CONFIG_BLK_WBT
200 unsigned short wbt_flags;
203 * rq sectors used for blk stats. It has the same value
204 * with blk_rq_sectors(rq), except that it never be zeroed
207 unsigned short stats_sectors;
210 * Number of scatter-gather DMA addr+len pairs after
211 * physical address coalescing is performed.
213 unsigned short nr_phys_segments;
215 #if defined(CONFIG_BLK_DEV_INTEGRITY)
216 unsigned short nr_integrity_segments;
219 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
220 struct bio_crypt_ctx *crypt_ctx;
221 struct blk_ksm_keyslot *crypt_keyslot;
224 unsigned short write_hint;
225 unsigned short ioprio;
227 enum mq_rq_state state;
230 unsigned int timeout;
231 unsigned long deadline;
234 struct __call_single_data csd;
239 * completion callback.
241 rq_end_io_fn *end_io;
245 static inline bool blk_op_is_scsi(unsigned int op)
247 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
250 static inline bool blk_op_is_private(unsigned int op)
252 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
255 static inline bool blk_rq_is_scsi(struct request *rq)
257 return blk_op_is_scsi(req_op(rq));
260 static inline bool blk_rq_is_private(struct request *rq)
262 return blk_op_is_private(req_op(rq));
265 static inline bool blk_rq_is_passthrough(struct request *rq)
267 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
270 static inline bool bio_is_passthrough(struct bio *bio)
272 unsigned op = bio_op(bio);
274 return blk_op_is_scsi(op) || blk_op_is_private(op);
277 static inline unsigned short req_get_ioprio(struct request *req)
282 #include <linux/elevator.h>
284 struct blk_queue_ctx;
288 enum blk_eh_timer_return {
289 BLK_EH_DONE, /* drivers has completed the command */
290 BLK_EH_RESET_TIMER, /* reset timer and try again */
293 enum blk_queue_state {
298 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
299 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
301 #define BLK_SCSI_MAX_CMDS (256)
302 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
305 * Zoned block device models (zoned limit).
307 * Note: This needs to be ordered from the least to the most severe
308 * restrictions for the inheritance in blk_stack_limits() to work.
310 enum blk_zoned_model {
311 BLK_ZONED_NONE = 0, /* Regular block device */
312 BLK_ZONED_HA, /* Host-aware zoned block device */
313 BLK_ZONED_HM, /* Host-managed zoned block device */
316 struct queue_limits {
317 unsigned long bounce_pfn;
318 unsigned long seg_boundary_mask;
319 unsigned long virt_boundary_mask;
321 unsigned int max_hw_sectors;
322 unsigned int max_dev_sectors;
323 unsigned int chunk_sectors;
324 unsigned int max_sectors;
325 unsigned int max_segment_size;
326 unsigned int physical_block_size;
327 unsigned int logical_block_size;
328 unsigned int alignment_offset;
331 unsigned int max_discard_sectors;
332 unsigned int max_hw_discard_sectors;
333 unsigned int max_write_same_sectors;
334 unsigned int max_write_zeroes_sectors;
335 unsigned int max_zone_append_sectors;
336 unsigned int discard_granularity;
337 unsigned int discard_alignment;
338 unsigned int zone_write_granularity;
340 unsigned short max_segments;
341 unsigned short max_integrity_segments;
342 unsigned short max_discard_segments;
344 unsigned char misaligned;
345 unsigned char discard_misaligned;
346 unsigned char raid_partial_stripes_expensive;
347 enum blk_zoned_model zoned;
350 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
353 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
355 #ifdef CONFIG_BLK_DEV_ZONED
357 #define BLK_ALL_ZONES ((unsigned int)-1)
358 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
359 unsigned int nr_zones, report_zones_cb cb, void *data);
360 unsigned int blkdev_nr_zones(struct gendisk *disk);
361 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
362 sector_t sectors, sector_t nr_sectors,
364 int blk_revalidate_disk_zones(struct gendisk *disk,
365 void (*update_driver_data)(struct gendisk *disk));
367 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
368 unsigned int cmd, unsigned long arg);
369 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
370 unsigned int cmd, unsigned long arg);
372 #else /* CONFIG_BLK_DEV_ZONED */
374 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
379 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
380 fmode_t mode, unsigned int cmd,
386 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
387 fmode_t mode, unsigned int cmd,
393 #endif /* CONFIG_BLK_DEV_ZONED */
395 struct request_queue {
396 struct request *last_merge;
397 struct elevator_queue *elevator;
399 struct percpu_ref q_usage_counter;
401 struct blk_queue_stats *stats;
402 struct rq_qos *rq_qos;
404 const struct blk_mq_ops *mq_ops;
407 struct blk_mq_ctx __percpu *queue_ctx;
409 unsigned int queue_depth;
411 /* hw dispatch queues */
412 struct blk_mq_hw_ctx **queue_hw_ctx;
413 unsigned int nr_hw_queues;
415 struct backing_dev_info *backing_dev_info;
418 * The queue owner gets to use this for whatever they like.
