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>
10 #include <linux/major.h>
11 #include <linux/genhd.h>
12 #include <linux/list.h>
13 #include <linux/llist.h>
14 #include <linux/timer.h>
15 #include <linux/workqueue.h>
16 #include <linux/pagemap.h>
17 #include <linux/backing-dev-defs.h>
18 #include <linux/wait.h>
19 #include <linux/mempool.h>
20 #include <linux/pfn.h>
21 #include <linux/bio.h>
22 #include <linux/stringify.h>
23 #include <linux/gfp.h>
24 #include <linux/bsg.h>
25 #include <linux/smp.h>
26 #include <linux/rcupdate.h>
27 #include <linux/percpu-refcount.h>
28 #include <linux/scatterlist.h>
29 #include <linux/blkzoned.h>
32 struct scsi_ioctl_command;
35 struct elevator_queue;
41 struct blk_flush_queue;
44 struct blk_queue_stats;
45 struct blk_stat_callback;
47 #define BLKDEV_MIN_RQ 4
48 #define BLKDEV_MAX_RQ 128 /* Default maximum */
50 /* Must be consistent with blk_mq_poll_stats_bkt() */
51 #define BLK_MQ_POLL_STATS_BKTS 16
53 /* Doing classic polling */
54 #define BLK_MQ_POLL_CLASSIC -1
57 * Maximum number of blkcg policies allowed to be registered concurrently.
58 * Defined here to simplify include dependency.
60 #define BLKCG_MAX_POLS 5
62 static inline int blk_validate_block_size(unsigned int bsize)
64 if (bsize < 512 || bsize > PAGE_SIZE || !is_power_of_2(bsize))
70 typedef void (rq_end_io_fn)(struct request *, blk_status_t);
74 typedef __u32 __bitwise req_flags_t;
76 /* elevator knows about this request */
77 #define RQF_SORTED ((__force req_flags_t)(1 << 0))
78 /* drive already may have started this one */
79 #define RQF_STARTED ((__force req_flags_t)(1 << 1))
80 /* may not be passed by ioscheduler */
81 #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3))
82 /* request for flush sequence */
83 #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4))
84 /* merge of different types, fail separately */
85 #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5))
86 /* track inflight for MQ */
87 #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6))
88 /* don't call prep for this one */
89 #define RQF_DONTPREP ((__force req_flags_t)(1 << 7))
90 /* set for "ide_preempt" requests and also for requests for which the SCSI
91 "quiesce" state must be ignored. */
92 #define RQF_PREEMPT ((__force req_flags_t)(1 << 8))
93 /* contains copies of user pages */
94 #define RQF_COPY_USER ((__force req_flags_t)(1 << 9))
95 /* vaguely specified driver internal error. Ignored by the block layer */
96 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
97 /* don't warn about errors */
98 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
99 /* elevator private data attached */
100 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
101 /* account into disk and partition IO statistics */
102 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
103 /* request came from our alloc pool */
104 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
105 /* runtime pm request */
106 #define RQF_PM ((__force req_flags_t)(1 << 15))
107 /* on IO scheduler merge hash */
108 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
109 /* track IO completion time */
110 #define RQF_STATS ((__force req_flags_t)(1 << 17))
111 /* Look at ->special_vec for the actual data payload instead of the
113 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
114 /* The per-zone write lock is held for this request */
115 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
116 /* already slept for hybrid poll */
117 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
118 /* ->timeout has been called, don't expire again */
119 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
121 /* flags that prevent us from merging requests: */
122 #define RQF_NOMERGE_FLAGS \
123 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
126 * Request state for blk-mq.
135 * Try to put the fields that are referenced together in the same cacheline.
137 * If you modify this structure, make sure to update blk_rq_init() and
138 * especially blk_mq_rq_ctx_init() to take care of the added fields.
141 struct request_queue *q;
142 struct blk_mq_ctx *mq_ctx;
143 struct blk_mq_hw_ctx *mq_hctx;
145 unsigned int cmd_flags; /* op and common flags */
146 req_flags_t rq_flags;
151 /* the following two fields are internal, NEVER access directly */
152 unsigned int __data_len; /* total data len */
153 sector_t __sector; /* sector cursor */
158 struct list_head queuelist;
161 * The hash is used inside the scheduler, and killed once the
162 * request reaches the dispatch list. The ipi_list is only used
163 * to queue the request for softirq completion, which is long
164 * after the request has been unhashed (and even removed from
165 * the dispatch list).
168 struct hlist_node hash; /* merge hash */
169 struct list_head ipi_list;
173 * The rb_node is only used inside the io scheduler, requests
174 * are pruned when moved to the dispatch queue. So let the
175 * completion_data share space with the rb_node.
178 struct rb_node rb_node; /* sort/lookup */
179 struct bio_vec special_vec;
180 void *completion_data;
181 int error_count; /* for legacy drivers, don't use */
185 * Three pointers are available for the IO schedulers, if they need
186 * more they have to dynamically allocate it. Flush requests are
187 * never put on the IO scheduler. So let the flush fields share
188 * space with the elevator data.
