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 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 /* contains copies of user pages */
86 #define RQF_COPY_USER ((__force req_flags_t)(1 << 9))
87 /* vaguely specified driver internal error. Ignored by the block layer */
88 #define RQF_FAILED ((__force req_flags_t)(1 << 10))
89 /* don't warn about errors */
90 #define RQF_QUIET ((__force req_flags_t)(1 << 11))
91 /* elevator private data attached */
92 #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12))
93 /* account into disk and partition IO statistics */
94 #define RQF_IO_STAT ((__force req_flags_t)(1 << 13))
95 /* request came from our alloc pool */
96 #define RQF_ALLOCED ((__force req_flags_t)(1 << 14))
97 /* runtime pm request */
98 #define RQF_PM ((__force req_flags_t)(1 << 15))
99 /* on IO scheduler merge hash */
100 #define RQF_HASHED ((__force req_flags_t)(1 << 16))
101 /* track IO completion time */
102 #define RQF_STATS ((__force req_flags_t)(1 << 17))
103 /* Look at ->special_vec for the actual data payload instead of the
105 #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18))
106 /* The per-zone write lock is held for this request */
107 #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19))
108 /* already slept for hybrid poll */
109 #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20))
110 /* ->timeout has been called, don't expire again */
111 #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21))
113 /* flags that prevent us from merging requests: */
114 #define RQF_NOMERGE_FLAGS \
115 (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD)
118 * Request state for blk-mq.
127 * Try to put the fields that are referenced together in the same cacheline.
129 * If you modify this structure, make sure to update blk_rq_init() and
130 * especially blk_mq_rq_ctx_init() to take care of the added fields.
133 struct request_queue *q;
134 struct blk_mq_ctx *mq_ctx;
135 struct blk_mq_hw_ctx *mq_hctx;
137 unsigned int cmd_flags; /* op and common flags */
138 req_flags_t rq_flags;
143 /* the following two fields are internal, NEVER access directly */
144 unsigned int __data_len; /* total data len */
145 sector_t __sector; /* sector cursor */
150 struct list_head queuelist;
153 * The hash is used inside the scheduler, and killed once the
154 * request reaches the dispatch list. The ipi_list is only used
155 * to queue the request for softirq completion, which is long
156 * after the request has been unhashed (and even removed from
157 * the dispatch list).
160 struct hlist_node hash; /* merge hash */
161 struct list_head ipi_list;
165 * The rb_node is only used inside the io scheduler, requests
166 * are pruned when moved to the dispatch queue. So let the
167 * completion_data share space with the rb_node.
170 struct rb_node rb_node; /* sort/lookup */
171 struct bio_vec special_vec;
172 void *completion_data;
173 int error_count; /* for legacy drivers, don't use */
177 * Three pointers are available for the IO schedulers, if they need
178 * more they have to dynamically allocate it. Flush requests are
179 * never put on the IO scheduler. So let the flush fields share
180 * space with the elevator data.
190 struct list_head list;
191 rq_end_io_fn *saved_end_io;
195 struct gendisk *rq_disk;
196 struct hd_struct *part;
197 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
198 /* Time that the first bio started allocating this request. */
201 /* Time that this request was allocated for this IO. */
203 /* Time that I/O was submitted to the device. */
204 u64 io_start_time_ns;
206 #ifdef CONFIG_BLK_WBT
207 unsigned short wbt_flags;
210 * rq sectors used for blk stats. It has the same value
211 * with blk_rq_sectors(rq), except that it never be zeroed
214 unsigned short stats_sectors;
217 * Number of scatter-gather DMA addr+len pairs after
218 * physical address coalescing is performed.
220 unsigned short nr_phys_segments;
222 #if defined(CONFIG_BLK_DEV_INTEGRITY)
223 unsigned short nr_integrity_segments;
226 unsigned short write_hint;
227 unsigned short ioprio;
229 unsigned int extra_len; /* length of alignment and padding */
231 enum mq_rq_state state;
234 unsigned int timeout;
235 unsigned long deadline;
238 struct __call_single_data csd;
243 * completion callback.
