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 /* elevator knows about this request */
69 #define RQF_SORTED ((__force req_flags_t)(1 << 0))
70 /* drive already may have started this one */
71 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
72 /* may not be passed by ioscheduler */
73 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
74 /* request for flush sequence */
75 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
76 /* merge of different types, fail separately */
77 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
78 /* track inflight for MQ */
79 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
80 /* don't call prep for this one */
81 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
82 /* set for "ide_preempt" requests and also for requests for which the SCSI
83 "quiesce" state must be ignored. */
84 #define RQF_PREEMPT ((__force req_flags_t)(1 << 8))
85 /* vaguely specified driver internal error. Ignored by the block layer */
86 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
87 /* don't warn about errors */
88 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
89 /* elevator private data attached */
90 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
91 /* account into disk and partition IO statistics */
92 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
93 /* request came from our alloc pool */
94 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
95 /* runtime pm request */
96 #define RQF_PM ((__force req_flags_t)(1 << 15))
97 /* on IO scheduler merge hash */
98 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
99 /* track IO completion time */
100 #define RQF_STATS ((__force req_flags_t)(1 << 17))
101 /* Look at ->special_vec for the actual data payload instead of the
103 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
104 /* The per-zone write lock is held for this request */
105 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
106 /* already slept for hybrid poll */
107 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
108 /* ->timeout has been called, don't expire again */
109 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
111 /* flags that prevent us from merging requests: */
112 #define RQF_NOMERGE_FLAGS \
113 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
116 * Request state for blk-mq.
125 * Try to put the fields that are referenced together in the same cacheline.
127 * If you modify this structure, make sure to update blk_rq_init() and
128 * especially blk_mq_rq_ctx_init() to take care of the added fields.
131 struct request_queue *q;
132 struct blk_mq_ctx *mq_ctx;
133 struct blk_mq_hw_ctx *mq_hctx;
135 unsigned int cmd_flags; /* op and common flags */
136 req_flags_t rq_flags;
141 /* the following two fields are internal, NEVER access directly */
142 unsigned int __data_len; /* total data len */
143 sector_t __sector; /* sector cursor */
148 struct list_head queuelist;
151 * The hash is used inside the scheduler, and killed once the
152 * request reaches the dispatch list. The ipi_list is only used
153 * to queue the request for softirq completion, which is long
154 * after the request has been unhashed (and even removed from
155 * the dispatch list).
158 struct hlist_node hash; /* merge hash */
159 struct list_head ipi_list;
163 * The rb_node is only used inside the io scheduler, requests
164 * are pruned when moved to the dispatch queue. So let the
165 * completion_data share space with the rb_node.
168 struct rb_node rb_node; /* sort/lookup */
169 struct bio_vec special_vec;
170 void *completion_data;
171 int error_count; /* for legacy drivers, don't use */
175 * Three pointers are available for the IO schedulers, if they need
176 * more they have to dynamically allocate it. Flush requests are
177 * never put on the IO scheduler. So let the flush fields share
178 * space with the elevator data.
188 struct list_head list;
189 rq_end_io_fn *saved_end_io;
193 struct gendisk *rq_disk;
194 struct hd_struct *part;
195 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
196 /* Time that the first bio started allocating this request. */
199 /* Time that this request was allocated for this IO. */
201 /* Time that I/O was submitted to the device. */
202 u64 io_start_time_ns;
204 #ifdef CONFIG_BLK_WBT
205 unsigned short wbt_flags;
208 * rq sectors used for blk stats. It has the same value
209 * with blk_rq_sectors(rq), except that it never be zeroed
212 unsigned short stats_sectors;
215 * Number of scatter-gather DMA addr+len pairs after
216 * physical address coalescing is performed.
218 unsigned short nr_phys_segments;
220 #if defined(CONFIG_BLK_DEV_INTEGRITY)
221 unsigned short nr_integrity_segments;
224 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
225 struct bio_crypt_ctx *crypt_ctx;
226 struct blk_ksm_keyslot *crypt_keyslot;
229 unsigned short write_hint;
230 unsigned short ioprio;
232 enum mq_rq_state state;
235 unsigned int timeout;
236 unsigned long deadline;
239 struct __call_single_data csd;
244 * completion callback.
246 rq_end_io_fn *end_io;
250 static inline bool blk_op_is_scsi(unsigned int op)
252 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
255 static inline bool blk_op_is_private(unsigned int op)
257 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
260 static inline bool blk_rq_is_scsi(struct request *rq)
262 return blk_op_is_scsi(req_op(rq));
265 static inline bool blk_rq_is_private(struct request *rq)
267 return blk_op_is_private(req_op(rq));
270 static inline bool blk_rq_is_passthrough(struct request *rq)
272 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
275 static inline bool bio_is_passthrough(struct bio *bio)
277 unsigned op = bio_op(bio);
279 return blk_op_is_scsi(op) || blk_op_is_private(op);
282 static inline unsigned short req_get_ioprio(struct request *req)
287 #include <linux/elevator.h>
289 struct blk_queue_ctx;
293 enum blk_eh_timer_return {
294 BLK_EH_DONE, /* drivers has completed the command */
295 BLK_EH_RESET_TIMER, /* reset timer and try again */
298 enum blk_queue_state {
303 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
304 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
306 #define BLK_SCSI_MAX_CMDS (256)
307 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
310 * Zoned block device models (zoned limit).
