1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef _LINUX_BLKDEV_H
3 #define _LINUX_BLKDEV_H
5 #include <linux/sched.h>
6 #include <linux/sched/clock.h>
7 #include <linux/major.h>
8 #include <linux/genhd.h>
9 #include <linux/list.h>
10 #include <linux/llist.h>
11 #include <linux/timer.h>
12 #include <linux/workqueue.h>
13 #include <linux/pagemap.h>
14 #include <linux/backing-dev-defs.h>
15 #include <linux/wait.h>
16 #include <linux/mempool.h>
17 #include <linux/pfn.h>
18 #include <linux/bio.h>
19 #include <linux/stringify.h>
20 #include <linux/gfp.h>
21 #include <linux/bsg.h>
22 #include <linux/smp.h>
23 #include <linux/rcupdate.h>
24 #include <linux/percpu-refcount.h>
25 #include <linux/scatterlist.h>
26 #include <linux/blkzoned.h>
30 struct scsi_ioctl_command;
33 struct elevator_queue;
39 struct blk_flush_queue;
42 struct blk_queue_stats;
43 struct blk_stat_callback;
44 struct blk_keyslot_manager;
46 #define BLKDEV_MIN_RQ 4
47 #define BLKDEV_MAX_RQ 128 /* Default maximum */
49 /* Must be consistent with blk_mq_poll_stats_bkt() */
50 #define BLK_MQ_POLL_STATS_BKTS 16
52 /* Doing classic polling */
53 #define BLK_MQ_POLL_CLASSIC -1
56 * Maximum number of blkcg policies allowed to be registered concurrently.
57 * Defined here to simplify include dependency.
59 #define BLKCG_MAX_POLS 5
61 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
65 typedef __u32 __bitwise req_flags_t;
67 /* elevator knows about this request */
68 #define RQF_SORTED ((__force req_flags_t)(1 << 0))
69 /* drive already may have started this one */
70 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
71 /* may not be passed by ioscheduler */
72 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
73 /* request for flush sequence */
74 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
75 /* merge of different types, fail separately */
76 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
77 /* track inflight for MQ */
78 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
79 /* don't call prep for this one */
80 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
81 /* set for "ide_preempt" requests and also for requests for which the SCSI
82 "quiesce" state must be ignored. */
83 #define RQF_PREEMPT ((__force req_flags_t)(1 << 8))
84 /* vaguely specified driver internal error. Ignored by the block layer */
85 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
86 /* don't warn about errors */
87 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
88 /* elevator private data attached */
89 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
90 /* account into disk and partition IO statistics */
91 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
92 /* request came from our alloc pool */
93 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
94 /* runtime pm request */
95 #define RQF_PM ((__force req_flags_t)(1 << 15))
96 /* on IO scheduler merge hash */
97 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
98 /* track IO completion time */
99 #define RQF_STATS ((__force req_flags_t)(1 << 17))
100 /* Look at ->special_vec for the actual data payload instead of the
102 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
103 /* The per-zone write lock is held for this request */
104 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
105 /* already slept for hybrid poll */
106 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
107 /* ->timeout has been called, don't expire again */
108 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
110 /* flags that prevent us from merging requests: */
111 #define RQF_NOMERGE_FLAGS \
112 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
115 * Request state for blk-mq.
124 * Try to put the fields that are referenced together in the same cacheline.
126 * If you modify this structure, make sure to update blk_rq_init() and
127 * especially blk_mq_rq_ctx_init() to take care of the added fields.
130 struct request_queue *q;
131 struct blk_mq_ctx *mq_ctx;
132 struct blk_mq_hw_ctx *mq_hctx;
134 unsigned int cmd_flags; /* op and common flags */
135 req_flags_t rq_flags;
140 /* the following two fields are internal, NEVER access directly */
141 unsigned int __data_len; /* total data len */
142 sector_t __sector; /* sector cursor */
147 struct list_head queuelist;
150 * The hash is used inside the scheduler, and killed once the
151 * request reaches the dispatch list. The ipi_list is only used
152 * to queue the request for softirq completion, which is long
153 * after the request has been unhashed (and even removed from
154 * the dispatch list).
157 struct hlist_node hash; /* merge hash */
158 struct list_head ipi_list;
162 * The rb_node is only used inside the io scheduler, requests
163 * are pruned when moved to the dispatch queue. So let the
164 * completion_data share space with the rb_node.
167 struct rb_node rb_node; /* sort/lookup */
168 struct bio_vec special_vec;
169 void *completion_data;
170 int error_count; /* for legacy drivers, don't use */
174 * Three pointers are available for the IO schedulers, if they need
175 * more they have to dynamically allocate it. Flush requests are
176 * never put on the IO scheduler. So let the flush fields share
177 * space with the elevator data.
187 struct list_head list;
188 rq_end_io_fn *saved_end_io;
192 struct gendisk *rq_disk;
193 struct hd_struct *part;
194 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
195 /* Time that the first bio started allocating this request. */
198 /* Time that this request was allocated for this IO. */
200 /* Time that I/O was submitted to the device. */
201 u64 io_start_time_ns;
203 #ifdef CONFIG_BLK_WBT
204 unsigned short wbt_flags;
207 * rq sectors used for blk stats. It has the same value
208 * with blk_rq_sectors(rq), except that it never be zeroed
211 unsigned short stats_sectors;
214 * Number of scatter-gather DMA addr+len pairs after
215 * physical address coalescing is performed.