198 struct list_head list;
199 rq_end_io_fn *saved_end_io;
203 struct gendisk *rq_disk;
204 struct hd_struct *part;
205 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
206 /* Time that the first bio started allocating this request. */
209 /* Time that this request was allocated for this IO. */
211 /* Time that I/O was submitted to the device. */
212 u64 io_start_time_ns;
214 #ifdef CONFIG_BLK_WBT
215 unsigned short wbt_flags;
218 * rq sectors used for blk stats. It has the same value
219 * with blk_rq_sectors(rq), except that it never be zeroed
222 unsigned short stats_sectors;
225 * Number of scatter-gather DMA addr+len pairs after
226 * physical address coalescing is performed.
228 unsigned short nr_phys_segments;
230 #if defined(CONFIG_BLK_DEV_INTEGRITY)
231 unsigned short nr_integrity_segments;
234 unsigned short write_hint;
235 unsigned short ioprio;
237 unsigned int extra_len; /* length of alignment and padding */
239 enum mq_rq_state state;
242 unsigned int timeout;
243 unsigned long deadline;
246 struct __call_single_data csd;
251 * completion callback.
253 rq_end_io_fn *end_io;
257 static inline bool blk_op_is_scsi(unsigned int op)
259 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
262 static inline bool blk_op_is_private(unsigned int op)
264 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
267 static inline bool blk_rq_is_scsi(struct request *rq)
269 return blk_op_is_scsi(req_op(rq));
272 static inline bool blk_rq_is_private(struct request *rq)
274 return blk_op_is_private(req_op(rq));
277 static inline bool blk_rq_is_passthrough(struct request *rq)
279 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
282 static inline bool bio_is_passthrough(struct bio *bio)
284 unsigned op = bio_op(bio);
286 return blk_op_is_scsi(op) || blk_op_is_private(op);
289 static inline unsigned short req_get_ioprio(struct request *req)
294 #include <linux/elevator.h>
296 struct blk_queue_ctx;
298 typedef blk_qc_t (make_request_fn) (struct request_queue *q, struct bio *bio);
301 typedef int (dma_drain_needed_fn)(struct request *);
303 enum blk_eh_timer_return {
304 BLK_EH_DONE, /* drivers has completed the command */
305 BLK_EH_RESET_TIMER, /* reset timer and try again */
308 enum blk_queue_state {
313 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
314 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
316 #define BLK_SCSI_MAX_CMDS (256)
317 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
320 * Zoned block device models (zoned limit).
322 enum blk_zoned_model {
323 BLK_ZONED_NONE, /* Regular block device */
324 BLK_ZONED_HA, /* Host-aware zoned block device */
325 BLK_ZONED_HM, /* Host-managed zoned block device */
328 struct queue_limits {
329 unsigned long bounce_pfn;
330 unsigned long seg_boundary_mask;
331 unsigned long virt_boundary_mask;
333 unsigned int max_hw_sectors;
334 unsigned int max_dev_sectors;
335 unsigned int chunk_sectors;
336 unsigned int max_sectors;
337 unsigned int max_segment_size;
338 unsigned int physical_block_size;
339 unsigned int logical_block_size;
340 unsigned int alignment_offset;
343 unsigned int max_discard_sectors;
344 unsigned int max_hw_discard_sectors;
345 unsigned int max_write_same_sectors;
346 unsigned int max_write_zeroes_sectors;
347 unsigned int discard_granularity;
348 unsigned int discard_alignment;
350 unsigned short max_segments;
351 unsigned short max_integrity_segments;
352 unsigned short max_discard_segments;
354 unsigned char misaligned;
355 unsigned char discard_misaligned;
356 unsigned char raid_partial_stripes_expensive;
357 enum blk_zoned_model zoned;
360 #ifdef CONFIG_BLK_DEV_ZONED
363 * Maximum number of zones to report with a single report zones command.
365 #define BLK_ZONED_REPORT_MAX_ZONES 8192U
367 extern unsigned int blkdev_nr_zones(struct block_device *bdev);
368 extern int blkdev_report_zones(struct block_device *bdev,
369 sector_t sector, struct blk_zone *zones,
370 unsigned int *nr_zones);
371 extern int blkdev_reset_zones(struct block_device *bdev, sector_t sectors,
372 sector_t nr_sectors, gfp_t gfp_mask);
373 extern int blk_revalidate_disk_zones(struct gendisk *disk);
375 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
376 unsigned int cmd, unsigned long arg);
377 extern int blkdev_reset_zones_ioctl(struct block_device *bdev, fmode_t mode,
378 unsigned int cmd, unsigned long arg);
380 #else /* CONFIG_BLK_DEV_ZONED */
382 static inline unsigned int blkdev_nr_zones(struct block_device *bdev)
387 static inline int blk_revalidate_disk_zones(struct gendisk *disk)
392 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
393 fmode_t mode, unsigned int cmd,
399 static inline int blkdev_reset_zones_ioctl(struct block_device *bdev,
400 fmode_t mode, unsigned int cmd,
406 #endif /* CONFIG_BLK_DEV_ZONED */
408 struct request_queue {
409 struct request *last_merge;
410 struct elevator_queue *elevator;
412 struct blk_queue_stats *stats;
413 struct rq_qos *rq_qos;
415 make_request_fn *make_request_fn;
416 dma_drain_needed_fn *dma_drain_needed;
418 const struct blk_mq_ops *mq_ops;
421 struct blk_mq_ctx __percpu *queue_ctx;
422 unsigned int nr_queues;
424 unsigned int queue_depth;
426 /* hw dispatch queues */
427 struct blk_mq_hw_ctx **queue_hw_ctx;
428 unsigned int nr_hw_queues;
430 struct backing_dev_info *backing_dev_info;
433 * The queue owner gets to use this for whatever they like.