312 * Note: This needs to be ordered from the least to the most severe
313 * restrictions for the inheritance in blk_stack_limits() to work.
315 enum blk_zoned_model {
316 BLK_ZONED_NONE = 0, /* Regular block device */
317 BLK_ZONED_HA, /* Host-aware zoned block device */
318 BLK_ZONED_HM, /* Host-managed zoned block device */
321 struct queue_limits {
322 unsigned long bounce_pfn;
323 unsigned long seg_boundary_mask;
324 unsigned long virt_boundary_mask;
326 unsigned int max_hw_sectors;
327 unsigned int max_dev_sectors;
328 unsigned int chunk_sectors;
329 unsigned int max_sectors;
330 unsigned int max_segment_size;
331 unsigned int physical_block_size;
332 unsigned int logical_block_size;
333 unsigned int alignment_offset;
336 unsigned int max_discard_sectors;
337 unsigned int max_hw_discard_sectors;
338 unsigned int max_write_same_sectors;
339 unsigned int max_write_zeroes_sectors;
340 unsigned int max_zone_append_sectors;
341 unsigned int discard_granularity;
342 unsigned int discard_alignment;
344 unsigned short max_segments;
345 unsigned short max_integrity_segments;
346 unsigned short max_discard_segments;
348 unsigned char misaligned;
349 unsigned char discard_misaligned;
350 unsigned char raid_partial_stripes_expensive;
351 enum blk_zoned_model zoned;
354 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
357 void blk_queue_set_zoned(struct gendisk *disk, enum blk_zoned_model model);
359 #ifdef CONFIG_BLK_DEV_ZONED
361 #define BLK_ALL_ZONES ((unsigned int)-1)
362 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
363 unsigned int nr_zones, report_zones_cb cb, void *data);
364 unsigned int blkdev_nr_zones(struct gendisk *disk);
365 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
366 sector_t sectors, sector_t nr_sectors,
368 int blk_revalidate_disk_zones(struct gendisk *disk,
369 void (*update_driver_data)(struct gendisk *disk));
371 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
372 unsigned int cmd, unsigned long arg);
373 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
374 unsigned int cmd, unsigned long arg);
376 #else /* CONFIG_BLK_DEV_ZONED */
378 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
383 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
384 fmode_t mode, unsigned int cmd,
390 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
391 fmode_t mode, unsigned int cmd,
397 #endif /* CONFIG_BLK_DEV_ZONED */
399 struct request_queue {
400 struct request *last_merge;
401 struct elevator_queue *elevator;
403 struct percpu_ref q_usage_counter;
405 struct blk_queue_stats *stats;
406 struct rq_qos *rq_qos;
408 const struct blk_mq_ops *mq_ops;
411 struct blk_mq_ctx __percpu *queue_ctx;
413 unsigned int queue_depth;
415 /* hw dispatch queues */
416 struct blk_mq_hw_ctx **queue_hw_ctx;
417 unsigned int nr_hw_queues;
419 struct backing_dev_info *backing_dev_info;
422 * The queue owner gets to use this for whatever they like.
423 * ll_rw_blk doesn't touch it.
428 * various queue flags, see QUEUE_* below
430 unsigned long queue_flags;
432 * Number of contexts that have called blk_set_pm_only(). If this
433 * counter is above zero then only RQF_PM and RQF_PREEMPT requests are
439 * ida allocated id for this queue. Used to index queues from
445 * queue needs bounce pages for pages above this limit
449 spinlock_t queue_lock;
459 struct kobject *mq_kobj;
461 #ifdef CONFIG_BLK_DEV_INTEGRITY
462 struct blk_integrity integrity;
463 #endif /* CONFIG_BLK_DEV_INTEGRITY */
467 enum rpm_status rpm_status;
468 unsigned int nr_pending;
474 unsigned long nr_requests; /* Max # of requests */
476 unsigned int dma_pad_mask;
477 unsigned int dma_alignment;
479 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
480 /* Inline crypto capabilities */
481 struct blk_keyslot_manager *ksm;
484 unsigned int rq_timeout;
487 struct blk_stat_callback *poll_cb;
488 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
490 struct timer_list timeout;
491 struct work_struct timeout_work;
493 atomic_t nr_active_requests_shared_sbitmap;
495 struct list_head icq_list;
496 #ifdef CONFIG_BLK_CGROUP
497 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
498 struct blkcg_gq *root_blkg;
499 struct list_head blkg_list;
502 struct queue_limits limits;
504 unsigned int required_elevator_features;
506 #ifdef CONFIG_BLK_DEV_ZONED
508 * Zoned block device information for request dispatch control.