419 * ll_rw_blk doesn't touch it.
424 * various queue flags, see QUEUE_* below
426 unsigned long queue_flags;
428 * Number of contexts that have called blk_set_pm_only(). If this
429 * counter is above zero then only RQF_PM requests are processed.
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;
468 unsigned long nr_requests; /* Max # of requests */
470 unsigned int dma_pad_mask;
471 unsigned int dma_alignment;
473 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
474 /* Inline crypto capabilities */
475 struct blk_keyslot_manager *ksm;
478 unsigned int rq_timeout;
481 struct blk_stat_callback *poll_cb;
482 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
484 struct timer_list timeout;
485 struct work_struct timeout_work;
487 atomic_t nr_active_requests_shared_sbitmap;
489 struct list_head icq_list;
490 #ifdef CONFIG_BLK_CGROUP
491 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
492 struct blkcg_gq *root_blkg;
493 struct list_head blkg_list;
496 struct queue_limits limits;
498 unsigned int required_elevator_features;
500 #ifdef CONFIG_BLK_DEV_ZONED
502 * Zoned block device information for request dispatch control.
503 * nr_zones is the total number of zones of the device. This is always
504 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
505 * bits which indicates if a zone is conventional (bit set) or
506 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
507 * bits which indicates if a zone is write locked, that is, if a write
508 * request targeting the zone was dispatched. All three fields are
509 * initialized by the low level device driver (e.g. scsi/sd.c).
510 * Stacking drivers (device mappers) may or may not initialize
513 * Reads of this information must be protected with blk_queue_enter() /
514 * blk_queue_exit(). Modifying this information is only allowed while
515 * no requests are being processed. See also blk_mq_freeze_queue() and
516 * blk_mq_unfreeze_queue().
518 unsigned int nr_zones;
519 unsigned long *conv_zones_bitmap;
520 unsigned long *seq_zones_wlock;
521 unsigned int max_open_zones;
522 unsigned int max_active_zones;
523 #endif /* CONFIG_BLK_DEV_ZONED */
528 unsigned int sg_timeout;
529 unsigned int sg_reserved_size;
531 struct mutex debugfs_mutex;
532 #ifdef CONFIG_BLK_DEV_IO_TRACE
533 struct blk_trace __rcu *blk_trace;
536 * for flush operations
538 struct blk_flush_queue *fq;
540 struct list_head requeue_list;
541 spinlock_t requeue_lock;
542 struct delayed_work requeue_work;
544 struct mutex sysfs_lock;
545 struct mutex sysfs_dir_lock;
548 * for reusing dead hctx instance in case of updating
551 struct list_head unused_hctx_list;
552 spinlock_t unused_hctx_lock;
556 #if defined(CONFIG_BLK_DEV_BSG)
557 struct bsg_class_device bsg_dev;
560 #ifdef CONFIG_BLK_DEV_THROTTLING
562 struct throtl_data *td;
564 struct rcu_head rcu_head;
565 wait_queue_head_t mq_freeze_wq;
567 * Protect concurrent access to q_usage_counter by
568 * percpu_ref_kill() and percpu_ref_reinit().