434 * ll_rw_blk doesn't touch it.
439 * various queue flags, see QUEUE_* below
441 unsigned long queue_flags;
443 * Number of contexts that have called blk_set_pm_only(). If this
444 * counter is above zero then only RQF_PM and RQF_PREEMPT requests are
450 * ida allocated id for this queue. Used to index queues from
456 * queue needs bounce pages for pages above this limit
460 spinlock_t queue_lock;
470 struct kobject *mq_kobj;
472 #ifdef CONFIG_BLK_DEV_INTEGRITY
473 struct blk_integrity integrity;
474 #endif /* CONFIG_BLK_DEV_INTEGRITY */
479 unsigned int nr_pending;
485 unsigned long nr_requests; /* Max # of requests */
487 unsigned int dma_drain_size;
488 void *dma_drain_buffer;
489 unsigned int dma_pad_mask;
490 unsigned int dma_alignment;
492 unsigned int rq_timeout;
495 struct blk_stat_callback *poll_cb;
496 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
498 struct timer_list timeout;
499 struct work_struct timeout_work;
501 struct list_head icq_list;
502 #ifdef CONFIG_BLK_CGROUP
503 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
504 struct blkcg_gq *root_blkg;
505 struct list_head blkg_list;
508 struct queue_limits limits;
510 unsigned int required_elevator_features;
512 #ifdef CONFIG_BLK_DEV_ZONED
514 * Zoned block device information for request dispatch control.
515 * nr_zones is the total number of zones of the device. This is always
516 * 0 for regular block devices. seq_zones_bitmap is a bitmap of nr_zones
517 * bits which indicates if a zone is conventional (bit clear) or
518 * sequential (bit set). seq_zones_wlock is a bitmap of nr_zones
519 * bits which indicates if a zone is write locked, that is, if a write
520 * request targeting the zone was dispatched. All three fields are
521 * initialized by the low level device driver (e.g. scsi/sd.c).
522 * Stacking drivers (device mappers) may or may not initialize
525 * Reads of this information must be protected with blk_queue_enter() /
526 * blk_queue_exit(). Modifying this information is only allowed while
527 * no requests are being processed. See also blk_mq_freeze_queue() and
528 * blk_mq_unfreeze_queue().
530 unsigned int nr_zones;
531 unsigned long *seq_zones_bitmap;
532 unsigned long *seq_zones_wlock;
533 #endif /* CONFIG_BLK_DEV_ZONED */
538 unsigned int sg_timeout;
539 unsigned int sg_reserved_size;
541 #ifdef CONFIG_BLK_DEV_IO_TRACE
542 struct blk_trace __rcu *blk_trace;
543 struct mutex blk_trace_mutex;
546 * for flush operations
548 struct blk_flush_queue *fq;
550 struct list_head requeue_list;
551 spinlock_t requeue_lock;
552 struct delayed_work requeue_work;
554 struct mutex sysfs_lock;
555 struct mutex sysfs_dir_lock;
558 * for reusing dead hctx instance in case of updating
561 struct list_head unused_hctx_list;
562 spinlock_t unused_hctx_lock;
566 #if defined(CONFIG_BLK_DEV_BSG)
567 struct bsg_class_device bsg_dev;
570 #ifdef CONFIG_BLK_DEV_THROTTLING
572 struct throtl_data *td;
574 struct rcu_head rcu_head;
575 wait_queue_head_t mq_freeze_wq;
577 * Protect concurrent access to q_usage_counter by
578 * percpu_ref_kill() and percpu_ref_reinit().