509 * nr_zones is the total number of zones of the device. This is always
510 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
511 * bits which indicates if a zone is conventional (bit set) or
512 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
513 * bits which indicates if a zone is write locked, that is, if a write
514 * request targeting the zone was dispatched. All three fields are
515 * initialized by the low level device driver (e.g. scsi/sd.c).
516 * Stacking drivers (device mappers) may or may not initialize
519 * Reads of this information must be protected with blk_queue_enter() /
520 * blk_queue_exit(). Modifying this information is only allowed while
521 * no requests are being processed. See also blk_mq_freeze_queue() and
522 * blk_mq_unfreeze_queue().
524 unsigned int nr_zones;
525 unsigned long *conv_zones_bitmap;
526 unsigned long *seq_zones_wlock;
527 unsigned int max_open_zones;
528 unsigned int max_active_zones;
529 #endif /* CONFIG_BLK_DEV_ZONED */
534 unsigned int sg_timeout;
535 unsigned int sg_reserved_size;
537 struct mutex debugfs_mutex;
538 #ifdef CONFIG_BLK_DEV_IO_TRACE
539 struct blk_trace __rcu *blk_trace;
542 * for flush operations
544 struct blk_flush_queue *fq;
546 struct list_head requeue_list;
547 spinlock_t requeue_lock;
548 struct delayed_work requeue_work;
550 struct mutex sysfs_lock;
551 struct mutex sysfs_dir_lock;
554 * for reusing dead hctx instance in case of updating
557 struct list_head unused_hctx_list;
558 spinlock_t unused_hctx_lock;
562 #if defined(CONFIG_BLK_DEV_BSG)
563 struct bsg_class_device bsg_dev;
566 #ifdef CONFIG_BLK_DEV_THROTTLING
568 struct throtl_data *td;
570 struct rcu_head rcu_head;
571 wait_queue_head_t mq_freeze_wq;
573 * Protect concurrent access to q_usage_counter by
574 * percpu_ref_kill() and percpu_ref_reinit().
576 struct mutex mq_freeze_lock;
578 struct blk_mq_tag_set *tag_set;
579 struct list_head tag_set_list;
580 struct bio_set bio_split;
582 struct dentry *debugfs_dir;
584 #ifdef CONFIG_BLK_DEBUG_FS
585 struct dentry *sched_debugfs_dir;
586 struct dentry *rqos_debugfs_dir;
589 bool mq_sysfs_init_done;
593 #define BLK_MAX_WRITE_HINTS 5
594 u64 write_hints[BLK_MAX_WRITE_HINTS];
597 /* Keep blk_queue_flag_name[] in sync with the definitions below */
598 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
599 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
600 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
601 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
602 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
603 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
604 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
605 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
606 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
607 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
608 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
609 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
610 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
611 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
612 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
613 #define QUEUE_FLAG_STABLE_WRITES 15 /* don't modify blks until WB is done */
614 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
615 #define QUEUE_FLAG_WC 17 /* Write back caching */
616 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
617 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
618 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
619 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
620 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
621 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
622 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
623 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
624 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
625 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
626 #define QUEUE_FLAG_HCTX_ACTIVE 28 /* at least one blk-mq hctx is active */
627 #define QUEUE_FLAG_NOWAIT 29 /* device supports NOWAIT */
629 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
630 (1 << QUEUE_FLAG_SAME_COMP) | \
631 (1 << QUEUE_FLAG_NOWAIT))
633 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
634 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
635 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
637 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
638 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
639 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
640 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
641 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
642 #define blk_queue_noxmerges(q) \
643 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
644 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
645 #define blk_queue_stable_writes(q) \
646 test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags)
647 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
648 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
649 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
650 #define blk_queue_zone_resetall(q) \
651 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
652 #define blk_queue_secure_erase(q) \
653 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
654 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
655 #define blk_queue_scsi_passthrough(q) \
656 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
657 #define blk_queue_pci_p2pdma(q) \
658 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
659 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
660 #define blk_queue_rq_alloc_time(q) \
661 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
663 #define blk_queue_rq_alloc_time(q) false
666 #define blk_noretry_request(rq) \
667 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
668 REQ_FAILFAST_DRIVER))
669 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
670 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
671 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
672 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
673 #define blk_queue_nowait(q) test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
675 extern void blk_set_pm_only(struct request_queue *q);
676 extern void blk_clear_pm_only(struct request_queue *q);
678 static inline bool blk_account_rq(struct request *rq)
680 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
683 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
685 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
687 #define rq_dma_dir(rq) \
688 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
690 #define dma_map_bvec(dev, bv, dir, attrs) \
691 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
694 static inline bool queue_is_mq(struct request_queue *q)