570 struct mutex mq_freeze_lock;
572 struct blk_mq_tag_set *tag_set;
573 struct list_head tag_set_list;
574 struct bio_set bio_split;
576 struct dentry *debugfs_dir;
578 #ifdef CONFIG_BLK_DEBUG_FS
579 struct dentry *sched_debugfs_dir;
580 struct dentry *rqos_debugfs_dir;
583 bool mq_sysfs_init_done;
587 #define BLK_MAX_WRITE_HINTS 5
588 u64 write_hints[BLK_MAX_WRITE_HINTS];
591 /* Keep blk_queue_flag_name[] in sync with the definitions below */
592 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
593 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
594 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
595 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
596 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
597 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
598 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
599 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
600 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
601 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
602 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
603 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
604 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
605 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
606 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
607 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
608 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
609 #define QUEUE_FLAG_WC 17 /* Write back caching */
610 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
611 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
612 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
613 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
614 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
615 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
616 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
617 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
618 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
619 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
620 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
621 #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */
623 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
624 (1 << QUEUE_FLAG_SAME_COMP) | \
625 (1 << QUEUE_FLAG_NOWAIT))
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)
667 #define blk_queue_nowait(q) test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
669 extern void blk_set_pm_only(struct request_queue *q);
670 extern void blk_clear_pm_only(struct request_queue *q);
672 static inline bool blk_account_rq(struct request *rq)
674 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
677 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
679 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
681 #define rq_dma_dir(rq) \
682 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
684 #define dma_map_bvec(dev, bv, dir, attrs) \
685 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
688 static inline bool queue_is_mq(struct request_queue *q)
694 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
696 return q->rpm_status;
699 static inline enum rpm_status queue_rpm_status(struct request_queue *q)
705 static inline enum blk_zoned_model
706 blk_queue_zoned_model(struct request_queue *q)
708 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
709 return q->limits.zoned;
710 return BLK_ZONED_NONE;
713 static inline bool blk_queue_is_zoned(struct request_queue *q)
715 switch (blk_queue_zoned_model(q)) {
724 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
726 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
729 #ifdef CONFIG_BLK_DEV_ZONED
730 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
732 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
735 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
738 if (!blk_queue_is_zoned(q))
740 return sector >> ilog2(q->limits.chunk_sectors);
743 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
746 if (!blk_queue_is_zoned(q))
748 if (!q->conv_zones_bitmap)
750 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
753 static inline void blk_queue_max_open_zones(struct request_queue *q,
754 unsigned int max_open_zones)
756 q->max_open_zones = max_open_zones;
759 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
761 return q->max_open_zones;
764 static inline void blk_queue_max_active_zones(struct request_queue *q,
765 unsigned int max_active_zones)
767 q->max_active_zones = max_active_zones;
770 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
772 return q->max_active_zones;
774 #else /* CONFIG_BLK_DEV_ZONED */
775 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
779 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
784 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
789 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
793 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
797 #endif /* CONFIG_BLK_DEV_ZONED */
799 static inline bool rq_is_sync(struct request *rq)
801 return op_is_sync(rq->cmd_flags);
804 static inline bool rq_mergeable(struct request *rq)
806 if (blk_rq_is_passthrough(rq))
809 if (req_op(rq) == REQ_OP_FLUSH)
812 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
815 if (req_op(rq) == REQ_OP_ZONE_APPEND)
818 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
820 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
826 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
828 if (bio_page(a) == bio_page(b) &&
829 bio_offset(a) == bio_offset(b))
835 static inline unsigned int blk_queue_depth(struct request_queue *q)
838 return q->queue_depth;
840 return q->nr_requests;
843 extern unsigned long blk_max_low_pfn, blk_max_pfn;
846 * standard bounce addresses:
848 * BLK_BOUNCE_HIGH : bounce all highmem pages
849 * BLK_BOUNCE_ANY : don't bounce anything
850 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
853 #if BITS_PER_LONG == 32
854 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
856 #define BLK_BOUNCE_HIGH -1ULL
858 #define BLK_BOUNCE_ANY (-1ULL)
859 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
862 * default timeout for SG_IO if none specified
864 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
865 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
871 unsigned long offset;
876 struct req_iterator {
877 struct bvec_iter iter;
881 /* This should not be used directly - use rq_for_each_segment */
882 #define for_each_bio(_bio) \
883 for (; _bio; _bio = _bio->bi_next)
884 #define __rq_for_each_bio(_bio, rq) \
886 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
888 #define rq_for_each_segment(bvl, _rq, _iter) \
889 __rq_for_each_bio(_iter.bio, _rq) \
890 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
892 #define rq_for_each_bvec(bvl, _rq, _iter) \
893 __rq_for_each_bio(_iter.bio, _rq) \
894 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
896 #define rq_iter_last(bvec, _iter) \
897 (_iter.bio->bi_next == NULL && \
898 bio_iter_last(bvec, _iter.iter))
900 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
901 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
903 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
904 extern void rq_flush_dcache_pages(struct request *rq);
906 static inline void rq_flush_dcache_pages(struct request *rq)
911 extern int blk_register_queue(struct gendisk *disk);
912 extern void blk_unregister_queue(struct gendisk *disk);
913 blk_qc_t submit_bio_noacct(struct bio *bio);
914 extern void blk_rq_init(struct request_queue *q, struct request *rq);
915 extern void blk_put_request(struct request *);
916 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
917 blk_mq_req_flags_t flags);
918 extern int blk_lld_busy(struct request_queue *q);
919 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
920 struct bio_set *bs, gfp_t gfp_mask,
921 int (*bio_ctr)(struct bio *, struct bio *, void *),
923 extern void blk_rq_unprep_clone(struct request *rq);
924 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
926 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
927 extern void blk_queue_split(struct bio **);
928 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
929 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
930 unsigned int, void __user *);
931 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
932 unsigned int, void __user *);
933 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
934 struct scsi_ioctl_command __user *);
935 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
936 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
938 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
939 extern void blk_queue_exit(struct request_queue *q);
940 extern void blk_sync_queue(struct request_queue *q);
941 extern int blk_rq_map_user(struct request_queue *, struct request *,
942 struct rq_map_data *, void __user *, unsigned long,
944 extern int blk_rq_unmap_user(struct bio *);
945 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
946 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
947 struct rq_map_data *, const struct iov_iter *,
949 extern void blk_execute_rq(struct gendisk *, struct request *, int);
950 extern void blk_execute_rq_nowait(struct gendisk *,
951 struct request *, int, rq_end_io_fn *);
953 /* Helper to convert REQ_OP_XXX to its string format XXX */
954 extern const char *blk_op_str(unsigned int op);
956 int blk_status_to_errno(blk_status_t status);
957 blk_status_t errno_to_blk_status(int errno);
959 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
961 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
963 return bdev->bd_disk->queue; /* this is never NULL */
967 * The basic unit of block I/O is a sector. It is used in a number of contexts
968 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
969 * bytes. Variables of type sector_t represent an offset or size that is a
970 * multiple of 512 bytes. Hence these two constants.