580 struct mutex mq_freeze_lock;
581 struct percpu_ref q_usage_counter;
583 struct blk_mq_tag_set *tag_set;
584 struct list_head tag_set_list;
585 struct bio_set bio_split;
587 #ifdef CONFIG_BLK_DEBUG_FS
588 struct dentry *debugfs_dir;
589 struct dentry *sched_debugfs_dir;
590 struct dentry *rqos_debugfs_dir;
593 bool mq_sysfs_init_done;
597 struct work_struct release_work;
599 #define BLK_MAX_WRITE_HINTS 5
600 u64 write_hints[BLK_MAX_WRITE_HINTS];
603 /* Keep blk_queue_flag_name[] in sync with the definitions below */
604 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
605 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
606 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
607 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
608 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
609 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
610 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
611 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
612 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
613 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
614 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
615 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
616 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
617 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
618 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
619 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
620 #define QUEUE_FLAG_WC 17 /* Write back caching */
621 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
622 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
623 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
624 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
625 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
626 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
627 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
628 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
629 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
630 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
632 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
633 (1 << QUEUE_FLAG_SAME_COMP))
635 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
636 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
637 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
639 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
640 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
641 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
642 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
643 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
644 #define blk_queue_noxmerges(q) \
645 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
646 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(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)
674 extern void blk_set_pm_only(struct request_queue *q);
675 extern void blk_clear_pm_only(struct request_queue *q);
677 static inline bool blk_account_rq(struct request *rq)
679 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
682 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
684 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
686 #define rq_dma_dir(rq) \
687 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
689 #define dma_map_bvec(dev, bv, dir, attrs) \
690 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
693 static inline bool queue_is_mq(struct request_queue *q)
698 static inline enum blk_zoned_model
699 blk_queue_zoned_model(struct request_queue *q)
701 return q->limits.zoned;
704 static inline bool blk_queue_is_zoned(struct request_queue *q)
706 switch (blk_queue_zoned_model(q)) {
715 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
717 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
720 #ifdef CONFIG_BLK_DEV_ZONED
721 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
723 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
726 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
729 if (!blk_queue_is_zoned(q))
731 return sector >> ilog2(q->limits.chunk_sectors);
734 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
737 if (!blk_queue_is_zoned(q) || !q->seq_zones_bitmap)
739 return test_bit(blk_queue_zone_no(q, sector), q->seq_zones_bitmap);
741 #else /* CONFIG_BLK_DEV_ZONED */
742 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
746 #endif /* CONFIG_BLK_DEV_ZONED */
748 static inline bool rq_is_sync(struct request *rq)
750 return op_is_sync(rq->cmd_flags);
753 static inline bool rq_mergeable(struct request *rq)
755 if (blk_rq_is_passthrough(rq))
758 if (req_op(rq) == REQ_OP_FLUSH)
761 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
764 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
766 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
772 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
774 if (bio_page(a) == bio_page(b) &&
775 bio_offset(a) == bio_offset(b))
781 static inline unsigned int blk_queue_depth(struct request_queue *q)
784 return q->queue_depth;
786 return q->nr_requests;
789 extern unsigned long blk_max_low_pfn, blk_max_pfn;
792 * standard bounce addresses:
794 * BLK_BOUNCE_HIGH : bounce all highmem pages
795 * BLK_BOUNCE_ANY : don't bounce anything
796 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
799 #if BITS_PER_LONG == 32
800 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
802 #define BLK_BOUNCE_HIGH -1ULL
804 #define BLK_BOUNCE_ANY (-1ULL)
805 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
808 * default timeout for SG_IO if none specified
810 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
811 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
817 unsigned long offset;
822 struct req_iterator {
823 struct bvec_iter iter;
827 /* This should not be used directly - use rq_for_each_segment */
828 #define for_each_bio(_bio) \
829 for (; _bio; _bio = _bio->bi_next)
830 #define __rq_for_each_bio(_bio, rq) \
832 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
834 #define rq_for_each_segment(bvl, _rq, _iter) \
835 __rq_for_each_bio(_iter.bio, _rq) \
836 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
838 #define rq_for_each_bvec(bvl, _rq, _iter) \
839 __rq_for_each_bio(_iter.bio, _rq) \
840 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
842 #define rq_iter_last(bvec, _iter) \
843 (_iter.bio->bi_next == NULL && \
844 bio_iter_last(bvec, _iter.iter))
846 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
847 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
849 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
850 extern void rq_flush_dcache_pages(struct request *rq);
852 static inline void rq_flush_dcache_pages(struct request *rq)
857 extern int blk_register_queue(struct gendisk *disk);
858 extern void blk_unregister_queue(struct gendisk *disk);
859 extern blk_qc_t generic_make_request(struct bio *bio);
860 extern blk_qc_t direct_make_request(struct bio *bio);
861 extern void blk_rq_init(struct request_queue *q, struct request *rq);
862 extern void blk_put_request(struct request *);
863 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
864 blk_mq_req_flags_t flags);
865 extern int blk_lld_busy(struct request_queue *q);
866 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
867 struct bio_set *bs, gfp_t gfp_mask,
868 int (*bio_ctr)(struct bio *, struct bio *, void *),
870 extern void blk_rq_unprep_clone(struct request *rq);
871 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
873 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
874 extern void blk_queue_split(struct request_queue *, struct bio **);
875 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
876 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
877 unsigned int, void __user *);
878 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
879 unsigned int, void __user *);
880 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
881 struct scsi_ioctl_command __user *);
883 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
884 extern void blk_queue_exit(struct request_queue *q);
885 extern void blk_sync_queue(struct request_queue *q);
886 extern int blk_rq_map_user(struct request_queue *, struct request *,
887 struct rq_map_data *, void __user *, unsigned long,
889 extern int blk_rq_unmap_user(struct bio *);
890 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
891 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
892 struct rq_map_data *, const struct iov_iter *,
894 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
895 struct request *, int);
896 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
897 struct request *, int, rq_end_io_fn *);
899 /* Helper to convert REQ_OP_XXX to its string format XXX */
900 extern const char *blk_op_str(unsigned int op);
902 int blk_status_to_errno(blk_status_t status);
903 blk_status_t errno_to_blk_status(int errno);
905 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
907 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
909 return bdev->bd_disk->queue; /* this is never NULL */
913 * The basic unit of block I/O is a sector. It is used in a number of contexts
914 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
915 * bytes. Variables of type sector_t represent an offset or size that is a
916 * multiple of 512 bytes. Hence these two constants.