217 unsigned short nr_phys_segments;
219 #if defined(CONFIG_BLK_DEV_INTEGRITY)
220 unsigned short nr_integrity_segments;
223 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
224 struct bio_crypt_ctx *crypt_ctx;
225 struct blk_ksm_keyslot *crypt_keyslot;
228 unsigned short write_hint;
229 unsigned short ioprio;
231 enum mq_rq_state state;
234 unsigned int timeout;
235 unsigned long deadline;
238 struct __call_single_data csd;
243 * completion callback.
245 rq_end_io_fn *end_io;
249 static inline bool blk_op_is_scsi(unsigned int op)
251 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
254 static inline bool blk_op_is_private(unsigned int op)
256 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
259 static inline bool blk_rq_is_scsi(struct request *rq)
261 return blk_op_is_scsi(req_op(rq));
264 static inline bool blk_rq_is_private(struct request *rq)
266 return blk_op_is_private(req_op(rq));
269 static inline bool blk_rq_is_passthrough(struct request *rq)
271 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
274 static inline bool bio_is_passthrough(struct bio *bio)
276 unsigned op = bio_op(bio);
278 return blk_op_is_scsi(op) || blk_op_is_private(op);
281 static inline unsigned short req_get_ioprio(struct request *req)
286 #include <linux/elevator.h>
288 struct blk_queue_ctx;
292 enum blk_eh_timer_return {
293 BLK_EH_DONE, /* drivers has completed the command */
294 BLK_EH_RESET_TIMER, /* reset timer and try again */
297 enum blk_queue_state {
302 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
303 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
305 #define BLK_SCSI_MAX_CMDS (256)
306 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
309 * Zoned block device models (zoned limit).
311 * Note: This needs to be ordered from the least to the most severe
312 * restrictions for the inheritance in blk_stack_limits() to work.
314 enum blk_zoned_model {
315 BLK_ZONED_NONE = 0, /* Regular block device */
316 BLK_ZONED_HA, /* Host-aware zoned block device */
317 BLK_ZONED_HM, /* Host-managed zoned block device */
320 struct queue_limits {
321 unsigned long bounce_pfn;
322 unsigned long seg_boundary_mask;
323 unsigned long virt_boundary_mask;
325 unsigned int max_hw_sectors;
326 unsigned int max_dev_sectors;
327 unsigned int chunk_sectors;
328 unsigned int max_sectors;
329 unsigned int max_segment_size;
330 unsigned int physical_block_size;
331 unsigned int logical_block_size;
332 unsigned int alignment_offset;
335 unsigned int max_discard_sectors;
336 unsigned int max_hw_discard_sectors;
337 unsigned int max_write_same_sectors;
338 unsigned int max_write_zeroes_sectors;
339 unsigned int max_zone_append_sectors;
340 unsigned int discard_granularity;
341 unsigned int discard_alignment;
343 unsigned short max_segments;
344 unsigned short max_integrity_segments;
345 unsigned short max_discard_segments;
347 unsigned char misaligned;
348 unsigned char discard_misaligned;
349 unsigned char raid_partial_stripes_expensive;
350 enum blk_zoned_model zoned;
353 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
356 #ifdef CONFIG_BLK_DEV_ZONED
358 #define BLK_ALL_ZONES ((unsigned int)-1)
359 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
360 unsigned int nr_zones, report_zones_cb cb, void *data);
361 unsigned int blkdev_nr_zones(struct gendisk *disk);
362 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
363 sector_t sectors, sector_t nr_sectors,
365 int blk_revalidate_disk_zones(struct gendisk *disk,
366 void (*update_driver_data)(struct gendisk *disk));
368 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
369 unsigned int cmd, unsigned long arg);
370 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
371 unsigned int cmd, unsigned long arg);
373 #else /* CONFIG_BLK_DEV_ZONED */
375 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
380 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
381 fmode_t mode, unsigned int cmd,
387 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
388 fmode_t mode, unsigned int cmd,
394 #endif /* CONFIG_BLK_DEV_ZONED */
396 struct request_queue {
397 struct request *last_merge;
398 struct elevator_queue *elevator;
400 struct blk_queue_stats *stats;
401 struct rq_qos *rq_qos;
403 const struct blk_mq_ops *mq_ops;
406 struct blk_mq_ctx __percpu *queue_ctx;
408 unsigned int queue_depth;
410 /* hw dispatch queues */
411 struct blk_mq_hw_ctx **queue_hw_ctx;
412 unsigned int nr_hw_queues;
414 struct backing_dev_info *backing_dev_info;
417 * The queue owner gets to use this for whatever they like.
418 * ll_rw_blk doesn't touch it.
423 * various queue flags, see QUEUE_* below
425 unsigned long queue_flags;
427 * Number of contexts that have called blk_set_pm_only(). If this
428 * counter is above zero then only RQF_PM and RQF_PREEMPT requests are
434 * ida allocated id for this queue. Used to index queues from
440 * queue needs bounce pages for pages above this limit
444 spinlock_t queue_lock;
454 struct kobject *mq_kobj;
456 #ifdef CONFIG_BLK_DEV_INTEGRITY
457 struct blk_integrity integrity;
458 #endif /* CONFIG_BLK_DEV_INTEGRITY */
462 enum rpm_status rpm_status;
463 unsigned int nr_pending;
469 unsigned long nr_requests; /* Max # of requests */
471 unsigned int dma_pad_mask;
472 unsigned int dma_alignment;
474 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
475 /* Inline crypto capabilities */
476 struct blk_keyslot_manager *ksm;
479 unsigned int rq_timeout;
482 struct blk_stat_callback *poll_cb;
483 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
485 struct timer_list timeout;
486 struct work_struct timeout_work;
488 atomic_t nr_active_requests_shared_sbitmap;
490 struct list_head icq_list;
491 #ifdef CONFIG_BLK_CGROUP
492 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
493 struct blkcg_gq *root_blkg;
494 struct list_head blkg_list;
497 struct queue_limits limits;
499 unsigned int required_elevator_features;
501 #ifdef CONFIG_BLK_DEV_ZONED
503 * Zoned block device information for request dispatch control.