973 #define SECTOR_SHIFT 9
976 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
980 * blk_rq_pos() : the current sector
981 * blk_rq_bytes() : bytes left in the entire request
982 * blk_rq_cur_bytes() : bytes left in the current segment
983 * blk_rq_err_bytes() : bytes left till the next error boundary
984 * blk_rq_sectors() : sectors left in the entire request
985 * blk_rq_cur_sectors() : sectors left in the current segment
986 * blk_rq_stats_sectors() : sectors of the entire request used for stats
988 static inline sector_t blk_rq_pos(const struct request *rq)
993 static inline unsigned int blk_rq_bytes(const struct request *rq)
995 return rq->__data_len;
998 static inline int blk_rq_cur_bytes(const struct request *rq)
1000 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
1003 extern unsigned int blk_rq_err_bytes(const struct request *rq);
1005 static inline unsigned int blk_rq_sectors(const struct request *rq)
1007 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
1010 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
1012 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
1015 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
1017 return rq->stats_sectors;
1020 #ifdef CONFIG_BLK_DEV_ZONED
1022 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
1023 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
1025 static inline unsigned int blk_rq_zone_no(struct request *rq)
1027 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1030 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1032 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1034 #endif /* CONFIG_BLK_DEV_ZONED */
1037 * Some commands like WRITE SAME have a payload or data transfer size which
1038 * is different from the size of the request. Any driver that supports such
1039 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1040 * calculate the data transfer size.
1042 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1044 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1045 return rq->special_vec.bv_len;
1046 return blk_rq_bytes(rq);
1050 * Return the first full biovec in the request. The caller needs to check that
1051 * there are any bvecs before calling this helper.
1053 static inline struct bio_vec req_bvec(struct request *rq)
1055 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1056 return rq->special_vec;
1057 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1060 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1063 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1064 return min(q->limits.max_discard_sectors,
1065 UINT_MAX >> SECTOR_SHIFT);
1067 if (unlikely(op == REQ_OP_WRITE_SAME))
1068 return q->limits.max_write_same_sectors;
1070 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1071 return q->limits.max_write_zeroes_sectors;
1073 return q->limits.max_sectors;
1077 * Return maximum size of a request at given offset. Only valid for
1078 * file system requests.
1080 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1082 unsigned int chunk_sectors)
1084 if (!chunk_sectors) {
1085 if (q->limits.chunk_sectors)
1086 chunk_sectors = q->limits.chunk_sectors;
1088 return q->limits.max_sectors;
1091 if (likely(is_power_of_2(chunk_sectors)))
1092 chunk_sectors -= offset & (chunk_sectors - 1);
1094 chunk_sectors -= sector_div(offset, chunk_sectors);
1096 return min(q->limits.max_sectors, chunk_sectors);
1099 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1102 struct request_queue *q = rq->q;
1104 if (blk_rq_is_passthrough(rq))
1105 return q->limits.max_hw_sectors;
1107 if (!q->limits.chunk_sectors ||
1108 req_op(rq) == REQ_OP_DISCARD ||
1109 req_op(rq) == REQ_OP_SECURE_ERASE)
1110 return blk_queue_get_max_sectors(q, req_op(rq));
1112 return min(blk_max_size_offset(q, offset, 0),
1113 blk_queue_get_max_sectors(q, req_op(rq)));
1116 static inline unsigned int blk_rq_count_bios(struct request *rq)
1118 unsigned int nr_bios = 0;
1121 __rq_for_each_bio(bio, rq)
1127 void blk_steal_bios(struct bio_list *list, struct request *rq);
1130 * Request completion related functions.