919 #define SECTOR_SHIFT 9
922 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
926 * blk_rq_pos() : the current sector
927 * blk_rq_bytes() : bytes left in the entire request
928 * blk_rq_cur_bytes() : bytes left in the current segment
929 * blk_rq_err_bytes() : bytes left till the next error boundary
930 * blk_rq_sectors() : sectors left in the entire request
931 * blk_rq_cur_sectors() : sectors left in the current segment
932 * blk_rq_stats_sectors() : sectors of the entire request used for stats
934 static inline sector_t blk_rq_pos(const struct request *rq)
939 static inline unsigned int blk_rq_bytes(const struct request *rq)
941 return rq->__data_len;
944 static inline int blk_rq_cur_bytes(const struct request *rq)
946 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
949 extern unsigned int blk_rq_err_bytes(const struct request *rq);
951 static inline unsigned int blk_rq_sectors(const struct request *rq)
953 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
956 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
958 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
961 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
963 return rq->stats_sectors;
966 #ifdef CONFIG_BLK_DEV_ZONED
967 static inline unsigned int blk_rq_zone_no(struct request *rq)
969 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
972 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
974 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
976 #endif /* CONFIG_BLK_DEV_ZONED */
979 * Some commands like WRITE SAME have a payload or data transfer size which
980 * is different from the size of the request. Any driver that supports such
981 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
982 * calculate the data transfer size.
984 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
986 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
987 return rq->special_vec.bv_len;
988 return blk_rq_bytes(rq);
992 * Return the first full biovec in the request. The caller needs to check that
993 * there are any bvecs before calling this helper.
995 static inline struct bio_vec req_bvec(struct request *rq)
997 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
998 return rq->special_vec;
999 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
1002 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
1005 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
1006 return min(q->limits.max_discard_sectors,
1007 UINT_MAX >> SECTOR_SHIFT);
1009 if (unlikely(op == REQ_OP_WRITE_SAME))
1010 return q->limits.max_write_same_sectors;
1012 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1013 return q->limits.max_write_zeroes_sectors;
1015 return q->limits.max_sectors;
1019 * Return maximum size of a request at given offset. Only valid for
1020 * file system requests.
1022 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1025 if (!q->limits.chunk_sectors)
1026 return q->limits.max_sectors;
1028 return min(q->limits.max_sectors, (unsigned int)(q->limits.chunk_sectors -
1029 (offset & (q->limits.chunk_sectors - 1))));
1032 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1035 struct request_queue *q = rq->q;
1037 if (blk_rq_is_passthrough(rq))
1038 return q->limits.max_hw_sectors;
1040 if (!q->limits.chunk_sectors ||
1041 req_op(rq) == REQ_OP_DISCARD ||
1042 req_op(rq) == REQ_OP_SECURE_ERASE)
1043 return blk_queue_get_max_sectors(q, req_op(rq));
1045 return min(blk_max_size_offset(q, offset),
1046 blk_queue_get_max_sectors(q, req_op(rq)));
1049 static inline unsigned int blk_rq_count_bios(struct request *rq)
1051 unsigned int nr_bios = 0;
1054 __rq_for_each_bio(bio, rq)
1060 void blk_steal_bios(struct bio_list *list, struct request *rq);
1063 * Request completion related functions.
1065 * blk_update_request() completes given number of bytes and updates
1066 * the request without completing it.
1068 extern bool blk_update_request(struct request *rq, blk_status_t error,
1069 unsigned int nr_bytes);
1071 extern void __blk_complete_request(struct request *);
1072 extern void blk_abort_request(struct request *);
1075 * Access functions for manipulating queue properties
1077 extern void blk_cleanup_queue(struct request_queue *);
1078 extern void blk_queue_make_request(struct request_queue *, make_request_fn *);
1079 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1080 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1081 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1082 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1083 extern void blk_queue_max_discard_segments(struct request_queue *,
1085 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1086 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1087 unsigned int max_discard_sectors);
1088 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1089 unsigned int max_write_same_sectors);
1090 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1091 unsigned int max_write_same_sectors);
1092 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1093 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1094 extern void blk_queue_alignment_offset(struct request_queue *q,
1095 unsigned int alignment);
1096 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1097 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1098 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1099 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1100 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1101 extern void blk_set_default_limits(struct queue_limits *lim);
1102 extern void blk_set_stacking_limits(struct queue_limits *lim);
1103 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1105 extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev,
1107 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1109 extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b);
1110 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1111 extern int blk_queue_dma_drain(struct request_queue *q,
1112 dma_drain_needed_fn *dma_drain_needed,
1113 void *buf, unsigned int size);
1114 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1115 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1116 extern void blk_queue_dma_alignment(struct request_queue *, int);
1117 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1118 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1119 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1120 extern void blk_queue_required_elevator_features(struct request_queue *q,
1121 unsigned int features);
1122 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1123 struct device *dev);
1126 * Number of physical segments as sent to the device.