699 static inline enum blk_zoned_model
700 blk_queue_zoned_model(struct request_queue *q)
702 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED))
703 return q->limits.zoned;
704 return BLK_ZONED_NONE;
707 static inline bool blk_queue_is_zoned(struct request_queue *q)
709 switch (blk_queue_zoned_model(q)) {
718 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
720 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
723 #ifdef CONFIG_BLK_DEV_ZONED
724 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
726 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
729 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
732 if (!blk_queue_is_zoned(q))
734 return sector >> ilog2(q->limits.chunk_sectors);
737 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
740 if (!blk_queue_is_zoned(q))
742 if (!q->conv_zones_bitmap)
744 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
747 static inline void blk_queue_max_open_zones(struct request_queue *q,
748 unsigned int max_open_zones)
750 q->max_open_zones = max_open_zones;
753 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
755 return q->max_open_zones;
758 static inline void blk_queue_max_active_zones(struct request_queue *q,
759 unsigned int max_active_zones)
761 q->max_active_zones = max_active_zones;
764 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
766 return q->max_active_zones;
768 #else /* CONFIG_BLK_DEV_ZONED */
769 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
773 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
778 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
783 static inline unsigned int queue_max_open_zones(const struct request_queue *q)
787 static inline unsigned int queue_max_active_zones(const struct request_queue *q)
791 #endif /* CONFIG_BLK_DEV_ZONED */
793 static inline bool rq_is_sync(struct request *rq)
795 return op_is_sync(rq->cmd_flags);
798 static inline bool rq_mergeable(struct request *rq)
800 if (blk_rq_is_passthrough(rq))
803 if (req_op(rq) == REQ_OP_FLUSH)
806 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
809 if (req_op(rq) == REQ_OP_ZONE_APPEND)
812 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
814 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
820 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
822 if (bio_page(a) == bio_page(b) &&
823 bio_offset(a) == bio_offset(b))
829 static inline unsigned int blk_queue_depth(struct request_queue *q)
832 return q->queue_depth;
834 return q->nr_requests;
837 extern unsigned long blk_max_low_pfn, blk_max_pfn;
840 * standard bounce addresses:
842 * BLK_BOUNCE_HIGH : bounce all highmem pages
843 * BLK_BOUNCE_ANY : don't bounce anything
844 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
847 #if BITS_PER_LONG == 32
848 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
850 #define BLK_BOUNCE_HIGH -1ULL
852 #define BLK_BOUNCE_ANY (-1ULL)
853 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
856 * default timeout for SG_IO if none specified
858 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
859 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
865 unsigned long offset;
870 struct req_iterator {
871 struct bvec_iter iter;
875 /* This should not be used directly - use rq_for_each_segment */
876 #define for_each_bio(_bio) \
877 for (; _bio; _bio = _bio->bi_next)
878 #define __rq_for_each_bio(_bio, rq) \
880 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
882 #define rq_for_each_segment(bvl, _rq, _iter) \
883 __rq_for_each_bio(_iter.bio, _rq) \
884 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
886 #define rq_for_each_bvec(bvl, _rq, _iter) \
887 __rq_for_each_bio(_iter.bio, _rq) \
888 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
890 #define rq_iter_last(bvec, _iter) \
891 (_iter.bio->bi_next == NULL && \
892 bio_iter_last(bvec, _iter.iter))
894 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
895 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
897 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
898 extern void rq_flush_dcache_pages(struct request *rq);
900 static inline void rq_flush_dcache_pages(struct request *rq)
905 extern int blk_register_queue(struct gendisk *disk);
906 extern void blk_unregister_queue(struct gendisk *disk);
907 blk_qc_t submit_bio_noacct(struct bio *bio);
908 extern void blk_rq_init(struct request_queue *q, struct request *rq);
909 extern void blk_put_request(struct request *);
910 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
911 blk_mq_req_flags_t flags);
912 extern int blk_lld_busy(struct request_queue *q);
913 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
914 struct bio_set *bs, gfp_t gfp_mask,
915 int (*bio_ctr)(struct bio *, struct bio *, void *),
917 extern void blk_rq_unprep_clone(struct request *rq);
918 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
920 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
921 extern void blk_queue_split(struct bio **);
922 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
923 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
924 unsigned int, void __user *);
925 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
926 unsigned int, void __user *);
927 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
928 struct scsi_ioctl_command __user *);
929 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
930 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
932 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
933 extern void blk_queue_exit(struct request_queue *q);
934 extern void blk_sync_queue(struct request_queue *q);
935 extern int blk_rq_map_user(struct request_queue *, struct request *,
936 struct rq_map_data *, void __user *, unsigned long,
938 extern int blk_rq_unmap_user(struct bio *);
939 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
940 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
941 struct rq_map_data *, const struct iov_iter *,
943 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
944 struct request *, int);
945 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
946 struct request *, int, rq_end_io_fn *);
948 /* Helper to convert REQ_OP_XXX to its string format XXX */
949 extern const char *blk_op_str(unsigned int op);
951 int blk_status_to_errno(blk_status_t status);
952 blk_status_t errno_to_blk_status(int errno);
954 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
956 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
958 return bdev->bd_disk->queue; /* this is never NULL */
962 * The basic unit of block I/O is a sector. It is used in a number of contexts
963 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
964 * bytes. Variables of type sector_t represent an offset or size that is a
965 * multiple of 512 bytes. Hence these two constants.