1132 * blk_update_request() completes given number of bytes and updates
1133 * the request without completing it.
1135 extern bool blk_update_request(struct request *rq, blk_status_t error,
1136 unsigned int nr_bytes);
1138 extern void blk_abort_request(struct request *);
1141 * Access functions for manipulating queue properties
1143 extern void blk_cleanup_queue(struct request_queue *);
1144 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1145 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1146 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1147 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1148 extern void blk_queue_max_discard_segments(struct request_queue *,
1150 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1151 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1152 unsigned int max_discard_sectors);
1153 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1154 unsigned int max_write_same_sectors);
1155 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1156 unsigned int max_write_same_sectors);
1157 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1158 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1159 unsigned int max_zone_append_sectors);
1160 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1161 void blk_queue_zone_write_granularity(struct request_queue *q,
1163 extern void blk_queue_alignment_offset(struct request_queue *q,
1164 unsigned int alignment);
1165 void blk_queue_update_readahead(struct request_queue *q);
1166 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1167 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1168 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1169 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1170 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1171 extern void blk_set_default_limits(struct queue_limits *lim);
1172 extern void blk_set_stacking_limits(struct queue_limits *lim);
1173 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1175 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1177 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1178 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1179 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1180 extern void blk_queue_dma_alignment(struct request_queue *, int);
1181 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1182 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1183 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1184 extern void blk_queue_required_elevator_features(struct request_queue *q,
1185 unsigned int features);
1186 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1187 struct device *dev);
1190 * Number of physical segments as sent to the device.
1192 * Normally this is the number of discontiguous data segments sent by the
1193 * submitter. But for data-less command like discard we might have no
1194 * actual data segments submitted, but the driver might have to add it's
1195 * own special payload. In that case we still return 1 here so that this
1196 * special payload will be mapped.
1198 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1200 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1202 return rq->nr_phys_segments;
1206 * Number of discard segments (or ranges) the driver needs to fill in.
1207 * Each discard bio merged into a request is counted as one segment.
1209 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1211 return max_t(unsigned short, rq->nr_phys_segments, 1);
1214 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1215 struct scatterlist *sglist, struct scatterlist **last_sg);
1216 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1217 struct scatterlist *sglist)
1219 struct scatterlist *last_sg = NULL;
1221 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1223 extern void blk_dump_rq_flags(struct request *, char *);
1225 bool __must_check blk_get_queue(struct request_queue *);
1226 struct request_queue *blk_alloc_queue(int node_id);
1227 extern void blk_put_queue(struct request_queue *);
1228 extern void blk_set_queue_dying(struct request_queue *);
1232 * blk_plug permits building a queue of related requests by holding the I/O
1233 * fragments for a short period. This allows merging of sequential requests
1234 * into single larger request. As the requests are moved from a per-task list to
1235 * the device's request_queue in a batch, this results in improved scalability
1236 * as the lock contention for request_queue lock is reduced.
1238 * It is ok not to disable preemption when adding the request to the plug list
1239 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1240 * the plug list when the task sleeps by itself. For details, please see
1241 * schedule() where blk_schedule_flush_plug() is called.
1244 struct list_head mq_list; /* blk-mq requests */
1245 struct list_head cb_list; /* md requires an unplug callback */
1246 unsigned short rq_count;
1247 bool multiple_queues;
1250 #define BLK_MAX_REQUEST_COUNT 16
1251 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1254 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1255 struct blk_plug_cb {
1256 struct list_head list;
1257 blk_plug_cb_fn callback;
1260 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1261 void *data, int size);
1262 extern void blk_start_plug(struct blk_plug *);
1263 extern void blk_finish_plug(struct blk_plug *);
1264 extern void blk_flush_plug_list(struct blk_plug *, bool);
1266 static inline void blk_flush_plug(struct task_struct *tsk)
1268 struct blk_plug *plug = tsk->plug;