504 * nr_zones is the total number of zones of the device. This is always
505 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
506 * bits which indicates if a zone is conventional (bit set) or
507 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
508 * bits which indicates if a zone is write locked, that is, if a write
509 * request targeting the zone was dispatched. All three fields are
510 * initialized by the low level device driver (e.g. scsi/sd.c).
511 * Stacking drivers (device mappers) may or may not initialize
514 * Reads of this information must be protected with blk_queue_enter() /
515 * blk_queue_exit(). Modifying this information is only allowed while
516 * no requests are being processed. See also blk_mq_freeze_queue() and
517 * blk_mq_unfreeze_queue().
519 unsigned int nr_zones;
520 unsigned long *conv_zones_bitmap;
521 unsigned long *seq_zones_wlock;
522 unsigned int max_open_zones;
523 unsigned int max_active_zones;
524 #endif /* CONFIG_BLK_DEV_ZONED */
529 unsigned int sg_timeout;
530 unsigned int sg_reserved_size;
532 struct mutex debugfs_mutex;
533 #ifdef CONFIG_BLK_DEV_IO_TRACE
534 struct blk_trace __rcu *blk_trace;
537 * for flush operations
539 struct blk_flush_queue *fq;
541 struct list_head requeue_list;
542 spinlock_t requeue_lock;
543 struct delayed_work requeue_work;
545 struct mutex sysfs_lock;
546 struct mutex sysfs_dir_lock;
549 * for reusing dead hctx instance in case of updating
552 struct list_head unused_hctx_list;
553 spinlock_t unused_hctx_lock;
557 #if defined(CONFIG_BLK_DEV_BSG)
558 struct bsg_class_device bsg_dev;
561 #ifdef CONFIG_BLK_DEV_THROTTLING
563 struct throtl_data *td;
565 struct rcu_head rcu_head;
566 wait_queue_head_t mq_freeze_wq;
568 * Protect concurrent access to q_usage_counter by
569 * percpu_ref_kill() and percpu_ref_reinit().
571 struct mutex mq_freeze_lock;
572 struct percpu_ref q_usage_counter;
574 struct blk_mq_tag_set *tag_set;
575 struct list_head tag_set_list;
576 struct bio_set bio_split;
578 struct dentry *debugfs_dir;
580 #ifdef CONFIG_BLK_DEBUG_FS
581 struct dentry *sched_debugfs_dir;
582 struct dentry *rqos_debugfs_dir;
585 bool mq_sysfs_init_done;
589 #define BLK_MAX_WRITE_HINTS 5
590 u64 write_hints[BLK_MAX_WRITE_HINTS];
593 /* Keep blk_queue_flag_name[] in sync with the definitions below */
594 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
595 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
596 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
597 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
598 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
599 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
600 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
601 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
602 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
603 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
604 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
605 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
606 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
607 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
608 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
609 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
610 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
611 #define QUEUE_FLAG_WC 17 /* Write back caching */
612 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
613 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
614 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
615 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
616 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
617 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
618 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
619 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
620 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
621 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
622 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
624 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
625 (1 << QUEUE_FLAG_SAME_COMP))
627 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
628 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
629 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
631 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
632 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
633 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
634 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
635 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
636 #define blk_queue_noxmerges(q) \
637 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
638 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
639 #define blk_queue_stable_writes(q) \
640 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
641 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
642 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
643 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
644 #define blk_queue_zone_resetall(q) \
645 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
646 #define blk_queue_secure_erase(q) \
647 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
648 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
649 #define blk_queue_scsi_passthrough(q) \
650 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
651 #define blk_queue_pci_p2pdma(q) \
652 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
653 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
654 #define blk_queue_rq_alloc_time(q) \
655 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
657 #define blk_queue_rq_alloc_time(q) false
660 #define blk_noretry_request(rq) \
661 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
662 REQ_FAILFAST_DRIVER))
663 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
664 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
665 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
666 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
668 extern void blk_set_pm_only(struct request_queue *q);
669 extern void blk_clear_pm_only(struct request_queue *q);
671 static inline bool blk_account_rq(struct request *rq)
673 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
676 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
678 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
680 #define rq_dma_dir(rq) \
681 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
683 #define dma_map_bvec(dev, bv, dir, attrs) \
684 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
687 static inline bool queue_is_mq(struct request_queue *q)
692 static inline enum blk_zoned_model