968 #define SECTOR_SHIFT 9
971 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
975 * blk_rq_pos() : the current sector
976 * blk_rq_bytes() : bytes left in the entire request
977 * blk_rq_cur_bytes() : bytes left in the current segment
978 * blk_rq_err_bytes() : bytes left till the next error boundary
979 * blk_rq_sectors() : sectors left in the entire request
980 * blk_rq_cur_sectors() : sectors left in the current segment
981 * blk_rq_stats_sectors() : sectors of the entire request used for stats
983 static inline sector_t blk_rq_pos(const struct request *rq)
988 static inline unsigned int blk_rq_bytes(const struct request *rq)
990 return rq->__data_len;
993 static inline int blk_rq_cur_bytes(const struct request *rq)
995 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
998 extern unsigned int blk_rq_err_bytes(const struct request *rq);
1000 static inline unsigned int blk_rq_sectors(const struct request *rq)
1002 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
1005 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
1007 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
1010 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
1012 return rq->stats_sectors;
1015 #ifdef CONFIG_BLK_DEV_ZONED
1017 /* Helper to convert BLK_ZONE_ZONE_XXX to its string format XXX */
1018 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond);
1020 static inline unsigned int blk_rq_zone_no(struct request *rq)
1022 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
1025 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
1027 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
1029 #endif /* CONFIG_BLK_DEV_ZONED */
1032 * Some commands like WRITE SAME have a payload or data transfer size which
1033 * is different from the size of the request. Any driver that supports such
1034 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
1035 * calculate the data transfer size.
1037 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
1039 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1040 return rq->special_vec.bv_len;
1041 return blk_rq_bytes(rq);
1045 * Return the first full biovec in the request. The caller needs to check that
1046 * there are any bvecs before calling this helper.
1048 static inline struct bio_vec req_bvec(struct request *rq)
1050 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1051 return rq->special_vec;
1052 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1055 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1058 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1059 return min(q->limits.max_discard_sectors,
1060 UINT_MAX >> SECTOR_SHIFT);
1062 if (unlikely(op == REQ_OP_WRITE_SAME))
1063 return q->limits.max_write_same_sectors;
1065 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1066 return q->limits.max_write_zeroes_sectors;
1068 return q->limits.max_sectors;
1072 * Return maximum size of a request at given offset. Only valid for
1073 * file system requests.
1075 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1078 unsigned int chunk_sectors = q->limits.chunk_sectors;
1081 return q->limits.max_sectors;
1083 if (likely(is_power_of_2(chunk_sectors)))
1084 chunk_sectors -= offset & (chunk_sectors - 1);
1086 chunk_sectors -= sector_div(offset, chunk_sectors);
1088 return min(q->limits.max_sectors, chunk_sectors);
1091 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1094 struct request_queue *q = rq->q;
1096 if (blk_rq_is_passthrough(rq))
1097 return q->limits.max_hw_sectors;
1099 if (!q->limits.chunk_sectors ||
1100 req_op(rq) == REQ_OP_DISCARD ||
1101 req_op(rq) == REQ_OP_SECURE_ERASE)
1102 return blk_queue_get_max_sectors(q, req_op(rq));
1104 return min(blk_max_size_offset(q, offset),
1105 blk_queue_get_max_sectors(q, req_op(rq)));
1108 static inline unsigned int blk_rq_count_bios(struct request *rq)
1110 unsigned int nr_bios = 0;
1113 __rq_for_each_bio(bio, rq)
1119 void blk_steal_bios(struct bio_list *list, struct request *rq);
1122 * Request completion related functions.
1124 * blk_update_request() completes given number of bytes and updates
1125 * the request without completing it.
1127 extern bool blk_update_request(struct request *rq, blk_status_t error,
1128 unsigned int nr_bytes);
1130 extern void blk_abort_request(struct request *);
1133 * Access functions for manipulating queue properties
1135 extern void blk_cleanup_queue(struct request_queue *);
1136 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1137 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1138 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1139 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1140 extern void blk_queue_max_discard_segments(struct request_queue *,
1142 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1143 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1144 unsigned int max_discard_sectors);
1145 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1146 unsigned int max_write_same_sectors);
1147 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1148 unsigned int max_write_same_sectors);
1149 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1150 extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
1151 unsigned int max_zone_append_sectors);
1152 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1153 extern void blk_queue_alignment_offset(struct request_queue *q,
1154 unsigned int alignment);
1155 void blk_queue_update_readahead(struct request_queue *q);
1156 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1157 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1158 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1159 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1160 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1161 extern void blk_set_default_limits(struct queue_limits *lim);
1162 extern void blk_set_stacking_limits(struct queue_limits *lim);
1163 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1165 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1167 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1168 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1169 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1170 extern void blk_queue_dma_alignment(struct request_queue *, int);
1171 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1172 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1173 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1174 extern void blk_queue_required_elevator_features(struct request_queue *q,
1175 unsigned int features);
1176 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1177 struct device *dev);
1180 * Number of physical segments as sent to the device.