1271 blk_flush_plug_list(plug, false);
1274 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1276 struct blk_plug *plug = tsk->plug;
1279 blk_flush_plug_list(plug, true);
1282 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1284 struct blk_plug *plug = tsk->plug;
1287 (!list_empty(&plug->mq_list) ||
1288 !list_empty(&plug->cb_list));
1291 int blkdev_issue_flush(struct block_device *bdev);
1292 long nr_blockdev_pages(void);
1293 #else /* CONFIG_BLOCK */
1297 static inline void blk_start_plug(struct blk_plug *plug)
1301 static inline void blk_finish_plug(struct blk_plug *plug)
1305 static inline void blk_flush_plug(struct task_struct *task)
1309 static inline void blk_schedule_flush_plug(struct task_struct *task)
1314 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1319 static inline int blkdev_issue_flush(struct block_device *bdev)
1324 static inline long nr_blockdev_pages(void)
1328 #endif /* CONFIG_BLOCK */
1330 extern void blk_io_schedule(void);
1332 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1333 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1335 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1337 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1338 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1339 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1340 sector_t nr_sects, gfp_t gfp_mask, int flags,
1343 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1344 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1346 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1347 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1349 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1350 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1352 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1353 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1355 return blkdev_issue_discard(sb->s_bdev,
1356 block << (sb->s_blocksize_bits -
1358 nr_blocks << (sb->s_blocksize_bits -
1362 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1363 sector_t nr_blocks, gfp_t gfp_mask)
1365 return blkdev_issue_zeroout(sb->s_bdev,
1366 block << (sb->s_blocksize_bits -
1368 nr_blocks << (sb->s_blocksize_bits -
1373 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1375 static inline bool bdev_is_partition(struct block_device *bdev)
1377 return bdev->bd_partno;
1380 enum blk_default_limits {
1381 BLK_MAX_SEGMENTS = 128,
1382 BLK_SAFE_MAX_SECTORS = 255,
1383 BLK_DEF_MAX_SECTORS = 2560,
1384 BLK_MAX_SEGMENT_SIZE = 65536,
1385 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1388 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1390 return q->limits.seg_boundary_mask;
1393 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1395 return q->limits.virt_boundary_mask;
1398 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1400 return q->limits.max_sectors;
1403 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1405 return q->limits.max_hw_sectors;
1408 static inline unsigned short queue_max_segments(const struct request_queue *q)
1410 return q->limits.max_segments;
1413 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1415 return q->limits.max_discard_segments;
1418 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1420 return q->limits.max_segment_size;
1423 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1426 const struct queue_limits *l = &q->limits;
1428 return min(l->max_zone_append_sectors, l->max_sectors);
1431 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1435 if (q && q->limits.logical_block_size)
1436 retval = q->limits.logical_block_size;
1441 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1443 return queue_logical_block_size(bdev_get_queue(bdev));
1446 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1448 return q->limits.physical_block_size;
1451 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1453 return queue_physical_block_size(bdev_get_queue(bdev));
1456 static inline unsigned int queue_io_min(const struct request_queue *q)
1458 return q->limits.io_min;
1461 static inline int bdev_io_min(struct block_device *bdev)
1463 return queue_io_min(bdev_get_queue(bdev));
1466 static inline unsigned int queue_io_opt(const struct request_queue *q)
1468 return q->limits.io_opt;
1471 static inline int bdev_io_opt(struct block_device *bdev)
1473 return queue_io_opt(bdev_get_queue(bdev));
1476 static inline unsigned int
1477 queue_zone_write_granularity(const struct request_queue *q)
1479 return q->limits.zone_write_granularity;
1482 static inline unsigned int
1483 bdev_zone_write_granularity(struct block_device *bdev)
1485 return queue_zone_write_granularity(bdev_get_queue(bdev));
1488 static inline int queue_alignment_offset(const struct request_queue *q)
1490 if (q->limits.misaligned)
1493 return q->limits.alignment_offset;
1496 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1498 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1499 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1502 return (granularity + lim->alignment_offset - alignment) % granularity;
1505 static inline int bdev_alignment_offset(struct block_device *bdev)
1507 struct request_queue *q = bdev_get_queue(bdev);
1509 if (q->limits.misaligned)
1511 if (bdev_is_partition(bdev))
1512 return queue_limit_alignment_offset(&q->limits,
1513 bdev->bd_start_sect);
1514 return q->limits.alignment_offset;
1517 static inline int queue_discard_alignment(const struct request_queue *q)
1519 if (q->limits.discard_misaligned)
1522 return q->limits.discard_alignment;