245 rq_end_io_fn *end_io;
249 static inline bool blk_op_is_scsi(unsigned int op)
251 return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT;
254 static inline bool blk_op_is_private(unsigned int op)
256 return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT;
259 static inline bool blk_rq_is_scsi(struct request *rq)
261 return blk_op_is_scsi(req_op(rq));
264 static inline bool blk_rq_is_private(struct request *rq)
266 return blk_op_is_private(req_op(rq));
269 static inline bool blk_rq_is_passthrough(struct request *rq)
271 return blk_rq_is_scsi(rq) || blk_rq_is_private(rq);
274 static inline bool bio_is_passthrough(struct bio *bio)
276 unsigned op = bio_op(bio);
278 return blk_op_is_scsi(op) || blk_op_is_private(op);
281 static inline unsigned short req_get_ioprio(struct request *req)
286 #include <linux/elevator.h>
288 struct blk_queue_ctx;
290 typedef blk_qc_t (make_request_fn) (struct request_queue *q, struct bio *bio);
293 typedef int (dma_drain_needed_fn)(struct request *);
295 enum blk_eh_timer_return {
296 BLK_EH_DONE, /* drivers has completed the command */
297 BLK_EH_RESET_TIMER, /* reset timer and try again */
300 enum blk_queue_state {
305 #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */
306 #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */
308 #define BLK_SCSI_MAX_CMDS (256)
309 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
312 * Zoned block device models (zoned limit).
314 enum blk_zoned_model {
315 BLK_ZONED_NONE, /* Regular block device */
316 BLK_ZONED_HA, /* Host-aware zoned block device */
317 BLK_ZONED_HM, /* Host-managed zoned block device */
320 struct queue_limits {
321 unsigned long bounce_pfn;
322 unsigned long seg_boundary_mask;
323 unsigned long virt_boundary_mask;
325 unsigned int max_hw_sectors;
326 unsigned int max_dev_sectors;
327 unsigned int chunk_sectors;
328 unsigned int max_sectors;
329 unsigned int max_segment_size;
330 unsigned int physical_block_size;
331 unsigned int alignment_offset;
334 unsigned int max_discard_sectors;
335 unsigned int max_hw_discard_sectors;
336 unsigned int max_write_same_sectors;
337 unsigned int max_write_zeroes_sectors;
338 unsigned int discard_granularity;
339 unsigned int discard_alignment;
341 unsigned short logical_block_size;
342 unsigned short max_segments;
343 unsigned short max_integrity_segments;
344 unsigned short max_discard_segments;
346 unsigned char misaligned;
347 unsigned char discard_misaligned;
348 unsigned char raid_partial_stripes_expensive;
349 enum blk_zoned_model zoned;
352 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
355 #ifdef CONFIG_BLK_DEV_ZONED
357 #define BLK_ALL_ZONES ((unsigned int)-1)
358 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
359 unsigned int nr_zones, report_zones_cb cb, void *data);
361 extern unsigned int blkdev_nr_zones(struct block_device *bdev);
362 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
363 sector_t sectors, sector_t nr_sectors,
365 extern int blk_revalidate_disk_zones(struct gendisk *disk);
367 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
368 unsigned int cmd, unsigned long arg);
369 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
370 unsigned int cmd, unsigned long arg);
372 #else /* CONFIG_BLK_DEV_ZONED */
374 static inline unsigned int blkdev_nr_zones(struct block_device *bdev)
379 static inline int blk_revalidate_disk_zones(struct gendisk *disk)
384 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
385 fmode_t mode, unsigned int cmd,
391 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
392 fmode_t mode, unsigned int cmd,
398 #endif /* CONFIG_BLK_DEV_ZONED */
400 struct request_queue {
401 struct request *last_merge;
402 struct elevator_queue *elevator;
404 struct blk_queue_stats *stats;
405 struct rq_qos *rq_qos;
407 make_request_fn *make_request_fn;
408 dma_drain_needed_fn *dma_drain_needed;
410 const struct blk_mq_ops *mq_ops;
413 struct blk_mq_ctx __percpu *queue_ctx;
415 unsigned int queue_depth;
417 /* hw dispatch queues */
418 struct blk_mq_hw_ctx **queue_hw_ctx;
419 unsigned int nr_hw_queues;
421 struct backing_dev_info *backing_dev_info;
424 * The queue owner gets to use this for whatever they like.
425 * ll_rw_blk doesn't touch it.
430 * various queue flags, see QUEUE_* below
432 unsigned long queue_flags;
434 * Number of contexts that have called blk_set_pm_only(). If this
435 * counter is above zero then only RQF_PM and RQF_PREEMPT requests are
441 * ida allocated id for this queue. Used to index queues from
447 * queue needs bounce pages for pages above this limit
451 spinlock_t queue_lock;
461 struct kobject *mq_kobj;
463 #ifdef CONFIG_BLK_DEV_INTEGRITY
464 struct blk_integrity integrity;
465 #endif /* CONFIG_BLK_DEV_INTEGRITY */
470 unsigned int nr_pending;
476 unsigned long nr_requests; /* Max # of requests */
478 unsigned int dma_drain_size;
479 void *dma_drain_buffer;
480 unsigned int dma_pad_mask;
481 unsigned int dma_alignment;
483 unsigned int rq_timeout;
486 struct blk_stat_callback *poll_cb;
487 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
489 struct timer_list timeout;
490 struct work_struct timeout_work;
492 struct list_head icq_list;
493 #ifdef CONFIG_BLK_CGROUP
494 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
495 struct blkcg_gq *root_blkg;
496 struct list_head blkg_list;
499 struct queue_limits limits;
501 unsigned int required_elevator_features;
503 #ifdef CONFIG_BLK_DEV_ZONED
505 * Zoned block device information for request dispatch control.