1128 * Normally this is the number of discontiguous data segments sent by the
1129 * submitter. But for data-less command like discard we might have no
1130 * actual data segments submitted, but the driver might have to add it's
1131 * own special payload. In that case we still return 1 here so that this
1132 * special payload will be mapped.
1134 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1136 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1138 return rq->nr_phys_segments;
1142 * Number of discard segments (or ranges) the driver needs to fill in.
1143 * Each discard bio merged into a request is counted as one segment.
1145 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1147 return max_t(unsigned short, rq->nr_phys_segments, 1);
1150 extern int blk_rq_map_sg(struct request_queue *, struct request *, struct scatterlist *);
1151 extern void blk_dump_rq_flags(struct request *, char *);
1152 extern long nr_blockdev_pages(void);
1154 bool __must_check blk_get_queue(struct request_queue *);
1155 struct request_queue *blk_alloc_queue(gfp_t);
1156 struct request_queue *blk_alloc_queue_node(gfp_t gfp_mask, int node_id);
1157 extern void blk_put_queue(struct request_queue *);
1158 extern void blk_set_queue_dying(struct request_queue *);
1161 * blk_plug permits building a queue of related requests by holding the I/O
1162 * fragments for a short period. This allows merging of sequential requests
1163 * into single larger request. As the requests are moved from a per-task list to
1164 * the device's request_queue in a batch, this results in improved scalability
1165 * as the lock contention for request_queue lock is reduced.
1167 * It is ok not to disable preemption when adding the request to the plug list
1168 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1169 * the plug list when the task sleeps by itself. For details, please see
1170 * schedule() where blk_schedule_flush_plug() is called.
1173 struct list_head mq_list; /* blk-mq requests */
1174 struct list_head cb_list; /* md requires an unplug callback */
1175 unsigned short rq_count;
1176 bool multiple_queues;
1178 #define BLK_MAX_REQUEST_COUNT 16
1179 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1182 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1183 struct blk_plug_cb {
1184 struct list_head list;
1185 blk_plug_cb_fn callback;
1188 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1189 void *data, int size);
1190 extern void blk_start_plug(struct blk_plug *);
1191 extern void blk_finish_plug(struct blk_plug *);
1192 extern void blk_flush_plug_list(struct blk_plug *, bool);
1194 static inline void blk_flush_plug(struct task_struct *tsk)
1196 struct blk_plug *plug = tsk->plug;
1199 blk_flush_plug_list(plug, false);
1202 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1204 struct blk_plug *plug = tsk->plug;
1207 blk_flush_plug_list(plug, true);
1210 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1212 struct blk_plug *plug = tsk->plug;
1215 (!list_empty(&plug->mq_list) ||
1216 !list_empty(&plug->cb_list));
1219 extern int blkdev_issue_flush(struct block_device *, gfp_t, sector_t *);
1220 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1221 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1223 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1225 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1226 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1227 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1228 sector_t nr_sects, gfp_t gfp_mask, int flags,
1231 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1232 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1234 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1235 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1237 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1238 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1240 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1241 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1243 return blkdev_issue_discard(sb->s_bdev,
1244 block << (sb->s_blocksize_bits -
1246 nr_blocks << (sb->s_blocksize_bits -
1250 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1251 sector_t nr_blocks, gfp_t gfp_mask)
1253 return blkdev_issue_zeroout(sb->s_bdev,
1254 block << (sb->s_blocksize_bits -
1256 nr_blocks << (sb->s_blocksize_bits -
1261 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1263 enum blk_default_limits {
1264 BLK_MAX_SEGMENTS = 128,
1265 BLK_SAFE_MAX_SECTORS = 255,
1266 BLK_DEF_MAX_SECTORS = 2560,
1267 BLK_MAX_SEGMENT_SIZE = 65536,
1268 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1271 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1273 return q->limits.seg_boundary_mask;
1276 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1278 return q->limits.virt_boundary_mask;
1281 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1283 return q->limits.max_sectors;
1286 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1288 return q->limits.max_hw_sectors;
1291 static inline unsigned short queue_max_segments(const struct request_queue *q)
1293 return q->limits.max_segments;
1296 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1298 return q->limits.max_discard_segments;
1301 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1303 return q->limits.max_segment_size;
1306 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1310 if (q && q->limits.logical_block_size)
1311 retval = q->limits.logical_block_size;
1316 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1318 return queue_logical_block_size(bdev_get_queue(bdev));
1321 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1323 return q->limits.physical_block_size;
1326 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1328 return queue_physical_block_size(bdev_get_queue(bdev));
1331 static inline unsigned int queue_io_min(const struct request_queue *q)
1333 return q->limits.io_min;
1336 static inline int bdev_io_min(struct block_device *bdev)
1338 return queue_io_min(bdev_get_queue(bdev));
1341 static inline unsigned int queue_io_opt(const struct request_queue *q)
1343 return q->limits.io_opt;
1346 static inline int bdev_io_opt(struct block_device *bdev)
1348 return queue_io_opt(bdev_get_queue(bdev));
1351 static inline int queue_alignment_offset(const struct request_queue *q)