1182 * Normally this is the number of discontiguous data segments sent by the
1183 * submitter. But for data-less command like discard we might have no
1184 * actual data segments submitted, but the driver might have to add it's
1185 * own special payload. In that case we still return 1 here so that this
1186 * special payload will be mapped.
1188 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1190 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1192 return rq->nr_phys_segments;
1196 * Number of discard segments (or ranges) the driver needs to fill in.
1197 * Each discard bio merged into a request is counted as one segment.
1199 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1201 return max_t(unsigned short, rq->nr_phys_segments, 1);
1204 int __blk_rq_map_sg(struct request_queue *q, struct request *rq,
1205 struct scatterlist *sglist, struct scatterlist **last_sg);
1206 static inline int blk_rq_map_sg(struct request_queue *q, struct request *rq,
1207 struct scatterlist *sglist)
1209 struct scatterlist *last_sg = NULL;
1211 return __blk_rq_map_sg(q, rq, sglist, &last_sg);
1213 extern void blk_dump_rq_flags(struct request *, char *);
1215 bool __must_check blk_get_queue(struct request_queue *);
1216 struct request_queue *blk_alloc_queue(int node_id);
1217 extern void blk_put_queue(struct request_queue *);
1218 extern void blk_set_queue_dying(struct request_queue *);
1222 * blk_plug permits building a queue of related requests by holding the I/O
1223 * fragments for a short period. This allows merging of sequential requests
1224 * into single larger request. As the requests are moved from a per-task list to
1225 * the device's request_queue in a batch, this results in improved scalability
1226 * as the lock contention for request_queue lock is reduced.
1228 * It is ok not to disable preemption when adding the request to the plug list
1229 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1230 * the plug list when the task sleeps by itself. For details, please see
1231 * schedule() where blk_schedule_flush_plug() is called.
1234 struct list_head mq_list; /* blk-mq requests */
1235 struct list_head cb_list; /* md requires an unplug callback */
1236 unsigned short rq_count;
1237 bool multiple_queues;
1240 #define BLK_MAX_REQUEST_COUNT 16
1241 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1244 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1245 struct blk_plug_cb {
1246 struct list_head list;
1247 blk_plug_cb_fn callback;
1250 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1251 void *data, int size);
1252 extern void blk_start_plug(struct blk_plug *);
1253 extern void blk_finish_plug(struct blk_plug *);
1254 extern void blk_flush_plug_list(struct blk_plug *, bool);
1256 static inline void blk_flush_plug(struct task_struct *tsk)
1258 struct blk_plug *plug = tsk->plug;
1261 blk_flush_plug_list(plug, false);
1264 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1266 struct blk_plug *plug = tsk->plug;
1269 blk_flush_plug_list(plug, true);
1272 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1274 struct blk_plug *plug = tsk->plug;
1277 (!list_empty(&plug->mq_list) ||
1278 !list_empty(&plug->cb_list));
1281 int blkdev_issue_flush(struct block_device *, gfp_t);
1282 long nr_blockdev_pages(void);
1283 #else /* CONFIG_BLOCK */
1287 static inline void blk_start_plug(struct blk_plug *plug)
1291 static inline void blk_finish_plug(struct blk_plug *plug)
1295 static inline void blk_flush_plug(struct task_struct *task)
1299 static inline void blk_schedule_flush_plug(struct task_struct *task)
1304 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1309 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask)
1314 static inline long nr_blockdev_pages(void)
1318 #endif /* CONFIG_BLOCK */
1320 extern void blk_io_schedule(void);
1322 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1323 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1325 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1327 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1328 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1329 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1330 sector_t nr_sects, gfp_t gfp_mask, int flags,
1333 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1334 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1336 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1337 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1339 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1340 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1342 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1343 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1345 return blkdev_issue_discard(sb->s_bdev,
1346 block << (sb->s_blocksize_bits -
1348 nr_blocks << (sb->s_blocksize_bits -
1352 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1353 sector_t nr_blocks, gfp_t gfp_mask)
1355 return blkdev_issue_zeroout(sb->s_bdev,
1356 block << (sb->s_blocksize_bits -
1358 nr_blocks << (sb->s_blocksize_bits -
1363 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1365 static inline bool bdev_is_partition(struct block_device *bdev)
1367 return bdev->bd_partno;
1370 enum blk_default_limits {
1371 BLK_MAX_SEGMENTS = 128,
1372 BLK_SAFE_MAX_SECTORS = 255,
1373 BLK_DEF_MAX_SECTORS = 2560,
1374 BLK_MAX_SEGMENT_SIZE = 65536,
1375 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1378 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1380 return q->limits.seg_boundary_mask;
1383 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1385 return q->limits.virt_boundary_mask;