693 blk_queue_zoned_model(struct request_queue *q)
695 return q->limits.zoned;
698 static inline bool blk_queue_is_zoned(struct request_queue *q)
700 switch (blk_queue_zoned_model(q)) {
709 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
711 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
714 #ifdef CONFIG_BLK_DEV_ZONED
715 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
717 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
720 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
723 if (!blk_queue_is_zoned(q))
725 return sector >> ilog2(q->limits.chunk_sectors);
728 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
731 if (!blk_queue_is_zoned(q))
733 if (!q->conv_zones_bitmap)
735 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
738 static inline void blk_queue_max_open_zones(struct request_queue *q,
739 unsigned int max_open_zones)
741 q->max_open_zones = max_open_zones;
744 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
746 return q->max_open_zones;
749 static inline void blk_queue_max_active_zones(struct request_queue *q,
750 unsigned int max_active_zones)
752 q->max_active_zones = max_active_zones;
755 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
757 return q->max_active_zones;
759 #else /* CONFIG_BLK_DEV_ZONED */
760 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
764 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
769 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
774 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
778 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
782 #endif /* CONFIG_BLK_DEV_ZONED */
784 static inline bool rq_is_sync(struct request *rq)
786 return op_is_sync(rq->cmd_flags);
789 static inline bool rq_mergeable(struct request *rq)
791 if (blk_rq_is_passthrough(rq))
794 if (req_op(rq) == REQ_OP_FLUSH)
797 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
800 if (req_op(rq) == REQ_OP_ZONE_APPEND)
803 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
805 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
811 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
813 if (bio_page(a) == bio_page(b) &&
814 bio_offset(a) == bio_offset(b))
820 static inline unsigned int blk_queue_depth(struct request_queue *q)
823 return q->queue_depth;
825 return q->nr_requests;
828 extern unsigned long blk_max_low_pfn, blk_max_pfn;
831 * standard bounce addresses:
833 * BLK_BOUNCE_HIGH : bounce all highmem pages
834 * BLK_BOUNCE_ANY : don't bounce anything
835 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
838 #if BITS_PER_LONG == 32
839 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
841 #define BLK_BOUNCE_HIGH -1ULL
843 #define BLK_BOUNCE_ANY (-1ULL)
844 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
847 * default timeout for SG_IO if none specified
849 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
850 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
856 unsigned long offset;
861 struct req_iterator {
862 struct bvec_iter iter;
866 /* This should not be used directly - use rq_for_each_segment */
867 #define for_each_bio(_bio) \
868 for (; _bio; _bio = _bio->bi_next)
869 #define __rq_for_each_bio(_bio, rq) \
871 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
873 #define rq_for_each_segment(bvl, _rq, _iter) \
874 __rq_for_each_bio(_iter.bio, _rq) \
875 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
877 #define rq_for_each_bvec(bvl, _rq, _iter) \
878 __rq_for_each_bio(_iter.bio, _rq) \
879 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
881 #define rq_iter_last(bvec, _iter) \
882 (_iter.bio->bi_next == NULL && \
883 bio_iter_last(bvec, _iter.iter))
885 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
886 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
888 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
889 extern void rq_flush_dcache_pages(struct request *rq);
891 static inline void rq_flush_dcache_pages(struct request *rq)
896 extern int blk_register_queue(struct gendisk *disk);
897 extern void blk_unregister_queue(struct gendisk *disk);
898 blk_qc_t submit_bio_noacct(struct bio *bio);
899 extern void blk_rq_init(struct request_queue *q, struct request *rq);
900 extern void blk_put_request(struct request *);
901 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
902 blk_mq_req_flags_t flags);
903 extern int blk_lld_busy(struct request_queue *q);
904 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
905 struct bio_set *bs, gfp_t gfp_mask,
906 int (*bio_ctr)(struct bio *, struct bio *, void *),
908 extern void blk_rq_unprep_clone(struct request *rq);
909 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
911 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
912 extern void blk_queue_split(struct bio **);
913 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
914 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
915 unsigned int, void __user *);
916 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
917 unsigned int, void __user *);
918 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
919 struct scsi_ioctl_command __user *);
920 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
921 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
923 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
924 extern void blk_queue_exit(struct request_queue *q);
925 extern void blk_sync_queue(struct request_queue *q);
926 extern int blk_rq_map_user(struct request_queue *, struct request *,
927 struct rq_map_data *, void __user *, unsigned long,
929 extern int blk_rq_unmap_user(struct bio *);
930 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
931 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
932 struct rq_map_data *, const struct iov_iter *,
934 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
935 struct request *, int);
936 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
937 struct request *, int, rq_end_io_fn *);
939 /* Helper to convert REQ_OP_XXX to its string format XXX */
940 extern const char *blk_op_str(unsigned int op);
942 int blk_status_to_errno(blk_status_t status);
943 blk_status_t errno_to_blk_status(int errno);
945 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
947 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
949 return bdev->bd_disk->queue; /* this is never NULL */
953 * The basic unit of block I/O is a sector. It is used in a number of contexts
954 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
955 * bytes. Variables of type sector_t represent an offset or size that is a
956 * multiple of 512 bytes. Hence these two constants.