1353 if (q->limits.misaligned)
1356 return q->limits.alignment_offset;
1359 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1361 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1362 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1365 return (granularity + lim->alignment_offset - alignment) % granularity;
1368 static inline int bdev_alignment_offset(struct block_device *bdev)
1370 struct request_queue *q = bdev_get_queue(bdev);
1372 if (q->limits.misaligned)
1375 if (bdev != bdev->bd_contains)
1376 return bdev->bd_part->alignment_offset;
1378 return q->limits.alignment_offset;
1381 static inline int queue_discard_alignment(const struct request_queue *q)
1383 if (q->limits.discard_misaligned)
1386 return q->limits.discard_alignment;
1389 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1391 unsigned int alignment, granularity, offset;
1393 if (!lim->max_discard_sectors)
1396 /* Why are these in bytes, not sectors? */
1397 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1398 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1402 /* Offset of the partition start in 'granularity' sectors */
1403 offset = sector_div(sector, granularity);
1405 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1406 offset = (granularity + alignment - offset) % granularity;
1408 /* Turn it back into bytes, gaah */
1409 return offset << SECTOR_SHIFT;
1413 * Two cases of handling DISCARD merge:
1414 * If max_discard_segments > 1, the driver takes every bio
1415 * as a range and send them to controller together. The ranges
1416 * needn't to be contiguous.
1417 * Otherwise, the bios/requests will be handled as same as
1418 * others which should be contiguous.
1420 static inline bool blk_discard_mergable(struct request *req)
1422 if (req_op(req) == REQ_OP_DISCARD &&
1423 queue_max_discard_segments(req->q) > 1)
1428 static inline int bdev_discard_alignment(struct block_device *bdev)
1430 struct request_queue *q = bdev_get_queue(bdev);
1432 if (bdev != bdev->bd_contains)
1433 return bdev->bd_part->discard_alignment;
1435 return q->limits.discard_alignment;
1438 static inline unsigned int bdev_write_same(struct block_device *bdev)
1440 struct request_queue *q = bdev_get_queue(bdev);
1443 return q->limits.max_write_same_sectors;
1448 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1450 struct request_queue *q = bdev_get_queue(bdev);
1453 return q->limits.max_write_zeroes_sectors;
1458 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1460 struct request_queue *q = bdev_get_queue(bdev);
1463 return blk_queue_zoned_model(q);
1465 return BLK_ZONED_NONE;
1468 static inline bool bdev_is_zoned(struct block_device *bdev)
1470 struct request_queue *q = bdev_get_queue(bdev);
1473 return blk_queue_is_zoned(q);
1478 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1480 struct request_queue *q = bdev_get_queue(bdev);
1483 return blk_queue_zone_sectors(q);
1487 static inline int queue_dma_alignment(const struct request_queue *q)
1489 return q ? q->dma_alignment : 511;
1492 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1495 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1496 return !(addr & alignment) && !(len & alignment);
1499 /* assumes size > 256 */
1500 static inline unsigned int blksize_bits(unsigned int size)
1502 unsigned int bits = 8;
1506 } while (size > 256);
1510 static inline unsigned int block_size(struct block_device *bdev)
1512 return bdev->bd_block_size;
1515 typedef struct {struct page *v;} Sector;
1517 unsigned char *read_dev_sector(struct block_device *, sector_t, Sector *);
1519 static inline void put_dev_sector(Sector p)
1524 int kblockd_schedule_work(struct work_struct *work);
1525 int kblockd_schedule_work_on(int cpu, struct work_struct *work);
1526 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1528 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1529 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1530 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1531 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1533 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1535 enum blk_integrity_flags {
1536 BLK_INTEGRITY_VERIFY = 1 << 0,
1537 BLK_INTEGRITY_GENERATE = 1 << 1,
1538 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1539 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1542 struct blk_integrity_iter {
1546 unsigned int data_size;
1547 unsigned short interval;
1548 const char *disk_name;
1551 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1552 typedef void (integrity_prepare_fn) (struct request *);
1553 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1555 struct blk_integrity_profile {
1556 integrity_processing_fn *generate_fn;
1557 integrity_processing_fn *verify_fn;
1558 integrity_prepare_fn *prepare_fn;
1559 integrity_complete_fn *complete_fn;
1563 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1564 extern void blk_integrity_unregister(struct gendisk *);
1565 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1566 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1567 struct scatterlist *);
1568 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1569 extern bool blk_integrity_merge_rq(struct request_queue *, struct request *,
1571 extern bool blk_integrity_merge_bio(struct request_queue *, struct request *,
1574 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1576 struct blk_integrity *bi = &disk->queue->integrity;
1585 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1587 return blk_get_integrity(bdev->bd_disk);
1590 static inline bool blk_integrity_rq(struct request *rq)
1592 return rq->cmd_flags & REQ_INTEGRITY;
1595 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1598 q->limits.max_integrity_segments = segs;
1601 static inline unsigned short
1602 queue_max_integrity_segments(const struct request_queue *q)
1604 return q->limits.max_integrity_segments;
1608 * bio_integrity_intervals - Return number of integrity intervals for a bio
1609 * @bi: blk_integrity profile for device
1610 * @sectors: Size of the bio in 512-byte sectors
1612 * Description: The block layer calculates everything in 512 byte
1613 * sectors but integrity metadata is done in terms of the data integrity
1614 * interval size of the storage device. Convert the block layer sectors
1615 * to the appropriate number of integrity intervals.