1388 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1390 return q->limits.max_sectors;
1393 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1395 return q->limits.max_hw_sectors;
1398 static inline unsigned short queue_max_segments(const struct request_queue *q)
1400 return q->limits.max_segments;
1403 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1405 return q->limits.max_discard_segments;
1408 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1410 return q->limits.max_segment_size;
1413 static inline unsigned int queue_max_zone_append_sectors(const struct request_queue *q)
1416 const struct queue_limits *l = &q->limits;
1418 return min(l->max_zone_append_sectors, l->max_sectors);
1421 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1425 if (q && q->limits.logical_block_size)
1426 retval = q->limits.logical_block_size;
1431 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1433 return queue_logical_block_size(bdev_get_queue(bdev));
1436 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1438 return q->limits.physical_block_size;
1441 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1443 return queue_physical_block_size(bdev_get_queue(bdev));
1446 static inline unsigned int queue_io_min(const struct request_queue *q)
1448 return q->limits.io_min;
1451 static inline int bdev_io_min(struct block_device *bdev)
1453 return queue_io_min(bdev_get_queue(bdev));
1456 static inline unsigned int queue_io_opt(const struct request_queue *q)
1458 return q->limits.io_opt;
1461 static inline int bdev_io_opt(struct block_device *bdev)
1463 return queue_io_opt(bdev_get_queue(bdev));
1466 static inline int queue_alignment_offset(const struct request_queue *q)
1468 if (q->limits.misaligned)
1471 return q->limits.alignment_offset;
1474 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1476 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1477 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1480 return (granularity + lim->alignment_offset - alignment) % granularity;
1483 static inline int bdev_alignment_offset(struct block_device *bdev)
1485 struct request_queue *q = bdev_get_queue(bdev);
1487 if (q->limits.misaligned)
1489 if (bdev_is_partition(bdev))
1490 return queue_limit_alignment_offset(&q->limits,
1491 bdev->bd_part->start_sect);
1492 return q->limits.alignment_offset;
1495 static inline int queue_discard_alignment(const struct request_queue *q)
1497 if (q->limits.discard_misaligned)
1500 return q->limits.discard_alignment;
1503 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1505 unsigned int alignment, granularity, offset;
1507 if (!lim->max_discard_sectors)
1510 /* Why are these in bytes, not sectors? */
1511 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1512 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1516 /* Offset of the partition start in 'granularity' sectors */
1517 offset = sector_div(sector, granularity);
1519 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1520 offset = (granularity + alignment - offset) % granularity;
1522 /* Turn it back into bytes, gaah */
1523 return offset << SECTOR_SHIFT;
1526 static inline int bdev_discard_alignment(struct block_device *bdev)
1528 struct request_queue *q = bdev_get_queue(bdev);
1530 if (bdev_is_partition(bdev))
1531 return queue_limit_discard_alignment(&q->limits,
1532 bdev->bd_part->start_sect);
1533 return q->limits.discard_alignment;
1536 static inline unsigned int bdev_write_same(struct block_device *bdev)
1538 struct request_queue *q = bdev_get_queue(bdev);
1541 return q->limits.max_write_same_sectors;
1546 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1548 struct request_queue *q = bdev_get_queue(bdev);
1551 return q->limits.max_write_zeroes_sectors;
1556 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1558 struct request_queue *q = bdev_get_queue(bdev);
1561 return blk_queue_zoned_model(q);
1563 return BLK_ZONED_NONE;
1566 static inline bool bdev_is_zoned(struct block_device *bdev)
1568 struct request_queue *q = bdev_get_queue(bdev);
1571 return blk_queue_is_zoned(q);
1576 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1578 struct request_queue *q = bdev_get_queue(bdev);
1581 return blk_queue_zone_sectors(q);
1585 static inline unsigned int bdev_max_open_zones(struct block_device *bdev)
1587 struct request_queue *q = bdev_get_queue(bdev);
1590 return queue_max_open_zones(q);
1594 static inline unsigned int bdev_max_active_zones(struct block_device *bdev)
1596 struct request_queue *q = bdev_get_queue(bdev);
1599 return queue_max_active_zones(q);
1603 static inline int queue_dma_alignment(const struct request_queue *q)
1605 return q ? q->dma_alignment : 511;
1608 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1611 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1612 return !(addr & alignment) && !(len & alignment);
1615 /* assumes size > 256 */
1616 static inline unsigned int blksize_bits(unsigned int size)
1618 unsigned int bits = 8;
1622 } while (size > 256);
1626 static inline unsigned int block_size(struct block_device *bdev)
1628 return 1 << bdev->bd_inode->i_blkbits;
1631 int kblockd_schedule_work(struct work_struct *work);
1632 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1634 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1635 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1636 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1637 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1639 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1641 enum blk_integrity_flags {
1642 BLK_INTEGRITY_VERIFY = 1 << 0,