959 #define SECTOR_SHIFT 9
962 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
966 * blk_rq_pos() : the current sector
967 * blk_rq_bytes() : bytes left in the entire request
968 * blk_rq_cur_bytes() : bytes left in the current segment
969 * blk_rq_err_bytes() : bytes left till the next error boundary
970 * blk_rq_sectors() : sectors left in the entire request
971 * blk_rq_cur_sectors() : sectors left in the current segment
972 * blk_rq_stats_sectors() : sectors of the entire request used for stats
974 static inline sector_t blk_rq_pos(const struct request *rq)
979 static inline unsigned int blk_rq_bytes(const struct request *rq)
981 return rq->__data_len;
984 static inline int blk_rq_cur_bytes(const struct request *rq)
986 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
989 extern unsigned int blk_rq_err_bytes(const struct request *rq);
991 static inline unsigned int blk_rq_sectors(const struct request *rq)
993 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
996 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
998 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
1001 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
1003 return rq->stats_sectors;
1006 #ifdef CONFIG_BLK_DEV_ZONED
1008 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
1009 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
1011 static inline unsigned int blk_rq_zone_no(struct request *rq)
1013 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1016 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1018 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1020 #endif /* CONFIG_BLK_DEV_ZONED */
1023 * Some commands like WRITE SAME have a payload or data transfer size which
1024 * is different from the size of the request. Any driver that supports such
1025 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1026 * calculate the data transfer size.
1028 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1030 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1031 return rq->special_vec.bv_len;
1032 return blk_rq_bytes(rq);
1036 * Return the first full biovec in the request. The caller needs to check that
1037 * there are any bvecs before calling this helper.
1039 static inline struct bio_vec req_bvec(struct request *rq)
1041 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1042 return rq->special_vec;
1043 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1046 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1049 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1050 return min(q->limits.max_discard_sectors,
1051 UINT_MAX >> SECTOR_SHIFT);
1053 if (unlikely(op == REQ_OP_WRITE_SAME))
1054 return q->limits.max_write_same_sectors;
1056 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1057 return q->limits.max_write_zeroes_sectors;
1059 return q->limits.max_sectors;
1063 * Return maximum size of a request at given offset. Only valid for
1064 * file system requests.
1066 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1069 unsigned int chunk_sectors = q->limits.chunk_sectors;
1072 return q->limits.max_sectors;
1074 if (likely(is_power_of_2(chunk_sectors)))
1075 chunk_sectors -= offset & (chunk_sectors - 1);
1077 chunk_sectors -= sector_div(offset, chunk_sectors);
1079 return min(q->limits.max_sectors, chunk_sectors);
1082 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1085 struct request_queue *q = rq->q;
1087 if (blk_rq_is_passthrough(rq))
1088 return q->limits.max_hw_sectors;
1090 if (!q->limits.chunk_sectors ||
1091 req_op(rq) == REQ_OP_DISCARD ||
1092 req_op(rq) == REQ_OP_SECURE_ERASE)
1093 return blk_queue_get_max_sectors(q, req_op(rq));
1095 return min(blk_max_size_offset(q, offset),
1096 blk_queue_get_max_sectors(q, req_op(rq)));
1099 static inline unsigned int blk_rq_count_bios(struct request *rq)
1101 unsigned int nr_bios = 0;
1104 __rq_for_each_bio(bio, rq)
1110 void blk_steal_bios(struct bio_list *list, struct request *rq);
1113 * Request completion related functions.
1115 * blk_update_request() completes given number of bytes and updates
1116 * the request without completing it.
1118 extern bool blk_update_request(struct request *rq, blk_status_t error,
1119 unsigned int nr_bytes);
1121 extern void blk_abort_request(struct request *);
1124 * Access functions for manipulating queue properties
1126 extern void blk_cleanup_queue(struct request_queue *);
1127 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1128 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1129 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1130 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1131 extern void blk_queue_max_discard_segments(struct request_queue *,
1133 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1134 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1135 unsigned int max_discard_sectors);
1136 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1137 unsigned int max_write_same_sectors);
1138 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1139 unsigned int max_write_same_sectors);
1140 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1141 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1142 unsigned int max_zone_append_sectors);
1143 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1144 extern void blk_queue_alignment_offset(struct request_queue *q,
1145 unsigned int alignment);
1146 void blk_queue_update_readahead(struct request_queue *q);
1147 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1148 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1149 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1150 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1151 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1152 extern void blk_set_default_limits(struct queue_limits *lim);
1153 extern void blk_set_stacking_limits(struct queue_limits *lim);
1154 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1156 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1158 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1159 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1160 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1161 extern void blk_queue_dma_alignment(struct request_queue *, int);
1162 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1163 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1164 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1165 extern void blk_queue_required_elevator_features(struct request_queue *q,
1166 unsigned int features);
1167 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1168 struct device *dev);
1171 * Number of physical segments as sent to the device.
1173 * Normally this is the number of discontiguous data segments sent by the
1174 * submitter. But for data-less command like discard we might have no
1175 * actual data segments submitted, but the driver might have to add it's
1176 * own special payload. In that case we still return 1 here so that this
1177 * special payload will be mapped.
1179 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1181 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1183 return rq->nr_phys_segments;
1187 * Number of discard segments (or ranges) the driver needs to fill in.
1188 * Each discard bio merged into a request is counted as one segment.
1190 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1192 return max_t(unsigned short, rq->nr_phys_segments, 1);
1195 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1196 struct scatterlist *sglist, struct scatterlist **last_sg);
1197 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1198 struct scatterlist *sglist)
1200 struct scatterlist *last_sg = NULL;
1202 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1204 extern void blk_dump_rq_flags(struct request *, char *);
1206 bool __must_check blk_get_queue(struct request_queue *);
1207 struct request_queue *blk_alloc_queue(int node_id);
1208 extern void blk_put_queue(struct request_queue *);
1209 extern void blk_set_queue_dying(struct request_queue *);
1213 * blk_plug permits building a queue of related requests by holding the I/O
1214 * fragments for a short period. This allows merging of sequential requests
1215 * into single larger request. As the requests are moved from a per-task list to
1216 * the device's request_queue in a batch, this results in improved scalability
1217 * as the lock contention for request_queue lock is reduced.