506 * nr_zones is the total number of zones of the device. This is always
507 * 0 for regular block devices. seq_zones_bitmap is a bitmap of nr_zones
508 * bits which indicates if a zone is conventional (bit clear) or
509 * sequential (bit set). seq_zones_wlock is a bitmap of nr_zones
510 * bits which indicates if a zone is write locked, that is, if a write
511 * request targeting the zone was dispatched. All three fields are
512 * initialized by the low level device driver (e.g. scsi/sd.c).
513 * Stacking drivers (device mappers) may or may not initialize
516 * Reads of this information must be protected with blk_queue_enter() /
517 * blk_queue_exit(). Modifying this information is only allowed while
518 * no requests are being processed. See also blk_mq_freeze_queue() and
519 * blk_mq_unfreeze_queue().
521 unsigned int nr_zones;
522 unsigned long *seq_zones_bitmap;
523 unsigned long *seq_zones_wlock;
524 #endif /* CONFIG_BLK_DEV_ZONED */
529 unsigned int sg_timeout;
530 unsigned int sg_reserved_size;
532 #ifdef CONFIG_BLK_DEV_IO_TRACE
533 struct blk_trace *blk_trace;
534 struct mutex blk_trace_mutex;
537 * for flush operations
539 struct blk_flush_queue *fq;
541 struct list_head requeue_list;
542 spinlock_t requeue_lock;
543 struct delayed_work requeue_work;
545 struct mutex sysfs_lock;
546 struct mutex sysfs_dir_lock;
549 * for reusing dead hctx instance in case of updating
552 struct list_head unused_hctx_list;
553 spinlock_t unused_hctx_lock;
557 #if defined(CONFIG_BLK_DEV_BSG)
558 struct bsg_class_device bsg_dev;
561 #ifdef CONFIG_BLK_DEV_THROTTLING
563 struct throtl_data *td;
565 struct rcu_head rcu_head;
566 wait_queue_head_t mq_freeze_wq;
568 * Protect concurrent access to q_usage_counter by
569 * percpu_ref_kill() and percpu_ref_reinit().
571 struct mutex mq_freeze_lock;
572 struct percpu_ref q_usage_counter;
574 struct blk_mq_tag_set *tag_set;
575 struct list_head tag_set_list;
576 struct bio_set bio_split;
578 #ifdef CONFIG_BLK_DEBUG_FS
579 struct dentry *debugfs_dir;
580 struct dentry *sched_debugfs_dir;
581 struct dentry *rqos_debugfs_dir;
584 bool mq_sysfs_init_done;
588 struct work_struct release_work;
590 #define BLK_MAX_WRITE_HINTS 5
591 u64 write_hints[BLK_MAX_WRITE_HINTS];
594 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
595 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
596 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
597 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
598 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
599 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
600 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
601 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
602 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
603 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
604 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
605 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
606 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
607 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
608 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
609 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
610 #define QUEUE_FLAG_WC 17 /* Write back caching */
611 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
612 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
613 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
614 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
615 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
616 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
617 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
618 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
619 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
620 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
622 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
623 (1 << QUEUE_FLAG_SAME_COMP))
625 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
626 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
627 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
629 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
630 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
631 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
632 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
633 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
634 #define blk_queue_noxmerges(q) \
635 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
636 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
637 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
638 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
639 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
640 #define blk_queue_zone_resetall(q) \
641 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
642 #define blk_queue_secure_erase(q) \
643 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
644 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
645 #define blk_queue_scsi_passthrough(q) \
646 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
647 #define blk_queue_pci_p2pdma(q) \
648 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
649 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
650 #define blk_queue_rq_alloc_time(q) \
651 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
653 #define blk_queue_rq_alloc_time(q) false
656 #define blk_noretry_request(rq) \
657 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
658 REQ_FAILFAST_DRIVER))
659 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
660 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
661 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
662 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
664 extern void blk_set_pm_only(struct request_queue *q);
665 extern void blk_clear_pm_only(struct request_queue *q);
667 static inline bool blk_account_rq(struct request *rq)
669 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
672 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
674 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
676 #define rq_dma_dir(rq) \
677 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