1643 BLK_INTEGRITY_GENERATE = 1 << 1,
1644 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1645 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1648 struct blk_integrity_iter {
1652 unsigned int data_size;
1653 unsigned short interval;
1654 const char *disk_name;
1657 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1658 typedef void (integrity_prepare_fn) (struct request *);
1659 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1661 struct blk_integrity_profile {
1662 integrity_processing_fn *generate_fn;
1663 integrity_processing_fn *verify_fn;
1664 integrity_prepare_fn *prepare_fn;
1665 integrity_complete_fn *complete_fn;
1669 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1670 extern void blk_integrity_unregister(struct gendisk *);
1671 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1672 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1673 struct scatterlist *);
1674 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1676 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1678 struct blk_integrity *bi = &disk->queue->integrity;
1687 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1689 return blk_get_integrity(bdev->bd_disk);
1693 blk_integrity_queue_supports_integrity(struct request_queue *q)
1695 return q->integrity.profile;
1698 static inline bool blk_integrity_rq(struct request *rq)
1700 return rq->cmd_flags & REQ_INTEGRITY;
1703 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1706 q->limits.max_integrity_segments = segs;
1709 static inline unsigned short
1710 queue_max_integrity_segments(const struct request_queue *q)
1712 return q->limits.max_integrity_segments;
1716 * bio_integrity_intervals - Return number of integrity intervals for a bio
1717 * @bi: blk_integrity profile for device
1718 * @sectors: Size of the bio in 512-byte sectors
1720 * Description: The block layer calculates everything in 512 byte
1721 * sectors but integrity metadata is done in terms of the data integrity
1722 * interval size of the storage device. Convert the block layer sectors
1723 * to the appropriate number of integrity intervals.
1725 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1726 unsigned int sectors)
1728 return sectors >> (bi->interval_exp - 9);
1731 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1732 unsigned int sectors)
1734 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1738 * Return the first bvec that contains integrity data. Only drivers that are
1739 * limited to a single integrity segment should use this helper.
1741 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1743 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1745 return rq->bio->bi_integrity->bip_vec;
1748 #else /* CONFIG_BLK_DEV_INTEGRITY */
1751 struct block_device;
1753 struct blk_integrity;
1755 static inline int blk_integrity_rq(struct request *rq)
1759 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1764 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1766 struct scatterlist *s)
1770 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1774 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1779 blk_integrity_queue_supports_integrity(struct request_queue *q)
1783 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1787 static inline void blk_integrity_register(struct gendisk *d,
1788 struct blk_integrity *b)
1791 static inline void blk_integrity_unregister(struct gendisk *d)
1794 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1798 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1803 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1804 unsigned int sectors)
1809 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1810 unsigned int sectors)
1815 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1820 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1822 #ifdef CONFIG_BLK_INLINE_ENCRYPTION
1824 bool blk_ksm_register(struct blk_keyslot_manager *ksm, struct request_queue *q);
1826 void blk_ksm_unregister(struct request_queue *q);
1828 #else /* CONFIG_BLK_INLINE_ENCRYPTION */
1830 static inline bool blk_ksm_register(struct blk_keyslot_manager *ksm,
1831 struct request_queue *q)
1836 static inline void blk_ksm_unregister(struct request_queue *q) { }
1838 #endif /* CONFIG_BLK_INLINE_ENCRYPTION */
1841 struct block_device_operations {
1842 blk_qc_t (*submit_bio) (struct bio *bio);
1843 int (*open) (struct block_device *, fmode_t);
1844 void (*release) (struct gendisk *, fmode_t);
1845 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1846 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1847 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1848 unsigned int (*check_events) (struct gendisk *disk,
1849 unsigned int clearing);
1850 void (*unlock_native_capacity) (struct gendisk *);
1851 int (*revalidate_disk) (struct gendisk *);
1852 int (*getgeo)(struct block_device *, struct hd_geometry *);
1853 int (*set_read_only)(struct block_device *bdev, bool ro);
1854 /* this callback is with swap_lock and sometimes page table lock held */
1855 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1856 int (*report_zones)(struct gendisk *, sector_t sector,
1857 unsigned int nr_zones, report_zones_cb cb, void *data);
1858 char *(*devnode)(struct gendisk *disk, umode_t *mode);
1859 struct module *owner;
1860 const struct pr_ops *pr_ops;
1863 #ifdef CONFIG_COMPAT
1864 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1865 unsigned int, unsigned long);
1867 #define blkdev_compat_ptr_ioctl NULL
1870 extern int 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_part(struct hd_struct *part);
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 */