1617 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1618 unsigned int sectors)
1620 return sectors >> (bi->interval_exp - 9);
1623 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1624 unsigned int sectors)
1626 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1630 * Return the first bvec that contains integrity data. Only drivers that are
1631 * limited to a single integrity segment should use this helper.
1633 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1635 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1637 return rq->bio->bi_integrity->bip_vec;
1640 #else /* CONFIG_BLK_DEV_INTEGRITY */
1643 struct block_device;
1645 struct blk_integrity;
1647 static inline int blk_integrity_rq(struct request *rq)
1651 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1656 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1658 struct scatterlist *s)
1662 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1666 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1670 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1674 static inline void blk_integrity_register(struct gendisk *d,
1675 struct blk_integrity *b)
1678 static inline void blk_integrity_unregister(struct gendisk *d)
1681 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1685 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1689 static inline bool blk_integrity_merge_rq(struct request_queue *rq,
1695 static inline bool blk_integrity_merge_bio(struct request_queue *rq,
1702 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1703 unsigned int sectors)
1708 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1709 unsigned int sectors)
1714 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1719 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1721 struct block_device_operations {
1722 int (*open) (struct block_device *, fmode_t);
1723 void (*release) (struct gendisk *, fmode_t);
1724 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1725 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1726 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1727 unsigned int (*check_events) (struct gendisk *disk,
1728 unsigned int clearing);
1729 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1730 int (*media_changed) (struct gendisk *);
1731 void (*unlock_native_capacity) (struct gendisk *);
1732 int (*revalidate_disk) (struct gendisk *);
1733 int (*getgeo)(struct block_device *, struct hd_geometry *);
1734 /* this callback is with swap_lock and sometimes page table lock held */
1735 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1736 int (*report_zones)(struct gendisk *, sector_t sector,
1737 struct blk_zone *zones, unsigned int *nr_zones);
1738 struct module *owner;
1739 const struct pr_ops *pr_ops;
1742 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1744 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1745 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1746 struct writeback_control *);
1748 #ifdef CONFIG_BLK_DEV_ZONED
1749 bool blk_req_needs_zone_write_lock(struct request *rq);
1750 void __blk_req_zone_write_lock(struct request *rq);
1751 void __blk_req_zone_write_unlock(struct request *rq);
1753 static inline void blk_req_zone_write_lock(struct request *rq)
1755 if (blk_req_needs_zone_write_lock(rq))
1756 __blk_req_zone_write_lock(rq);
1759 static inline void blk_req_zone_write_unlock(struct request *rq)
1761 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1762 __blk_req_zone_write_unlock(rq);
1765 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1767 return rq->q->seq_zones_wlock &&
1768 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1771 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1773 if (!blk_req_needs_zone_write_lock(rq))
1775 return !blk_req_zone_is_write_locked(rq);
1778 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1783 static inline void blk_req_zone_write_lock(struct request *rq)
1787 static inline void blk_req_zone_write_unlock(struct request *rq)
1790 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1795 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1799 #endif /* CONFIG_BLK_DEV_ZONED */
1801 #else /* CONFIG_BLOCK */
1803 struct block_device;
1806 * stubs for when the block layer is configured out
1808 #define buffer_heads_over_limit 0
1810 static inline long nr_blockdev_pages(void)
1818 static inline void blk_start_plug(struct blk_plug *plug)
1822 static inline void blk_finish_plug(struct blk_plug *plug)
1826 static inline void blk_flush_plug(struct task_struct *task)
1830 static inline void blk_schedule_flush_plug(struct task_struct *task)
1835 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1840 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask,
1841 sector_t *error_sector)
1846 #endif /* CONFIG_BLOCK */
1848 static inline void blk_wake_io_task(struct task_struct *waiter)
1851 * If we're polling, the task itself is doing the completions. For
1852 * that case, we don't need to signal a wakeup, it's enough to just
1853 * mark us as RUNNING.
1855 if (waiter == current)
1856 __set_current_state(TASK_RUNNING);
1858 wake_up_process(waiter);