679 #define dma_map_bvec(dev, bv, dir, attrs) \
680 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
683 static inline bool queue_is_mq(struct request_queue *q)
688 static inline enum blk_zoned_model
689 blk_queue_zoned_model(struct request_queue *q)
691 return q->limits.zoned;
694 static inline bool blk_queue_is_zoned(struct request_queue *q)
696 switch (blk_queue_zoned_model(q)) {
705 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
707 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
710 #ifdef CONFIG_BLK_DEV_ZONED
711 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
713 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
716 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
719 if (!blk_queue_is_zoned(q))
721 return sector >> ilog2(q->limits.chunk_sectors);
724 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
727 if (!blk_queue_is_zoned(q) || !q->seq_zones_bitmap)
729 return test_bit(blk_queue_zone_no(q, sector), q->seq_zones_bitmap);
731 #else /* CONFIG_BLK_DEV_ZONED */
732 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
736 #endif /* CONFIG_BLK_DEV_ZONED */
738 static inline bool rq_is_sync(struct request *rq)
740 return op_is_sync(rq->cmd_flags);
743 static inline bool rq_mergeable(struct request *rq)
745 if (blk_rq_is_passthrough(rq))
748 if (req_op(rq) == REQ_OP_FLUSH)
751 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
754 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
756 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
762 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
764 if (bio_page(a) == bio_page(b) &&
765 bio_offset(a) == bio_offset(b))
771 static inline unsigned int blk_queue_depth(struct request_queue *q)
774 return q->queue_depth;
776 return q->nr_requests;
779 extern unsigned long blk_max_low_pfn, blk_max_pfn;
782 * standard bounce addresses:
784 * BLK_BOUNCE_HIGH : bounce all highmem pages
785 * BLK_BOUNCE_ANY : don't bounce anything
786 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
789 #if BITS_PER_LONG == 32
790 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
792 #define BLK_BOUNCE_HIGH -1ULL
794 #define BLK_BOUNCE_ANY (-1ULL)
795 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
798 * default timeout for SG_IO if none specified
800 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
801 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
807 unsigned long offset;
812 struct req_iterator {
813 struct bvec_iter iter;
817 /* This should not be used directly - use rq_for_each_segment */
818 #define for_each_bio(_bio) \
819 for (; _bio; _bio = _bio->bi_next)
820 #define __rq_for_each_bio(_bio, rq) \
822 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
824 #define rq_for_each_segment(bvl, _rq, _iter) \
825 __rq_for_each_bio(_iter.bio, _rq) \
826 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
828 #define rq_for_each_bvec(bvl, _rq, _iter) \
829 __rq_for_each_bio(_iter.bio, _rq) \
830 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
832 #define rq_iter_last(bvec, _iter) \
833 (_iter.bio->bi_next == NULL && \
834 bio_iter_last(bvec, _iter.iter))
836 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
837 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
839 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
840 extern void rq_flush_dcache_pages(struct request *rq);
842 static inline void rq_flush_dcache_pages(struct request *rq)
847 extern int blk_register_queue(struct gendisk *disk);
848 extern void blk_unregister_queue(struct gendisk *disk);
849 extern blk_qc_t generic_make_request(struct bio *bio);
850 extern blk_qc_t direct_make_request(struct bio *bio);
851 extern void blk_rq_init(struct request_queue *q, struct request *rq);
852 extern void blk_put_request(struct request *);
853 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
854 blk_mq_req_flags_t flags);
855 extern int blk_lld_busy(struct request_queue *q);
856 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
857 struct bio_set *bs, gfp_t gfp_mask,
858 int (*bio_ctr)(struct bio *, struct bio *, void *),
860 extern void blk_rq_unprep_clone(struct request *rq);
861 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
863 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
864 extern void blk_queue_split(struct request_queue *, struct bio **);
865 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
866 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
867 unsigned int, void __user *);
868 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
869 unsigned int, void __user *);
870 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
871 struct scsi_ioctl_command __user *);
872 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
873 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
875 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
876 extern void blk_queue_exit(struct request_queue *q);
877 extern void blk_sync_queue(struct request_queue *q);
878 extern int blk_rq_map_user(struct request_queue *, struct request *,
879 struct rq_map_data *, void __user *, unsigned long,
881 extern int blk_rq_unmap_user(struct bio *);
882 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
883 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
884 struct rq_map_data *, const struct iov_iter *,
886 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
887 struct request *, int);
888 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
889 struct request *, int, rq_end_io_fn *);
891 /* Helper to convert REQ_OP_XXX to its string format XXX */
892 extern const char *blk_op_str(unsigned int op);
894 int blk_status_to_errno(blk_status_t status);
895 blk_status_t errno_to_blk_status(int errno);
897 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
899 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
901 return bdev->bd_disk->queue; /* this is never NULL */
905 * The basic unit of block I/O is a sector. It is used in a number of contexts
906 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
907 * bytes. Variables of type sector_t represent an offset or size that is a
908 * multiple of 512 bytes. Hence these two constants.