1525 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1527 unsigned int alignment, granularity, offset;
1529 if (!lim->max_discard_sectors)
1532 /* Why are these in bytes, not sectors? */
1533 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1534 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1538 /* Offset of the partition start in 'granularity' sectors */
1539 offset = sector_div(sector, granularity);
1541 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1542 offset = (granularity + alignment - offset) % granularity;
1544 /* Turn it back into bytes, gaah */
1545 return offset << SECTOR_SHIFT;
1548 static inline int bdev_discard_alignment(struct block_device *bdev)
1550 struct request_queue *q = bdev_get_queue(bdev);
1552 if (bdev_is_partition(bdev))
1553 return queue_limit_discard_alignment(&q->limits,
1554 bdev->bd_start_sect);
1555 return q->limits.discard_alignment;
1558 static inline unsigned int bdev_write_same(struct block_device *bdev)
1560 struct request_queue *q = bdev_get_queue(bdev);
1563 return q->limits.max_write_same_sectors;
1568 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1570 struct request_queue *q = bdev_get_queue(bdev);
1573 return q->limits.max_write_zeroes_sectors;
1578 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1580 struct request_queue *q = bdev_get_queue(bdev);
1583 return blk_queue_zoned_model(q);
1585 return BLK_ZONED_NONE;
1588 static inline bool bdev_is_zoned(struct block_device *bdev)
1590 struct request_queue *q = bdev_get_queue(bdev);
1593 return blk_queue_is_zoned(q);
1598 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1600 struct request_queue *q = bdev_get_queue(bdev);
1603 return blk_queue_zone_sectors(q);
1607 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1609 struct request_queue *q = bdev_get_queue(bdev);
1612 return queue_max_open_zones(q);
1616 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1618 struct request_queue *q = bdev_get_queue(bdev);
1621 return queue_max_active_zones(q);
1625 static inline int queue_dma_alignment(const struct request_queue *q)
1627 return q ? q->dma_alignment : 511;
1630 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1633 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1634 return !(addr & alignment) && !(len & alignment);
1637 /* assumes size > 256 */
1638 static inline unsigned int blksize_bits(unsigned int size)
1640 unsigned int bits = 8;
1644 } while (size > 256);
1648 static inline unsigned int block_size(struct block_device *bdev)
1650 return 1 << bdev->bd_inode->i_blkbits;
1653 int kblockd_schedule_work(struct work_struct *work);
1654 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1656 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1657 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1658 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1659 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1661 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1663 enum blk_integrity_flags {
1664 BLK_INTEGRITY_VERIFY = 1 << 0,
1665 BLK_INTEGRITY_GENERATE = 1 << 1,
1666 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1667 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1670 struct blk_integrity_iter {
1674 unsigned int data_size;
1675 unsigned short interval;
1676 const char *disk_name;
1679 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1680 typedef void (integrity_prepare_fn) (struct request *);
1681 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1683 struct blk_integrity_profile {
1684 integrity_processing_fn *generate_fn;
1685 integrity_processing_fn *verify_fn;
1686 integrity_prepare_fn *prepare_fn;
1687 integrity_complete_fn *complete_fn;
1691 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1692 extern void blk_integrity_unregister(struct gendisk *);
1693 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1694 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1695 struct scatterlist *);
1696 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1698 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1700 struct blk_integrity *bi = &disk->queue->integrity;
1709 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1711 return blk_get_integrity(bdev->bd_disk);
1715 blk_integrity_queue_supports_integrity(struct request_queue *q)
1717 return q->integrity.profile;
1720 static inline bool blk_integrity_rq(struct request *rq)
1722 return rq->cmd_flags & REQ_INTEGRITY;
1725 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1728 q->limits.max_integrity_segments = segs;
1731 static inline unsigned short
1732 queue_max_integrity_segments(const struct request_queue *q)
1734 return q->limits.max_integrity_segments;
1738 * bio_integrity_intervals - Return number of integrity intervals for a bio
1739 * @bi: blk_integrity profile for device
1740 * @sectors: Size of the bio in 512-byte sectors
1742 * Description: The block layer calculates everything in 512 byte
1743 * sectors but integrity metadata is done in terms of the data integrity
1744 * interval size of the storage device. Convert the block layer sectors
1745 * to the appropriate number of integrity intervals.
1747 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1748 unsigned int sectors)
1750 return sectors >> (bi->interval_exp - 9);
1753 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1754 unsigned int sectors)
1756 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1760 * Return the first bvec that contains integrity data. Only drivers that are
1761 * limited to a single integrity segment should use this helper.