1219 * It is ok not to disable preemption when adding the request to the plug list
1220 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1221 * the plug list when the task sleeps by itself. For details, please see
1222 * schedule() where blk_schedule_flush_plug() is called.
1225 struct list_head mq_list; /* blk-mq requests */
1226 struct list_head cb_list; /* md requires an unplug callback */
1227 unsigned short rq_count;
1228 bool multiple_queues;
1231 #define BLK_MAX_REQUEST_COUNT 16
1232 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1235 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1236 struct blk_plug_cb {
1237 struct list_head list;
1238 blk_plug_cb_fn callback;
1241 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1242 void *data, int size);
1243 extern void blk_start_plug(struct blk_plug *);
1244 extern void blk_finish_plug(struct blk_plug *);
1245 extern void blk_flush_plug_list(struct blk_plug *, bool);
1247 static inline void blk_flush_plug(struct task_struct *tsk)
1249 struct blk_plug *plug = tsk->plug;
1252 blk_flush_plug_list(plug, false);
1255 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1257 struct blk_plug *plug = tsk->plug;
1260 blk_flush_plug_list(plug, true);
1263 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1265 struct blk_plug *plug = tsk->plug;
1268 (!list_empty(&plug->mq_list) ||
1269 !list_empty(&plug->cb_list));
1272 int blkdev_issue_flush(struct block_device *, gfp_t);
1273 long nr_blockdev_pages(void);
1274 #else /* CONFIG_BLOCK */
1278 static inline void blk_start_plug(struct blk_plug *plug)
1282 static inline void blk_finish_plug(struct blk_plug *plug)
1286 static inline void blk_flush_plug(struct task_struct *task)
1290 static inline void blk_schedule_flush_plug(struct task_struct *task)
1295 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1300 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask)
1305 static inline long nr_blockdev_pages(void)
1309 #endif /* CONFIG_BLOCK */
1311 extern void blk_io_schedule(void);
1313 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1314 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1316 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1318 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1319 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1320 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1321 sector_t nr_sects, gfp_t gfp_mask, int flags,
1324 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1325 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1327 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1328 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1330 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1331 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1333 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1334 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1336 return blkdev_issue_discard(sb->s_bdev,
1337 block << (sb->s_blocksize_bits -
1339 nr_blocks << (sb->s_blocksize_bits -
1343 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1344 sector_t nr_blocks, gfp_t gfp_mask)
1346 return blkdev_issue_zeroout(sb->s_bdev,
1347 block << (sb->s_blocksize_bits -
1349 nr_blocks << (sb->s_blocksize_bits -
1354 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1356 enum blk_default_limits {
1357 BLK_MAX_SEGMENTS = 128,
1358 BLK_SAFE_MAX_SECTORS = 255,
1359 BLK_DEF_MAX_SECTORS = 2560,
1360 BLK_MAX_SEGMENT_SIZE = 65536,
1361 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1364 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1366 return q->limits.seg_boundary_mask;
1369 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1371 return q->limits.virt_boundary_mask;
1374 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1376 return q->limits.max_sectors;
1379 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1381 return q->limits.max_hw_sectors;
1384 static inline unsigned short queue_max_segments(const struct request_queue *q)
1386 return q->limits.max_segments;
1389 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1391 return q->limits.max_discard_segments;
1394 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1396 return q->limits.max_segment_size;
1399 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1401 return q->limits.max_zone_append_sectors;
1404 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1408 if (q && q->limits.logical_block_size)
1409 retval = q->limits.logical_block_size;
1414 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1416 return queue_logical_block_size(bdev_get_queue(bdev));
1419 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1421 return q->limits.physical_block_size;
1424 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1426 return queue_physical_block_size(bdev_get_queue(bdev));
1429 static inline unsigned int queue_io_min(const struct request_queue *q)
1431 return q->limits.io_min;
1434 static inline int bdev_io_min(struct block_device *bdev)
1436 return queue_io_min(bdev_get_queue(bdev));
1439 static inline unsigned int queue_io_opt(const struct request_queue *q)
1441 return q->limits.io_opt;
1444 static inline int bdev_io_opt(struct block_device *bdev)
1446 return queue_io_opt(bdev_get_queue(bdev));
1449 static inline int queue_alignment_offset(const struct request_queue *q)
1451 if (q->limits.misaligned)
1454 return q->limits.alignment_offset;
1457 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1459 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1460 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1463 return (granularity + lim->alignment_offset - alignment) % granularity;
1466 static inline int bdev_alignment_offset(struct block_device *bdev)
1468 struct request_queue *q = bdev_get_queue(bdev);
1470 if (q->limits.misaligned)
1472 if (bdev != bdev->bd_contains)
1473 return queue_limit_alignment_offset(&q->limits,
1474 bdev->bd_part->start_sect);