911 #define SECTOR_SHIFT 9
914 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
918 * blk_rq_pos() : the current sector
919 * blk_rq_bytes() : bytes left in the entire request
920 * blk_rq_cur_bytes() : bytes left in the current segment
921 * blk_rq_err_bytes() : bytes left till the next error boundary
922 * blk_rq_sectors() : sectors left in the entire request
923 * blk_rq_cur_sectors() : sectors left in the current segment
924 * blk_rq_stats_sectors() : sectors of the entire request used for stats
926 static inline sector_t blk_rq_pos(const struct request *rq)
931 static inline unsigned int blk_rq_bytes(const struct request *rq)
933 return rq->__data_len;
936 static inline int blk_rq_cur_bytes(const struct request *rq)
938 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
941 extern unsigned int blk_rq_err_bytes(const struct request *rq);
943 static inline unsigned int blk_rq_sectors(const struct request *rq)
945 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
948 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
950 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
953 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
955 return rq->stats_sectors;
958 #ifdef CONFIG_BLK_DEV_ZONED
959 static inline unsigned int blk_rq_zone_no(struct request *rq)
961 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
964 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
966 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
968 #endif /* CONFIG_BLK_DEV_ZONED */
971 * Some commands like WRITE SAME have a payload or data transfer size which
972 * is different from the size of the request. Any driver that supports such
973 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
974 * calculate the data transfer size.
976 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
978 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
979 return rq->special_vec.bv_len;
980 return blk_rq_bytes(rq);
984 * Return the first full biovec in the request. The caller needs to check that
985 * there are any bvecs before calling this helper.
987 static inline struct bio_vec req_bvec(struct request *rq)
989 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
990 return rq->special_vec;
991 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
994 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
997 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
998 return min(q->limits.max_discard_sectors,
999 UINT_MAX >> SECTOR_SHIFT);
1001 if (unlikely(op == REQ_OP_WRITE_SAME))
1002 return q->limits.max_write_same_sectors;
1004 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1005 return q->limits.max_write_zeroes_sectors;
1007 return q->limits.max_sectors;
1011 * Return maximum size of a request at given offset. Only valid for
1012 * file system requests.
1014 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1017 if (!q->limits.chunk_sectors)
1018 return q->limits.max_sectors;
1020 return min(q->limits.max_sectors, (unsigned int)(q->limits.chunk_sectors -
1021 (offset & (q->limits.chunk_sectors - 1))));
1024 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1027 struct request_queue *q = rq->q;
1029 if (blk_rq_is_passthrough(rq))
1030 return q->limits.max_hw_sectors;
1032 if (!q->limits.chunk_sectors ||
1033 req_op(rq) == REQ_OP_DISCARD ||
1034 req_op(rq) == REQ_OP_SECURE_ERASE)
1035 return blk_queue_get_max_sectors(q, req_op(rq));
1037 return min(blk_max_size_offset(q, offset),
1038 blk_queue_get_max_sectors(q, req_op(rq)));
1041 static inline unsigned int blk_rq_count_bios(struct request *rq)
1043 unsigned int nr_bios = 0;
1046 __rq_for_each_bio(bio, rq)
1052 void blk_steal_bios(struct bio_list *list, struct request *rq);
1055 * Request completion related functions.
1057 * blk_update_request() completes given number of bytes and updates
1058 * the request without completing it.
1060 extern bool blk_update_request(struct request *rq, blk_status_t error,
1061 unsigned int nr_bytes);
1063 extern void __blk_complete_request(struct request *);
1064 extern void blk_abort_request(struct request *);
1067 * Access functions for manipulating queue properties
1069 extern void blk_cleanup_queue(struct request_queue *);
1070 extern void blk_queue_make_request(struct request_queue *, make_request_fn *);
1071 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1072 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1073 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1074 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1075 extern void blk_queue_max_discard_segments(struct request_queue *,
1077 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1078 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1079 unsigned int max_discard_sectors);
1080 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1081 unsigned int max_write_same_sectors);
1082 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1083 unsigned int max_write_same_sectors);
1084 extern void blk_queue_logical_block_size(struct request_queue *, unsigned short);
1085 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1086 extern void blk_queue_alignment_offset(struct request_queue *q,
1087 unsigned int alignment);
1088 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1089 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1090 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1091 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1092 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1093 extern void blk_set_default_limits(struct queue_limits *lim);
1094 extern void blk_set_stacking_limits(struct queue_limits *lim);
1095 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1097 extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev,
1099 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1101 extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b);
1102 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1103 extern int blk_queue_dma_drain(struct request_queue *q,
1104 dma_drain_needed_fn *dma_drain_needed,
1105 void *buf, unsigned int size);
1106 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1107 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1108 extern void blk_queue_dma_alignment(struct request_queue *, int);
1109 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1110 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1111 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1112 extern void blk_queue_required_elevator_features(struct request_queue *q,
1113 unsigned int features);
1114 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1115 struct device *dev);
1118 * Number of physical segments as sent to the device.