1763 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1765 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1767 return rq->bio->bi_integrity->bip_vec;
1770 #else /* CONFIG_BLK_DEV_INTEGRITY */
1773 struct block_device;
1775 struct blk_integrity;
1777 static inline int blk_integrity_rq(struct request *rq)
1781 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1786 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1788 struct scatterlist *s)
1792 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1796 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1801 blk_integrity_queue_supports_integrity(struct request_queue *q)
1805 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1809 static inline void blk_integrity_register(struct gendisk *d,
1810 struct blk_integrity *b)
1813 static inline void blk_integrity_unregister(struct gendisk *d)
1816 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1820 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1825 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1826 unsigned int sectors)
1831 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1832 unsigned int sectors)
1837 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1842 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1844 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1846 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1848 void blk_ksm_unregister(struct request_queue *q);
1850 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1852 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1853 struct request_queue *q)
1858 static inline void blk_ksm_unregister(struct request_queue *q) { }
1860 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1863 struct block_device_operations {
1864 blk_qc_t (*submit_bio) (struct bio *bio);
1865 int (*open) (struct block_device *, fmode_t);
1866 void (*release) (struct gendisk *, fmode_t);
1867 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1868 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1869 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1870 unsigned int (*check_events) (struct gendisk *disk,
1871 unsigned int clearing);
1872 void (*unlock_native_capacity) (struct gendisk *);
1873 int (*getgeo)(struct block_device *, struct hd_geometry *);
1874 int (*set_read_only)(struct block_device *bdev, bool ro);
1875 /* this callback is with swap_lock and sometimes page table lock held */
1876 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1877 int (*report_zones)(struct gendisk *, sector_t sector,
1878 unsigned int nr_zones, report_zones_cb cb, void *data);
1879 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1880 struct module *owner;
1881 const struct pr_ops *pr_ops;
1884 #ifdef CONFIG_COMPAT
1885 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1886 unsigned int, unsigned long);
1888 #define blkdev_compat_ptr_ioctl NULL
1891 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1892 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1893 struct writeback_control *);
1895 #ifdef CONFIG_BLK_DEV_ZONED
1896 bool blk_req_needs_zone_write_lock(struct request *rq);
1897 bool blk_req_zone_write_trylock(struct request *rq);
1898 void __blk_req_zone_write_lock(struct request *rq);
1899 void __blk_req_zone_write_unlock(struct request *rq);
1901 static inline void blk_req_zone_write_lock(struct request *rq)
1903 if (blk_req_needs_zone_write_lock(rq))
1904 __blk_req_zone_write_lock(rq);
1907 static inline void blk_req_zone_write_unlock(struct request *rq)
1909 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1910 __blk_req_zone_write_unlock(rq);
1913 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1915 return rq->q->seq_zones_wlock &&
1916 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1919 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1921 if (!blk_req_needs_zone_write_lock(rq))
1923 return !blk_req_zone_is_write_locked(rq);
1926 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1931 static inline void blk_req_zone_write_lock(struct request *rq)
1935 static inline void blk_req_zone_write_unlock(struct request *rq)
1938 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1943 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1947 #endif /* CONFIG_BLK_DEV_ZONED */
1949 static inline void blk_wake_io_task(struct task_struct *waiter)
1952 * If we're polling, the task itself is doing the completions. For
1953 * that case, we don't need to signal a wakeup, it's enough to just
1954 * mark us as RUNNING.
1956 if (waiter == current)
1957 __set_current_state(TASK_RUNNING);
1959 wake_up_process(waiter);
1962 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1964 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1965 unsigned long start_time);
1967 unsigned long bio_start_io_acct(struct bio *bio);
1968 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1969 struct block_device *orig_bdev);
1972 * bio_end_io_acct - end I/O accounting for bio based drivers
1973 * @bio: bio to end account for
1974 * @start: start time returned by bio_start_io_acct()
1976 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1978 return bio_end_io_acct_remapped(bio, start_time, bio->bi_bdev);
1981 int bdev_read_only(struct block_device *bdev);
1982 int set_blocksize(struct block_device *bdev, int size);
1984 const char *bdevname(struct block_device *bdev, char *buffer);
1985 int lookup_bdev(const char *pathname, dev_t *dev);
1987 void blkdev_show(struct seq_file *seqf, off_t offset);
1989 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1990 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1992 #define BLKDEV_MAJOR_MAX 512
1994 #define BLKDEV_MAJOR_MAX 0
1997 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1999 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
2000 int bd_prepare_to_claim(struct block_device *bdev, void *holder);
2001 void bd_abort_claiming(struct block_device *bdev, void *holder);
2002 void blkdev_put(struct block_device *bdev, fmode_t mode);
2004 /* just for blk-cgroup, don't use elsewhere */
2005 struct block_device *blkdev_get_no_open(dev_t dev);
2006 void blkdev_put_no_open(struct block_device *bdev);
2008 struct block_device *bdev_alloc(struct gendisk *disk, u8 partno);
2009 void bdev_add(struct block_device *bdev, dev_t dev);
2010 struct block_device *I_BDEV(struct inode *inode);
2011 struct block_device *bdgrab(struct block_device *bdev);
2012 void bdput(struct block_device *);
2013 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
2017 void invalidate_bdev(struct block_device *bdev);
2018 int sync_blockdev(struct block_device *bdev);
2020 static inline void invalidate_bdev(struct block_device *bdev)
2023 static inline int sync_blockdev(struct block_device *bdev)
2028 int fsync_bdev(struct block_device *bdev);
2030 int freeze_bdev(struct block_device *bdev);
2031 int thaw_bdev(struct block_device *bdev);
2033 #endif /* _LINUX_BLKDEV_H */