1475 return q->limits.alignment_offset;
1478 static inline int queue_discard_alignment(const struct request_queue *q)
1480 if (q->limits.discard_misaligned)
1483 return q->limits.discard_alignment;
1486 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1488 unsigned int alignment, granularity, offset;
1490 if (!lim->max_discard_sectors)
1493 /* Why are these in bytes, not sectors? */
1494 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1495 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1499 /* Offset of the partition start in 'granularity' sectors */
1500 offset = sector_div(sector, granularity);
1502 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1503 offset = (granularity + alignment - offset) % granularity;
1505 /* Turn it back into bytes, gaah */
1506 return offset << SECTOR_SHIFT;
1509 static inline int bdev_discard_alignment(struct block_device *bdev)
1511 struct request_queue *q = bdev_get_queue(bdev);
1513 if (bdev != bdev->bd_contains)
1514 return queue_limit_discard_alignment(&q->limits,
1515 bdev->bd_part->start_sect);
1516 return q->limits.discard_alignment;
1519 static inline unsigned int bdev_write_same(struct block_device *bdev)
1521 struct request_queue *q = bdev_get_queue(bdev);
1524 return q->limits.max_write_same_sectors;
1529 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1531 struct request_queue *q = bdev_get_queue(bdev);
1534 return q->limits.max_write_zeroes_sectors;
1539 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1541 struct request_queue *q = bdev_get_queue(bdev);
1544 return blk_queue_zoned_model(q);
1546 return BLK_ZONED_NONE;
1549 static inline bool bdev_is_zoned(struct block_device *bdev)
1551 struct request_queue *q = bdev_get_queue(bdev);
1554 return blk_queue_is_zoned(q);
1559 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1561 struct request_queue *q = bdev_get_queue(bdev);
1564 return blk_queue_zone_sectors(q);
1568 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1570 struct request_queue *q = bdev_get_queue(bdev);
1573 return queue_max_open_zones(q);
1577 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1579 struct request_queue *q = bdev_get_queue(bdev);
1582 return queue_max_active_zones(q);
1586 static inline int queue_dma_alignment(const struct request_queue *q)
1588 return q ? q->dma_alignment : 511;
1591 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1594 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1595 return !(addr & alignment) && !(len & alignment);
1598 /* assumes size > 256 */
1599 static inline unsigned int blksize_bits(unsigned int size)
1601 unsigned int bits = 8;
1605 } while (size > 256);
1609 static inline unsigned int block_size(struct block_device *bdev)
1611 return 1 << bdev->bd_inode->i_blkbits;
1614 int kblockd_schedule_work(struct work_struct *work);
1615 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1617 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1618 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1619 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1620 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1622 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1624 enum blk_integrity_flags {
1625 BLK_INTEGRITY_VERIFY = 1 << 0,
1626 BLK_INTEGRITY_GENERATE = 1 << 1,
1627 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1628 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1631 struct blk_integrity_iter {
1635 unsigned int data_size;
1636 unsigned short interval;
1637 const char *disk_name;
1640 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1641 typedef void (integrity_prepare_fn) (struct request *);
1642 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1644 struct blk_integrity_profile {
1645 integrity_processing_fn *generate_fn;
1646 integrity_processing_fn *verify_fn;
1647 integrity_prepare_fn *prepare_fn;
1648 integrity_complete_fn *complete_fn;
1652 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1653 extern void blk_integrity_unregister(struct gendisk *);
1654 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1655 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1656 struct scatterlist *);
1657 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1658 extern bool blk_integrity_merge_rq(struct request_queue *, struct request *,
1660 extern bool blk_integrity_merge_bio(struct request_queue *, struct request *,
1663 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1665 struct blk_integrity *bi = &disk->queue->integrity;
1674 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1676 return blk_get_integrity(bdev->bd_disk);
1680 blk_integrity_queue_supports_integrity(struct request_queue *q)
1682 return q->integrity.profile;
1685 static inline bool blk_integrity_rq(struct request *rq)
1687 return rq->cmd_flags & REQ_INTEGRITY;
1690 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1693 q->limits.max_integrity_segments = segs;
1696 static inline unsigned short
1697 queue_max_integrity_segments(const struct request_queue *q)
1699 return q->limits.max_integrity_segments;
1703 * bio_integrity_intervals - Return number of integrity intervals for a bio
1704 * @bi: blk_integrity profile for device
1705 * @sectors: Size of the bio in 512-byte sectors
1707 * Description: The block layer calculates everything in 512 byte
1708 * sectors but integrity metadata is done in terms of the data integrity
1709 * interval size of the storage device. Convert the block layer sectors
1710 * to the appropriate number of integrity intervals.
1712 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1713 unsigned int sectors)
1715 return sectors >> (bi->interval_exp - 9);
1718 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1719 unsigned int sectors)
1721 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1725 * Return the first bvec that contains integrity data. Only drivers that are
1726 * limited to a single integrity segment should use this helper.