1120 * Normally this is the number of discontiguous data segments sent by the
1121 * submitter. But for data-less command like discard we might have no
1122 * actual data segments submitted, but the driver might have to add it's
1123 * own special payload. In that case we still return 1 here so that this
1124 * special payload will be mapped.
1126 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1128 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1130 return rq->nr_phys_segments;
1134 * Number of discard segments (or ranges) the driver needs to fill in.
1135 * Each discard bio merged into a request is counted as one segment.
1137 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1139 return max_t(unsigned short, rq->nr_phys_segments, 1);
1142 extern int blk_rq_map_sg(struct request_queue *, struct request *, struct scatterlist *);
1143 extern void blk_dump_rq_flags(struct request *, char *);
1144 extern long nr_blockdev_pages(void);
1146 bool __must_check blk_get_queue(struct request_queue *);
1147 struct request_queue *blk_alloc_queue(gfp_t);
1148 struct request_queue *blk_alloc_queue_node(gfp_t gfp_mask, int node_id);
1149 extern void blk_put_queue(struct request_queue *);
1150 extern void blk_set_queue_dying(struct request_queue *);
1153 * blk_plug permits building a queue of related requests by holding the I/O
1154 * fragments for a short period. This allows merging of sequential requests
1155 * into single larger request. As the requests are moved from a per-task list to
1156 * the device's request_queue in a batch, this results in improved scalability
1157 * as the lock contention for request_queue lock is reduced.
1159 * It is ok not to disable preemption when adding the request to the plug list
1160 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1161 * the plug list when the task sleeps by itself. For details, please see
1162 * schedule() where blk_schedule_flush_plug() is called.
1165 struct list_head mq_list; /* blk-mq requests */
1166 struct list_head cb_list; /* md requires an unplug callback */
1167 unsigned short rq_count;
1168 bool multiple_queues;
1170 #define BLK_MAX_REQUEST_COUNT 16
1171 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1174 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1175 struct blk_plug_cb {
1176 struct list_head list;
1177 blk_plug_cb_fn callback;
1180 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1181 void *data, int size);
1182 extern void blk_start_plug(struct blk_plug *);
1183 extern void blk_finish_plug(struct blk_plug *);
1184 extern void blk_flush_plug_list(struct blk_plug *, bool);
1186 static inline void blk_flush_plug(struct task_struct *tsk)
1188 struct blk_plug *plug = tsk->plug;
1191 blk_flush_plug_list(plug, false);
1194 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1196 struct blk_plug *plug = tsk->plug;
1199 blk_flush_plug_list(plug, true);
1202 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1204 struct blk_plug *plug = tsk->plug;
1207 (!list_empty(&plug->mq_list) ||
1208 !list_empty(&plug->cb_list));
1211 extern int blkdev_issue_flush(struct block_device *, gfp_t, sector_t *);
1212 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1213 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1215 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1217 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1218 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1219 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1220 sector_t nr_sects, gfp_t gfp_mask, int flags,
1223 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1224 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1226 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1227 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1229 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1230 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1232 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1233 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1235 return blkdev_issue_discard(sb->s_bdev,
1236 block << (sb->s_blocksize_bits -
1238 nr_blocks << (sb->s_blocksize_bits -
1242 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1243 sector_t nr_blocks, gfp_t gfp_mask)
1245 return blkdev_issue_zeroout(sb->s_bdev,
1246 block << (sb->s_blocksize_bits -
1248 nr_blocks << (sb->s_blocksize_bits -
1253 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1255 enum blk_default_limits {
1256 BLK_MAX_SEGMENTS = 128,
1257 BLK_SAFE_MAX_SECTORS = 255,
1258 BLK_DEF_MAX_SECTORS = 2560,
1259 BLK_MAX_SEGMENT_SIZE = 65536,
1260 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1263 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1265 return q->limits.seg_boundary_mask;
1268 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1270 return q->limits.virt_boundary_mask;
1273 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1275 return q->limits.max_sectors;
1278 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1280 return q->limits.max_hw_sectors;
1283 static inline unsigned short queue_max_segments(const struct request_queue *q)
1285 return q->limits.max_segments;
1288 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1290 return q->limits.max_discard_segments;
1293 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1295 return q->limits.max_segment_size;
1298 static inline unsigned short queue_logical_block_size(const struct request_queue *q)
1302 if (q && q->limits.logical_block_size)
1303 retval = q->limits.logical_block_size;
1308 static inline unsigned short bdev_logical_block_size(struct block_device *bdev)
1310 return queue_logical_block_size(bdev_get_queue(bdev));
1313 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1315 return q->limits.physical_block_size;
1318 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1320 return queue_physical_block_size(bdev_get_queue(bdev));
1323 static inline unsigned int queue_io_min(const struct request_queue *q)
1325 return q->limits.io_min;
1328 static inline int bdev_io_min(struct block_device *bdev)
1330 return queue_io_min(bdev_get_queue(bdev));
1333 static inline unsigned int queue_io_opt(const struct request_queue *q)
1335 return q->limits.io_opt;
1338 static inline int bdev_io_opt(struct block_device *bdev)
1340 return queue_io_opt(bdev_get_queue(bdev));
1343 static inline int queue_alignment_offset(const struct request_queue *q)
1345 if (q->limits.misaligned)
1348 return q->limits.alignment_offset;
1351 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1353 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1354 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1357 return (granularity + lim->alignment_offset - alignment) % granularity;
1360 static inline int bdev_alignment_offset(struct block_device *bdev)
1362 struct request_queue *q = bdev_get_queue(bdev);
1364 if (q->limits.misaligned)
1367 if (bdev != bdev->bd_contains)
1368 return bdev->bd_part->alignment_offset;
1370 return q->limits.alignment_offset;
1373 static inline int queue_discard_alignment(const struct request_queue *q)
1375 if (q->limits.discard_misaligned)
1378 return q->limits.discard_alignment;
1381 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1383 unsigned int alignment, granularity, offset;
1385 if (!lim->max_discard_sectors)
1388 /* Why are these in bytes, not sectors? */
1389 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1390 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1394 /* Offset of the partition start in 'granularity' sectors */
1395 offset = sector_div(sector, granularity);
1397 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1398 offset = (granularity + alignment - offset) % granularity;
1400 /* Turn it back into bytes, gaah */
1401 return offset << SECTOR_SHIFT;
1404 static inline int bdev_discard_alignment(struct block_device *bdev)
1406 struct request_queue *q = bdev_get_queue(bdev);
1408 if (bdev != bdev->bd_contains)
1409 return bdev->bd_part->discard_alignment;
1411 return q->limits.discard_alignment;
1414 static inline unsigned int bdev_write_same(struct block_device *bdev)
1416 struct request_queue *q = bdev_get_queue(bdev);
1419 return q->limits.max_write_same_sectors;
1424 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1426 struct request_queue *q = bdev_get_queue(bdev);
1429 return q->limits.max_write_zeroes_sectors;
1434 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1436 struct request_queue *q = bdev_get_queue(bdev);
1439 return blk_queue_zoned_model(q);
1441 return BLK_ZONED_NONE;
1444 static inline bool bdev_is_zoned(struct block_device *bdev)
1446 struct request_queue *q = bdev_get_queue(bdev);
1449 return blk_queue_is_zoned(q);
1454 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1456 struct request_queue *q = bdev_get_queue(bdev);
1459 return blk_queue_zone_sectors(q);
1463 static inline int queue_dma_alignment(const struct request_queue *q)
1465 return q ? q->dma_alignment : 511;
1468 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1471 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1472 return !(addr & alignment) && !(len & alignment);
1475 /* assumes size > 256 */
1476 static inline unsigned int blksize_bits(unsigned int size)
1478 unsigned int bits = 8;
1482 } while (size > 256);
1486 static inline unsigned int block_size(struct block_device *bdev)
1488 return bdev->bd_block_size;
1491 typedef struct {struct page *v;} Sector;
1493 unsigned char *read_dev_sector(struct block_device *, sector_t, Sector *);
1495 static inline void put_dev_sector(Sector p)
1500 int kblockd_schedule_work(struct work_struct *work);
1501 int kblockd_schedule_work_on(int cpu, struct work_struct *work);
1502 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1504 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1505 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1506 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1507 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1509 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1511 enum blk_integrity_flags {
1512 BLK_INTEGRITY_VERIFY = 1 << 0,
1513 BLK_INTEGRITY_GENERATE = 1 << 1,
1514 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1515 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1518 struct blk_integrity_iter {
1522 unsigned int data_size;
1523 unsigned short interval;
1524 const char *disk_name;
1527 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1528 typedef void (integrity_prepare_fn) (struct request *);
1529 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1531 struct blk_integrity_profile {
1532 integrity_processing_fn *generate_fn;
1533 integrity_processing_fn *verify_fn;
1534 integrity_prepare_fn *prepare_fn;
1535 integrity_complete_fn *complete_fn;
1539 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1540 extern void blk_integrity_unregister(struct gendisk *);
1541 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1542 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1543 struct scatterlist *);
1544 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1545 extern bool blk_integrity_merge_rq(struct request_queue *, struct request *,
1547 extern bool blk_integrity_merge_bio(struct request_queue *, struct request *,
1550 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1552 struct blk_integrity *bi = &disk->queue->integrity;
1561 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1563 return blk_get_integrity(bdev->bd_disk);
1566 static inline bool blk_integrity_rq(struct request *rq)
1568 return rq->cmd_flags & REQ_INTEGRITY;
1571 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1574 q->limits.max_integrity_segments = segs;
1577 static inline unsigned short
1578 queue_max_integrity_segments(const struct request_queue *q)
1580 return q->limits.max_integrity_segments;
1584 * bio_integrity_intervals - Return number of integrity intervals for a bio
1585 * @bi: blk_integrity profile for device
1586 * @sectors: Size of the bio in 512-byte sectors
1588 * Description: The block layer calculates everything in 512 byte
1589 * sectors but integrity metadata is done in terms of the data integrity
1590 * interval size of the storage device. Convert the block layer sectors
1591 * to the appropriate number of integrity intervals.