1728 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1730 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1732 return rq->bio->bi_integrity->bip_vec;
1735 #else /* CONFIG_BLK_DEV_INTEGRITY */
1738 struct block_device;
1740 struct blk_integrity;
1742 static inline int blk_integrity_rq(struct request *rq)
1746 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1751 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1753 struct scatterlist *s)
1757 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1761 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1766 blk_integrity_queue_supports_integrity(struct request_queue *q)
1770 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1774 static inline void blk_integrity_register(struct gendisk *d,
1775 struct blk_integrity *b)
1778 static inline void blk_integrity_unregister(struct gendisk *d)
1781 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1785 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1789 static inline bool blk_integrity_merge_rq(struct request_queue *rq,
1795 static inline bool blk_integrity_merge_bio(struct request_queue *rq,
1802 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1803 unsigned int sectors)
1808 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1809 unsigned int sectors)
1814 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1819 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1821 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1823 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1825 void blk_ksm_unregister(struct request_queue *q);
1827 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1829 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1830 struct request_queue *q)
1835 static inline void blk_ksm_unregister(struct request_queue *q) { }
1837 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1840 struct block_device_operations {
1841 blk_qc_t (*submit_bio) (struct bio *bio);
1842 int (*open) (struct block_device *, fmode_t);
1843 void (*release) (struct gendisk *, fmode_t);
1844 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1845 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1846 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1847 unsigned int (*check_events) (struct gendisk *disk,
1848 unsigned int clearing);
1849 void (*unlock_native_capacity) (struct gendisk *);
1850 int (*revalidate_disk) (struct gendisk *);
1851 int (*getgeo)(struct block_device *, struct hd_geometry *);
1852 /* this callback is with swap_lock and sometimes page table lock held */
1853 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1854 int (*report_zones)(struct gendisk *, sector_t sector,
1855 unsigned int nr_zones, report_zones_cb cb, void *data);
1856 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1857 struct module *owner;
1858 const struct pr_ops *pr_ops;
1861 #ifdef CONFIG_COMPAT
1862 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1863 unsigned int, unsigned long);
1865 #define blkdev_compat_ptr_ioctl NULL
1868 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1870 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1871 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1872 struct writeback_control *);
1874 #ifdef CONFIG_BLK_DEV_ZONED
1875 bool blk_req_needs_zone_write_lock(struct request *rq);
1876 bool blk_req_zone_write_trylock(struct request *rq);
1877 void __blk_req_zone_write_lock(struct request *rq);
1878 void __blk_req_zone_write_unlock(struct request *rq);
1880 static inline void blk_req_zone_write_lock(struct request *rq)
1882 if (blk_req_needs_zone_write_lock(rq))
1883 __blk_req_zone_write_lock(rq);
1886 static inline void blk_req_zone_write_unlock(struct request *rq)
1888 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1889 __blk_req_zone_write_unlock(rq);
1892 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1894 return rq->q->seq_zones_wlock &&
1895 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1898 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1900 if (!blk_req_needs_zone_write_lock(rq))
1902 return !blk_req_zone_is_write_locked(rq);
1905 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1910 static inline void blk_req_zone_write_lock(struct request *rq)
1914 static inline void blk_req_zone_write_unlock(struct request *rq)
1917 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1922 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1926 #endif /* CONFIG_BLK_DEV_ZONED */
1928 static inline void blk_wake_io_task(struct task_struct *waiter)
1931 * If we're polling, the task itself is doing the completions. For
1932 * that case, we don't need to signal a wakeup, it's enough to just
1933 * mark us as RUNNING.
1935 if (waiter == current)
1936 __set_current_state(TASK_RUNNING);
1938 wake_up_process(waiter);
1941 unsigned long disk_start_io_acct(struct gendisk *disk, unsigned int sectors,
1943 void disk_end_io_acct(struct gendisk *disk, unsigned int op,
1944 unsigned long start_time);
1946 unsigned long part_start_io_acct(struct gendisk *disk, struct hd_struct **part,
1948 void part_end_io_acct(struct hd_struct *part, struct bio *bio,
1949 unsigned long start_time);
1952 * bio_start_io_acct - start I/O accounting for bio based drivers
1953 * @bio: bio to start account for
1955 * Returns the start time that should be passed back to bio_end_io_acct().
1957 static inline unsigned long bio_start_io_acct(struct bio *bio)
1959 return disk_start_io_acct(bio->bi_disk, bio_sectors(bio), bio_op(bio));
1963 * bio_end_io_acct - end I/O accounting for bio based drivers
1964 * @bio: bio to end account for
1965 * @start: start time returned by bio_start_io_acct()
1967 static inline void bio_end_io_acct(struct bio *bio, unsigned long start_time)
1969 return disk_end_io_acct(bio->bi_disk, bio_op(bio), start_time);
1972 int bdev_read_only(struct block_device *bdev);
1973 int set_blocksize(struct block_device *bdev, int size);
1975 const char *bdevname(struct block_device *bdev, char *buffer);
1976 struct block_device *lookup_bdev(const char *);
1978 void blkdev_show(struct seq_file *seqf, off_t offset);
1980 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1981 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
1983 #define BLKDEV_MAJOR_MAX 512
1985 #define BLKDEV_MAJOR_MAX 0
1988 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
1990 struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder);
1991 int bd_prepare_to_claim(struct block_device *bdev, struct block_device *whole,
1993 void bd_abort_claiming(struct block_device *bdev, struct block_device *whole,
1995 void blkdev_put(struct block_device *bdev, fmode_t mode);
1997 struct block_device *I_BDEV(struct inode *inode);
1998 struct block_device *bdget(dev_t);
1999 struct block_device *bdgrab(struct block_device *bdev);
2000 void bdput(struct block_device *);
2003 void invalidate_bdev(struct block_device *bdev);
2004 int truncate_bdev_range(struct block_device *bdev, fmode_t mode, loff_t lstart,
2006 int sync_blockdev(struct block_device *bdev);
2008 static inline void invalidate_bdev(struct block_device *bdev)
2011 static inline int truncate_bdev_range(struct block_device *bdev, fmode_t mode,
2012 loff_t lstart, loff_t lend)
2016 static inline int sync_blockdev(struct block_device *bdev)
2021 int fsync_bdev(struct block_device *bdev);
2023 struct super_block *freeze_bdev(struct block_device *bdev);
2024 int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2026 #endif /* _LINUX_BLKDEV_H */