1593 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1594 unsigned int sectors)
1596 return sectors >> (bi->interval_exp - 9);
1599 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1600 unsigned int sectors)
1602 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1606 * Return the first bvec that contains integrity data. Only drivers that are
1607 * limited to a single integrity segment should use this helper.
1609 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1611 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1613 return rq->bio->bi_integrity->bip_vec;
1616 #else /* CONFIG_BLK_DEV_INTEGRITY */
1619 struct block_device;
1621 struct blk_integrity;
1623 static inline int blk_integrity_rq(struct request *rq)
1627 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1632 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1634 struct scatterlist *s)
1638 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1642 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1646 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1650 static inline void blk_integrity_register(struct gendisk *d,
1651 struct blk_integrity *b)
1654 static inline void blk_integrity_unregister(struct gendisk *d)
1657 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1661 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1665 static inline bool blk_integrity_merge_rq(struct request_queue *rq,
1671 static inline bool blk_integrity_merge_bio(struct request_queue *rq,
1678 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1679 unsigned int sectors)
1684 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1685 unsigned int sectors)
1690 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1695 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1697 struct block_device_operations {
1698 int (*open) (struct block_device *, fmode_t);
1699 void (*release) (struct gendisk *, fmode_t);
1700 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1701 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1702 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1703 unsigned int (*check_events) (struct gendisk *disk,
1704 unsigned int clearing);
1705 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1706 int (*media_changed) (struct gendisk *);
1707 void (*unlock_native_capacity) (struct gendisk *);
1708 int (*revalidate_disk) (struct gendisk *);
1709 int (*getgeo)(struct block_device *, struct hd_geometry *);
1710 /* this callback is with swap_lock and sometimes page table lock held */
1711 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1712 int (*report_zones)(struct gendisk *, sector_t sector,
1713 unsigned int nr_zones, report_zones_cb cb, void *data);
1714 struct module *owner;
1715 const struct pr_ops *pr_ops;
1718 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1720 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1721 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1722 struct writeback_control *);
1724 #ifdef CONFIG_BLK_DEV_ZONED
1725 bool blk_req_needs_zone_write_lock(struct request *rq);
1726 void __blk_req_zone_write_lock(struct request *rq);
1727 void __blk_req_zone_write_unlock(struct request *rq);
1729 static inline void blk_req_zone_write_lock(struct request *rq)
1731 if (blk_req_needs_zone_write_lock(rq))
1732 __blk_req_zone_write_lock(rq);
1735 static inline void blk_req_zone_write_unlock(struct request *rq)
1737 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1738 __blk_req_zone_write_unlock(rq);
1741 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1743 return rq->q->seq_zones_wlock &&
1744 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1747 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1749 if (!blk_req_needs_zone_write_lock(rq))
1751 return !blk_req_zone_is_write_locked(rq);
1754 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1759 static inline void blk_req_zone_write_lock(struct request *rq)
1763 static inline void blk_req_zone_write_unlock(struct request *rq)
1766 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1771 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1775 #endif /* CONFIG_BLK_DEV_ZONED */
1777 #else /* CONFIG_BLOCK */
1779 struct block_device;
1782 * stubs for when the block layer is configured out
1784 #define buffer_heads_over_limit 0
1786 static inline long nr_blockdev_pages(void)
1794 static inline void blk_start_plug(struct blk_plug *plug)
1798 static inline void blk_finish_plug(struct blk_plug *plug)
1802 static inline void blk_flush_plug(struct task_struct *task)
1806 static inline void blk_schedule_flush_plug(struct task_struct *task)
1811 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1816 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask,
1817 sector_t *error_sector)
1822 #endif /* CONFIG_BLOCK */
1824 static inline void blk_wake_io_task(struct task_struct *waiter)
1827 * If we're polling, the task itself is doing the completions. For
1828 * that case, we don't need to signal a wakeup, it's enough to just
1829 * mark us as RUNNING.
1831 if (waiter == current)
1832 __set_current_state(TASK_RUNNING);
1834 wake_up_process(waiter);